This page contains discussions that have been archived from Village pump (technical). Please do not edit the contents of this page. If you wish to revive any of these discussions, either start a new thread or use the talk page associated with that topic.
Latest comment: 4 years ago6 comments3 people in discussion
Hey all, if I think that an editor is restoring a much earlier version of an article, is there a tool that could look at Entire Article A and scan back through the edit history to find all the versions (Entire Article B-Z) that match? WikiBlame doesn't quite do the trick, since it requires a specific phrase, and typically stops as soon as it finds the first instance of that phrase. Context: Let's say I think they might be evading a block, and I want to see if they're restoring a version that they manipulated under a different account at some point in the distant past. Thanks, Cyphoidbomb (talk) 02:12, 20 March 2020 (UTC)
@Cyphoidbomb: WikiBlame should be able to help with this. Set the "Start date" back a bit, to a date before the old version was re-instated, and tick the box "Look for removal of text". It should then find the last of the old revisions that contained your phrase. -- John of Reading (talk) 07:28, 20 March 2020 (UTC)
You could also run a database query for this, if you're searching for previous revisions that exactly match a given revision in its entirety. See quarry:query/43132; this one finds all revisions of my sandbox where it was a blank page (i.e. matching the content of Special:PermaLink/944834000). Just fork that query and replace the page_title, page_namespace and rev_id accordingly (see comments). Hope this helps, — MusikAnimaltalk16:51, 20 March 2020 (UTC)
@MusikAnimal: I think this might be exactly what I need, thank you. I gave it a whirl and ran into some issues. If I was searching an article, should the page_namespace value be zero? That seemed to work, but I wanted to double-check. Where can I find the other appropriate values for, say, Wikipedia space, Article space, draft space, etc.? Thanks! Cyphoidbomb (talk) 19:35, 20 March 2020 (UTC)
Latest comment: 4 years ago4 comments2 people in discussion
I created a template for the {{Honda Sports Award}}, which has 620 entries.
I exported to Excel so I could see how many names were duplicated, but I would also like to count red links.
While I could do so manually, I would like to monitor the count over time, so I'd like to know if there is an easy way to count them.--S Philbrick(Talk)21:36, 20 March 2020 (UTC)
document.querySelectorAll('a.new').length <-- gets you all of them
Latest comment: 4 years ago7 comments2 people in discussion
Hi everyone. I've recently discovered this Javascript code that recognises bolded keywords and adds symbols in front of them. I made a fork out of it on my common/js by adding icons and altering keywords. Here is my understanding of it:
Under var vs=, each icon corresponds to an integer starting from 0, with each new line being incremented by 1.
Under var la=, each item associates a string with the icon's integer.
I'm having some success with changing and adding icons and strings, but my question is how would it prioritise which string to use? Say I have different icons that I want to use for two terms, "delete" and "speedy delete". Is there a way to suppress the delete icon when "speedy delete" is typed? --Tenryuu 🐲 ( 💬 • 📝) 04:32, 22 March 2020 (UTC)
var vs= ... actually makes a giant string where "+" is string concatention. var vt=vs.split("#"); then splits the string at each "#" to make an array with numbering from 0. The result is as you say. The script is currently based on individual words and selects an icon for each matching word. If you set an icon for 'speedy' alone then you get both the speedy icon and the delete icon when the page says "speedy delete". Is that OK? You would similarly get two icons for "speedy keep". PrimeHunter (talk) 13:19, 22 March 2020 (UTC)
PrimeHunter, would it be appropriate to essentially consider it as an OR statement? If the test string "speedy keep" exists, then
the "keep" icon activates
the "speedy keep" icon recognises "keep" but waits on a "speedy" somewhere in the string
the "speedy keep" icon recognises speedy as well and activates
I also noticed that it doesn't appear to do exact string matching. I'm guessing the script doesn't allow for it? --Tenryuu 🐲 ( 💬 • 📝) 15:12, 22 March 2020 (UTC)
That is not how it works. A "speedy keep" icon doesn't recognize or activate anything. Icons must be associated with a single word and they activate on any bolded string containing that word. A "speedy" icon activates on a "speedy keep" comment. So does a "keep" icon. Both icons will be shown. Capitalization doesn't matter. PrimeHunter (talk) 16:11, 22 March 2020 (UTC)
Alright, thanks for the help. I'll learn a bit more Javascript when the time comes then. In the meantime I consider this Resolved. :) --Tenryuu 🐲 ( 💬 • 📝) 20:15, 22 March 2020 (UTC)
Gadget-MenuTabsToggle
Latest comment: 4 years ago3 comments2 people in discussion
They were just moved again, so is there really a need to rename them all from commons? I guess the svgs are MIA. But does this gadget even still work? ~ Amory(u • t • c)10:12, 21 March 2020 (UTC)
@Amorymeltzer: Can you put the new links they have? I tested the gadget on some another wikipedias, but it didn't work. The gadget work on our project after changing the files urls. ⇒ AramTalk21:40, 22 March 2020 (UTC)
"Strike out usernames that have been blocked" option
Latest comment: 4 years ago3 comments2 people in discussion
If you have the "Strike out usernames that have been blocked" option enabled in the appearance section of your preferences, it strikes out... well, blocked users. HOWEVER, it doesn't strike out globally locked users.
Those should also be striken. I'm posting here because I don't know where else this should be posted. Headbomb {t · c · p · b}00:03, 23 March 2020 (UTC)
By the way, it's not in the appearance section of preferences, it's in the gadgets section (in the appearance subsection). Sorry to be pedantic but this is critical to where the code is made and can be modified. The actual appearance section is part of MediaWiki and you would have to file a Phabricator request for a change there. We rarely mention the subsection of gadgets but just say gadgets. They are JavaScript and/or CSS made here at the English Wikipedia. PrimeHunter (talk) 01:50, 23 March 2020 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Problems
Some development will be slower than planned. This is because of the current pandemic. You can see the new deployment guidelines. This is to avoid risks when some persons could be unavailable.
There was a problem when adding interwiki links. The tool you use to add interwiki links could suggest the wrong project to link to. This has now been fixed. [1][2]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 24 March. It will be on non-Wikipedia wikis and some Wikipedias from 25 March. It will be on all wikis from 26 March (calendar).
Future changes
There is a project to make editing easier for newcomers. The developers are trying to understand what initiatives different Wikipedias have to welcome newcomers. They also want to know which templates are often used for maintenance activities. You can help this project by checking if your wiki's pages are listed on Wikidata.
Help Needed (read, begged for): MediaWiki JS Interface
Latest comment: 4 years ago3 comments2 people in discussion
Resolved
Ironically, I've been using AJAX to make synchronous API calls. I'd really like to swap over to using MediaWiki's JavaScript API interface, but I don't know how to make synchronous calls to it.
Of course, 'undefined' is logged to the console since get() or done() do not wait for the request to be 'completed'. Is it possible to make synchronous calls to the API via this method? I am aware that I could write }).done(result => { console.log(result); });, but let us ignore that for argument's sake. Regards, —guywan (talk • contribs)15:15, 23 March 2020 (UTC)
@Guywan: mw.Api does not support synchronous API calls. Rightfully so, since synchronous calls block the JavaScript runtime's event loop which causes the browser to freeze until the API response is received.
However, you can use the ES6 async/await functions to make your asynchronous code look synchronous. But note that ES6+ stuff is not supported by older browsers (IE 11, etc) and are thus disallowed on site-wide javascript.
Latest comment: 4 years ago4 comments3 people in discussion
When I start editing an article, visit another web page in the same tab, then use the Back button to return to my edits, my edits are gone -- I see the previous state of the page. This behavior is new. Until a week or so ago, I would routinely visit another page, then use Back to return to my incomplete edits. I really liked that functionality.
This appears to be a change in Wikipedia (different HTTP cache-control header, perhaps?), not the browser (I use Chrome 80.0.3987.149 on Windows 10). Here's a quick test I tried. I created the following HTML page:
<!DOCTYPE html><html><ahref="https://en.wiki.x.io/wiki/Main_Page">Go to WP</a><form><textarea></textarea></form></html>
navigated to it, entered something in the form, clicked on Go to WP, then hit Back, and the textarea content was still there.
But if I do the same thing with, say, my Sandbox page, the textbox content is reset. Has something changed recently? Thanks, --Macrakis (talk) 21:20, 22 March 2020 (UTC)
I don't know if anything has changed recently, but ctrl-click or whatever your browser's shortcut is for opening a link in a new tab is a safer alternative for maintaining the state of any web page than trusting the back button. Using the "open link in new tab" feature is a good habit to build. – Jonesey95 (talk) 22:13, 22 March 2020 (UTC)
I'm also seeing this and yeah, it is damned annoying to lose edits because chrome changed or because mediawiki changed; I don't know which. On win 10 and chrome, it's Ctrl+⇧ Shift-click for a new tab that gets the focus. Still, habits are very hard to break so there is the 'Warn me when I leave an edit page with unsaved changes' at Special:Preferences#mw-prefsection-editing.
Latest comment: 4 years ago4 comments3 people in discussion
I'd like to propose that a bot be used to allow users to schedule reminders for themselves. The basic spec is:
User adds an entry to /reminders.json
User adds user page to a specific category (not created yet)
Each day, but retrieves the user pages in that category, transforms them to the reminder pages ("User:Example" => "User:Example/reminders.json"), retrieves the scheduled reminders, and posts any for the day
User is responsible for removing old reminders from their /reminders.json page, but the bot only checks to see if there are reminders for "today" (UTC at noon) and so doesn't care if there are old ones left
Why not just make a gadget that stores things in the user's private JSON store (userjs-foo in the options API), and checks that every day? --Yair rand (talk) 20:00, 23 March 2020 (UTC)
A userscript already exists: User:SD0001/W-Ping, which does precisely that. The advantage of storing stuff using the options API is that reminders remain private. It is not viewable by anyone like userspace subpages. SD0001 (talk) 04:17, 24 March 2020 (UTC)
Exclude Templates in Search
Latest comment: 4 years ago5 comments3 people in discussion
Is there some way to search article text, but exclude navbox templates? For example, if I want to search Wikipedia for rumaki, I typically don't want to find all the pages which include Template:Bacon. One trick is to find an obscure term that appears in the template and search for, e.g., [rumaki -samgyeopsal]; but that isn't right, either, because it excludes pages that mention rumaki in the text of the page as well as in the Bacon template. Thanks, --Macrakis (talk) 22:09, 23 March 2020 (UTC)
insource:rumaki searches for pages saying rumaki in the source text. rumaki -hastemplate:Bacon searches pages which say rumaki in the renderd text but do not have the template. It does not find pages which both have the template and say rumaki elsewhere on the page. You may also be interested in User:PrimeHunter/Source links.js. It searches pages which link to Rumaki in the source of the page and not only via a template. It includes pages which link both in the source and a template. On Rumaki the script says Source links. PrimeHunter (talk) 22:25, 23 March 2020 (UTC)
(edit conflict) I just had to do this recently, and it was a bit easier, because the page had a disambiguator which could be searched for as well since it would only appear in links and not normal text. It's not a perfect solution, but you can try a regexp by searching insource:/\[\[[Rr]umaki/, which should match anything that links to that page, but it won't catch anything that links to a redirect to it, which may or may not be a problem. If there's a better way, I'd certainly like to know. –Deacon Vorbis (carbon • videos) 22:28, 23 March 2020 (UTC)
insource regex is expensive by itself and may time out. mw:Help:CirrusSearch#Insource says: "When possible, please avoid running a bare regexp search". If the goal was to find links (I doubt it was although a link was posted), you can search insource:rumaki insource:/\[\[[Rr]umaki/. My script does something similar. PrimeHunter (talk) 22:40, 23 March 2020 (UTC)
pqasb.pqarchiver.com is gone. Last captured 2018-05-16.
About 896 pages on English Wikipedia contain the word "pqarchiver". (list) The first few I found have unrepaired dead links.
Humans could handle these with best results, but there are so many.
A bot can do almost as well. Humans could optionally intervene or follow-up to make better results.
It's a little tangled for humans and bots because some of the {cite}s have incorrect tags like "publisher=pqasb.pqarchiver.com" (it should be Newsday), "title=Newsday..." (it should be the title of the article), no "author=", no "title=", etc.
Here, the live copy on chicagotribune.com seems best, but it had to be searched-for.
If the URL at chicagotribune.com goes away, Internet Archive of chicagotribune.com is a fallback.
The "automatable" fix would be pqasb.pqarchiver.com (if found). An archive of an archive seems ugly, but it is better than a broken link. A bot might or might not be able to find a "live" link, depending on the original publisher. If someone wants to manually search out the "live copy" afterward, the archive of the archive can make it easier. - A876 (talk) 08:13, 24 March 2020 (UTC)
It looks like it tries to make two edit areas. Is that normal for you? I haven't seen it before. Does safemode work? Does it work to log out? Try to disable syntax highlighting on the highlighter marker button to the left of "Advanced". Did you clear cache with only F5, with Ctrl+F5, or clear the entire browser cache? See Wikipedia:Bypass your cache. PrimeHunter (talk) 17:41, 22 March 2020 (UTC)
I've seen something like this before. You've got Editor / "Enable the editing toolbar" checked under Special:Preferences, yes? I think it's related to that. I used to use that, but found it didn't work reliably and turned it off. Try unchecking it and see if that helps? -- RoySmith(talk)17:47, 22 March 2020 (UTC)
Jo-Jo Eumerus, I've started using the Visual Editor most of the time, especially when I'm creating citations. The built-in citation tool works pretty well. I've also been using User:V111P/js/webRef.js sometimes. One of those might be a useful work-around for you. -- RoySmith(talk)21:37, 22 March 2020 (UTC)
I have seen this before, though not recently. I use chrome if that matters. For me, if I recall correctly, it appears to happen when the last character added to the edit window (bottom right corner) is a space character that fills or is the first to overflow edit window. Adding that character causes the browser to enable the elevators in the scrollbars. But, it has been a while since I've seen this so it is highly possible that I am mis-remembering the details. When I did encounter these problems, inserting a crlf (enter key) restored the display. After I finished writing whatever it was that I was writing, go back and delete the crlf and bob's-yer-uncle.
The thing I'm seeing most often with this highlighter is mis-registering of the highlighting by a few characters, which, once the highlighting is offset from where it is supposed to be, carries on to the end of the wikisource.
For what it's worth I am going with Firefox. I see from the page history that other editors were able to edit the page in the meantime. Jo-Jo Eumerus (talk) 19:59, 22 March 2020 (UTC)
A little update: It seems like not all versions of the history have the bug at the same time. Maybe it's something with Firefox?
I did manage to complete that article with RoySmith's suggstion, but it won't work for a large article or a multi-page source if the bug appears there as well. Jo-Jo Eumerus (talk) 09:17, 24 March 2020 (UTC)
Excessive use of style attributes makes custom theme changes (e.g. dark theme) difficult-to-impossible
Latest comment: 4 years ago2 comments2 people in discussion
Is there a reason why style="..." is used everywhere, instead of using CSS classes or IDs? (For example, on this page, there are hundreds of occurences, including color changes.) Doing so makes making large theme customizations, especially color changes such as dark-theming, for which color contrast is often an issue, extremely difficult and error-prone. — Preceding unsigned comment added by Sollyucko (talk • contribs) 19:54, 24 March 2020 (UTC)
Because creating per-template CSS classes wasn't possible in the software until very recently, and modifying site-wide CSS justifiably takes a huge amount of effort to gain consensus and work through the bureaucracy. --Ahecht (TALK PAGE) 21:13, 24 March 2020 (UTC)
New traffic report: Daily article pageviews from social media
Latest comment: 4 years ago2 comments2 people in discussion
The WMF Research team has published a new report of inbound traffic coming from Facebook, Twitter, YouTube, and Reddit.
The report contains a list of all articles that received at least 500 views from one or more of these sites (i.e. someone clicked a link on Twitter that sent them directly to a Wikipedia article). The report will be updated daily at around 14:00 UTC with traffic counts from the previous calendar day.
We believe this report provides editors with a valuable new information source. Daily inbound social media traffic stats can help editors monitor edits to articles that are going viral on social media sites and/or are being linked to by the social media platform itself in order to fact-check disinformation and other controversial content.
The social media traffic report also contains additional public article metadata that may be useful in the context of monitoring articles that are receiving unexpected attention from social media sites, such as...
the total number of pageviews (from all sources) that article received in the same period of time
the number of pageviews the article received from the same platform (e.g. Facebook) the previous day (two days ago)
The number of editors who have the page on their watchlist
The number of editors who have watchlisted the page AND recently visited it
We are currently actively seeking feedback on this report! We have some ideas of our own for how to improve the report, but we want to hear yours. If you have feature suggestions, questions, or other comments please add them to the project talkpage on Meta or ping Jonathan Morgan on his talkpage. Also be sure to check out our growing FAQ.
We intend to maintain this daily report for at least the next two months. If we receive feedback that the report is useful, we are considering making it available indefinitely. Cheers, Jmorgan (WMF) (talk) 18:56, 23 March 2020 (UTC)
Well, there is style="display:inline-table;" but something in the ombox class prevents it from working here, and we probably shouldn't try to make it work. PrimeHunter (talk) 00:17, 25 March 2020 (UTC)
Has there been a change to the configuration of the mobile site? Starting in the last 24 hours, I'm seeing very over-zealous hyphenation being automatically inserted in mobile views. Taking the current Main Page as an example, 30% of the TFA lines and 50% of the ITN lines of text end with an auto-generated hyphen, but look fine on desktop. Turning my phone to landscape view roughly halves those percentages, and moves the hyphens to different places, but it's still an off-puttingly high number of hyphens that makes the text difficult to read. They're also being inserted in very odd locations within the word e.g. im-agery and mu-sic. Modest Geniustalk16:38, 20 March 2020 (UTC)
I am using Safari on iPhone, so that's probably the same issue. I only log in on desktop, remaining logged out on mobile, so cannot make manual CSS changes there. And if there's a problem, it should probably be fixed for more than just the user who reports it... Modest Geniustalk17:22, 20 March 2020 (UTC)
PrimeHunter, I think Minerva.css didn't use to get loaded on the mobile site (only in the desktop version of the minerva skin). Judging from responses now, it seems this has been rectified last week (I know some skin rework was done, i'm guessing this is part of it). If I remember correctly, there were multiple people asking for hyphenation back then, and I made that change to test it out and asked people to look at the result. I believe we already concluded back than, that auto hypens was terrible, because it has no sense of measure, but apparently I never reverted it. Suggest we undo that change. —TheDJ (talk • contribs) 10:05, 21 March 2020 (UTC)
Again it's unclear if there was a deliberate change, but it's looking a lot better today. TFA has just two auto-hyphens, and ITN three (portrait mode, two and two on landscape), which is far more manageable. This level is probably what was originally intended. Modest Geniustalk10:56, 23 March 2020 (UTC)
Sure, but that doesn't explain why it went haywire for a few days, on Wikipedia only, affecting multiple users. There was no Safari update in that time either. Modest Geniustalk11:18, 23 March 2020 (UTC)
Modest Genius, my comment was about today. There is no difference in 'aggresivness' of the hyphenation between today and yesterday, because there is no such setting as 'aggressiveness' for auto hyphenation. As to what changed before that, that is already being discussed in thread. —TheDJ (talk • contribs) 19:22, 23 March 2020 (UTC)
The reason why we only noticed it lately was because our browsers couldn't make use of that feature till they were updated, then we took notice to the 2 year-old edit. The feature might be helpful in certain contexts, like in extremely long German compound words, but not in every syllable, and not in English, in which vowels and sometimes consonants depend on the whole word to be correctly pronounced. I wouldn't utter a T in "often", but it was split to of-ten. --Mahmudmasri (talk) 20:26, 25 March 2020 (UTC)
Articles nearly unreadable
In the past couple of days, it appears that changes have been made to Wikipedia that have made articles almost unreadable on the mobile site. Words are now allowed to be broken up across lines, connected by a hyphen (I forget the name for this). This is perhaps most noticeable in tables, such as Template:Episode table. One of the column headers in this table is "Directed by", but with the recent changes, it now appears like this:
@Mclarenfan17: I moved your post to the existing section about the issue. The hyphenation may soon be removed. You can add the above CSS to remove it now for yourself. PrimeHunter (talk) 13:14, 21 March 2020 (UTC)
Latest comment: 4 years ago3 comments2 people in discussion
Ok. The Infobox Pandemic template has something - something very wrong with it. The website linkage isn't working. All it says on every single article I have checked is Official Website. BUT the underlying URL isn't presenting. I don't want to try to tinker with it and break the myriad coronavirus-pandemic articles somehow. Help! Shearonink (talk) 04:16, 26 March 2020 (UTC)
Latest comment: 4 years ago10 comments5 people in discussion
I'm looking to customize {{Afc talk}} so that, rather than saying "If your account is [autoconfirmed] you can create articles yourself", it gives advice specific to the user's situation. Is there a template similar to {{IsIPAddress}} that identifies autoconfirmed users? Sdkb (talk) 06:44, 24 March 2020 (UTC)
Unfortunately, it is not possible for any template or module to check whether a user is autoconfirmed or not. SD0001 (talk) 09:44, 24 March 2020 (UTC)
@Sdkb: Probably could depending on the tradeoff you have to make. Basically you'd put the text of the autocofirmed user and the non-autoconfirmed and then wrap them in the appropriate class, without any if statements or similar. The source wikitext has more "junk" in it but to a new user they should only see the content they care about on the page. --Izno (talk) 21:10, 24 March 2020 (UTC)
@Izno: A little extra code sounds like a worthwhile tradeoff. (I'm assuming it would display based on the status of the viewer, not status of the user whose talk page it's on, so it'd display differently for AfC reviewers than reviewees. That's a bigger downside to me, but I'd still like to try implementing.) I found MediaWiki:Group-autoconfirmed.css, which is hopefully right, but I'm not sure where the group is for all non-autoconfirmed editors; is there a class for that or a way to code it? Sdkb (talk) 23:33, 24 March 2020 (UTC)
Latest comment: 4 years ago5 comments2 people in discussion
WP:Sockpuppet investigations/Osatmusic was archived yesterday, but there's no link to the archive page. I noticed something like this a while ago, and was told it was just a cache problem. In that case, when I looked again, sure enough the link was there, so I assumed that was the case. But, this is 14 hours ago, which seems like more than long enough for any page cache to time out. Even odder, what links here doesn't show the archive page either. I only get:
User talk:Osatmusic (links | edit)
User:Osatmusic (links | edit)
User:Krealkayln (links | edit)
User talk:Krealkayln (links | edit)
Category:Wikipedia sockpuppets of Osatmusic (links | edit)
Xaosflux, Yes, it's working for me now too. So, this is essentially the same experience I had the last time. Somebody else tries it, it works for them, and then it works for me too. That smells like some kind of cache invalidation problem inside the server stack, not my browser.
The other thing that's confusing is why the archive doesn't show up in the What Links Here listing. Looking at some other SPI pages, they're all like that. I guess the "< Wikipedia:Sockpuppet investigations | Karmaisking" line at the top of the archive is generated in some way that doesn't create an entry in the pagelinks table? -- RoySmith(talk)17:01, 26 March 2020 (UTC)
@RoySmith: those links are coming from a included template, so it certainly can be server side caching, you can always WP:PURGE a page or give it a null edit to force a page recache. — xaosfluxTalk17:14, 26 March 2020 (UTC)
Likely what is happening is: the content is edited out of the page, and the page is saved, but the new page created doesn't appear to have finished yet (or was very very close to the same time, in your example the same second) - so when it saves the page the template has nothing to do and saves it blank. The next purge or edit to the page would cause the template to recalculate. A way to avoid this is to make sure the new archive page is successfully created before removing the content from the original page. — xaosfluxTalk17:18, 26 March 2020 (UTC)
Template include size limit
Latest comment: 4 years ago19 comments8 people in discussion
2019–20 coronavirus pandemic and 2019–20 coronavirus pandemic in Mainland China are still each exceeding the template include size limit, which is generally not a good thing because it produces reader-facing errors and we aren't supposed to like those. I've tried to reduce the template size of the former article (mainly from removing inline CSS), but I've only gotten about one quarter of the way there.
The main ways I could see the situation improving:
Replacement of {{Flagdeco}}, which would result in a reduction of about 300KB. The template is impressively inefficient and has multiple nested layers. I think it could be appropriate to just substitute all the instances, but it would probably be worthwhile to improve the template as well.
Replacement of {{Medical cases chart/Row}}, which suffers from the same sort of issue; its improvement would result in a similar reduction. It's based on {{Bar stacked}}, but I think it would be worth removing the dependency if it would improve the efficiency of the template.
Reduction of the navbox sizes, specifically {{2019–20 coronavirus pandemic}}. The navboxes at the bottom of the former article take up 500KB. This navbox in particular seems to be excessively large and would probably benefit from being split into several smaller navboxes. Removing the links to the data templates alone would result in a reduction of 130KB.
None of these seem like low-hanging fruit, since both templates are fairly convoluted and the navbox split might take a while and/or be difficult to organize, so maybe there are still other things that could be reduced first. Jc86035 (talk) 22:44, 24 March 2020 (UTC)
I checked out substing the Flagdeco template; it needs to be substed 3 times, and some useless "if" statements cut to end up with a simpler file and size spec. It looks to be achievable if we do it in a sandbox end then replace the flags. There has been consensus to keep the flag images. I am prepared to this over a period of time if there is agreement here. Graeme Bartlett (talk) 00:27, 25 March 2020 (UTC)
There are 224 of flagdeco in the three sidepanel templates. I could readily replace them with much more efficient code that I would put in a module, however, previewing the 224 templates in a sandbox requires only "CPU time usage: 0.824 seconds" and not much else so I don't think working on them would help much. I'll look at the other options. I have fixed several large articles by replacing key parts with cut-down equivalents in a module and something will be needed here as the article requires "CPU time usage: 9.328 seconds" which makes edits very hard apart from the technical limit. Johnuniq (talk) 02:27, 25 March 2020 (UTC)
@Johnuniq: I'm specifically looking at the template include size. I don't think the CPU usage is really relevant (that is, even a fast template could still be inefficient in terms of template size). Directly because of the layers of nesting, {{flagdeco|United States}} becomes ~810 bytes when used in the articles but would become ~190 bytes if it were substituted all the way down. This is primarily where the improvement would come from. Jc86035 (talk) 05:52, 25 March 2020 (UTC)
To fix the post‐expand include size problem, the only options are to remove some of the above from the article, or to include the wikitext of some of the above templates in the article. Those options are ugly and the second option would not save much. Commenting out {{2019–20 coronavirus pandemic data}} and previewing the page gives 2,002,635 bytes with all the other templates expanded. Johnuniq (talk) 04:11, 25 March 2020 (UTC)
@Johnuniq: I would disagree with your assessment that these are the "only options", and I'm not sure how you would get to that conclusion. There is a decent amount of room to make many of those templates more efficient in terms of template size (I had already made two of them more efficient before starting this thread), and I think the reduction would be (barely) achievable without omitting any actual content. ({{Flagdeco}} is used in three of those templates.) Jc86035 (talk) 06:02, 25 March 2020 (UTC)
You could even cut the navboxes' template size by close to a quarter just by invoking the modules directly from the navbox templates instead of using {{Navbox}}. Jc86035 (talk) 06:08, 25 March 2020 (UTC)
@Jc86035: I replaced all 178 occurrences of {{flagdeco|...}} in {{2019–20 coronavirus pandemic data}} with the fully expanded output of one of the flags, then previewed the result at 2019–20 coronavirus pandemic. There was no significant change to the post‐expand include size in the result although it did manage to expand the first two navboxes. By "only options", I was thinking that things like halving the template expansion of the navboxes by directly calling the navbox module would not give enough benefit. Johnuniq (talk) 06:16, 25 March 2020 (UTC)
@Johnuniq: If you previewed the result in the article, because it would be exceeding the limit both before and after the change, it would just display some number close to the maximum each time (if I'm understanding the situation correctly). Testing by previewing the template without the documentation and then doubling the number is probably more accurate. Jc86035 (talk) 06:25, 25 March 2020 (UTC)
Yes, the expansion size was near the maximum but, as mentioned, it managed to expand the first two of the navboxes ({{2019–20 coronavirus pandemic}} and {{PHEIC}}). Replacing every {{flagdeco}} with its expansion would save the space from those two navboxes, namely 285,450 + 8,588 = 294,038 bytes. However, a bit more saving is needed to include all the navboxes, and the article is going to keep expanding. Johnuniq (talk) 06:34, 25 March 2020 (UTC)
Latest comment: 4 years ago3 comments2 people in discussion
I have the option to show additions/removals to categories turned on (or better, the inverse not turned off). That gives now a line in my watchlist:
[revid] [user] ([talk] [contribs] [block]) [article] added to category
I, severely, miss there a diff of the edit that caused that change in the category. I now either have to click 'contribs', and find the edit of the editor, or click 'article', load it's history, and there find the edit. If one editor made multiple edits to the same article, the actual edit that caused the category change is quite a job to find. For maintenance categories, you sometimes want to be able to revert the edit that caused the categorisation change, not having to do that manual. Are there any options to add a 'diff' link to these lines in the watchlist? --Dirk BeetstraTC06:15, 26 March 2020 (UTC)
Click the revid to see the revision. Then click "diff" to the left of "Previous revision" at the top to see the diff for the revision (there is no diff link if it's the page creation). This works for all revision links. There is also a diff link to the next and current revision. You appear to have enabled "Group changes by page in recent changes and watchlist" at Special:Preferences#mw-prefsection-rc. It's worse if it's disabled. Then you get unlinked "(diff | hist)" (this is phab:T148533) and have to do the contributions or history search you mention. PrimeHunter (talk) 15:20, 26 March 2020 (UTC)
Latest comment: 4 years ago7 comments6 people in discussion
Hello, i have a suggestion for the wikipedia's UI. When we have a long article needing a task done, it is frustrating to read it upto the end then scroll back up to click on the edit button or publish button. I wanted to suggest that can't we have a scroll up button at the bottom right corner in every wikipedia page? I think this will help a lot. Lightbluerain (talk) 18:01, 25 March 2020 (UTC)
Your browser or operating system probably also has a keyboard shortcut for "go back to the top of this window". On my computer, for example, it is "command-up arrow". – Jonesey95 (talk) 21:45, 25 March 2020 (UTC)
Yeah, "Home" or "Ctrl-Home" should work on non-Macs. (except for mobile devices. If anyone knows how to go to "top of page" on Android Lollipop, I'm dying to know it.) Elizium23 (talk) 06:25, 26 March 2020 (UTC)
Elizium23 "I use wikipedia on mobile". If you use iOS, you can double tap on the top of the screen, and it will jump to the top in any app or website. Its a not well known trick but once you know it, you won't stop using it. —TheDJ (talk • contribs) 08:43, 26 March 2020 (UTC)
Latest comment: 4 years ago10 comments4 people in discussion
I'd like to propose a change to the automated notification new editors receive after creating their account and the one they receive after making their first edit. Is there a page buried deep in MediaWiki or something where I could propose such a change? Sdkb (talk) 06:45, 26 March 2020 (UTC)
@Sdkb: Depends. Do you talk about email notifications or about in-browser notifications? Do you want to change the default in the MediaWiki software itself, or only for English Wikipedia? To change the in-browser first-edit one, see the "notification-header-thank-you" strings in the English translation of the Echo/Notifications extension at phab:diffusion/ECHO/browse/master/i18n/en.json (and see mw:How to report a bug if you want to propose changes). To change the account creation email on English Wikipedia only, see MediaWiki:Confirmemail_body. (PS: For future reference and as a courtesy to readers, please avoid "this" in summaries but name things, as nobody can know what "this" is about until having read the complete comment - thanks a lot!) --Malyacko (talk) 10:31, 26 March 2020 (UTC)
Thanks so much; you all are wizards! I'll make the proposal sometime soon at VPP for visibility, but glad to have somewhere concrete to point to. Sdkb (talk) 17:48, 26 March 2020 (UTC)
Will not support a change to an accessibility nightmare set of pages. Not sure you get it yet! --Moxy🍁20:42, 26 March 2020 (UTC)
@Moxy: I had a feeling you'd oppose this haha. You've made your views very clear, and please trust that I am reading all your comments/the resources you're linking; I sincerely appreciate the attention you're giving to the proposals, and I think it's made them stronger. On some matters like the buttons we are just going to have to disagree; I've already responded and don't have anything else to add beyond them. Regarding the new issue of accessibility you've started bringing up, my thought is that, by adding some fragment tags to Template:Intro to and then using excerpts, it should be possible to create a readable single-page display of the intro to series that could be linked from the menu. Overall, I'm not sure the technical pump is the best place for us to get into this discussion too deeply; we'll both have a chance to have our say if I get around to making it a proposal. Sdkb (talk) 22:27, 26 March 2020 (UTC)
I am really of the oppnion you need much more experience before proposing whole sale changes all over. It's wonderful you found some help pages to fix recently but I need you to read what people are saying and poating about accessibility and reader retention. Take some time to learn some coding and what new ediotrs do and need and how people navigate pages. What your proposing is the forth multiple page tutorial and the biggest one to date. We know how bad this work for our readers because we have already done this three times. Only change I would support would be a link to Help:Content as its the page with the most community input and discussion over the years. As has been said a few times now...slow down learn and test.....don't just push what you think looks pretty...learn what works. Pls review Wikipedia usable for the blind? --Moxy🍁23:08, 26 March 2020 (UTC)
Dated statements
Latest comment: 4 years ago4 comments2 people in discussion
Category:Articles containing potentially dated statements from 26 March 2020 has now reappeared in the article, perhaps someone is messing up the template. Brandmeistertalk20:14, 26 March 2020 (UTC)
often ejects the comma to a new line on a mobile display, and thus has attempted to circumvent this sort of behaviour by using the piped markup
[[Primal Scream|Primal Scream,]]
As I see it, this manner of piping ought not to be necessary and is probably in breach of WP:PIPE. Incidentally, the colleague contacted me because I have a script that removes redundant piping and which has disturbed their workaround. I have not personally observed the line ejects as described, but if such ejects are "normal behaviour" for the MW software, this may be an issue for the wider community. Any comments would be welcome -- Ohc ¡digame!09:02, 26 March 2020 (UTC)
I hardly ever encounter this behaviour on iOS. But yes, browsers sometimes do this. They break at word boundaries when needed and because , cannot be part of a word (especially when precede by a link) it is a valid ‘breakpoint’, especially when CSS word-break:break-all is used to avoid long words from overflowing outside of the viewport (used on mobile). When width is constrained, you might see it more often, simply because more breaks are required. Making workarounds is not the right strategy however and first we should check the exact spot and determine which browser does this and how often it happens. —TheDJ (talk • contribs) 12:10, 26 March 2020 (UTC)
Me too. {{nowrap|[[Primal Scream]],}} is another workaround which may be better but I don't like cluttering up the wikitext for it. It prevents wrapping inside the whole link. Whether this is desirable may depend on the length and other factors. PrimeHunter (talk) 14:21, 26 March 2020 (UTC)
Many thanks for the comments, especially those of PrimeHunter re. another possible workaround. To TheDJ, I note that wben using Chrome on Android it's very common to see hanging punctuation marks as I described, therefore I would certainly say this an issue which needs addressing. To Ohconfucius, I don't see anything on WP:PIPE to suggest that the two workarounds now suggested are proscribed. What the guidelines do mention are cases where an entire word or phrase is piped to create an internal link and that link pertains only to part of the text in question. The rationale of prohibition here is clearer, since in theory a reader might be given the impression that they're being redirected to an article on one subject (possibly as yet unwritten) when in fact it concerns quite another. However, I've never encountered this situation in practice and would argue that editors should be able to exercise their own judgement in such instances (with the caveat, perhaps, that WP:PIPE advises the exercise of caution). Punctuation inside piped internal links is a different matter entirely, since it doesn't introduce this kind of ambiguity.
Since, when I see poor typesetting on any web page I immediately begin to question the reliability of its content, my personal opinion is that it's actually rather important to avoid hanging punctuation, aside from the fact it just looks nasty!! I can't immediately think of a way of avoiding this unless the same style is applied to the word immediately preceding a punctuation mark (as far as I'm aware the "hanging-punctuation" property has, to date, only only been implemented by Safari). A workaround would therefore seem the only course of action, though it may not be ideal – in life, the "least worse" solution to most problems involves some degree of compromise
Any further thoughts on this subject would be most welcome.
To those who have observed the line break between a word and its trailing comma, I'd ask whether it occurs in the absence of a link (Edwin implies this although I don't see why it should happen ), and whether it happens on other websites of reliable sources. -- Ohc ¡digame!16:38, 26 March 2020 (UTC)
Edwin of Northumbria please file a bug with Chrome mobile. Doing weird workarounds in our content isn't sustainable long term solution honestly especially when it involves problems that are device specific. "Punctuation inside piped internal links is a different matter entirely, since it doesn't introduce this kind of ambiguity" No, but it does other things, like telling google search and Apple dictionary or screen scraping bots, that the display value of a link contains a comma. Also consider that your personal 'least worst' might be still be someone else's 'most worst'. If we don't complain with browser vendors then the problem will never be fixed and there will be a never ending workload of adding commas to links and then desktop users removing them again. —TheDJ (talk • contribs) 16:53, 26 March 2020 (UTC)
For the record, [[Primal Scream]], produces this html in both desktop and mobile: <a href="/wiki/Primal_Scream" title="Primal Scream">Primal Scream</a>, . Wrapping before the comma looks like a poor browser decision. PrimeHunter (talk) 17:51, 26 March 2020 (UTC)
Ohc ¡digame! and TheDJ – Further research indicates that the problem occurs only when a link is involved and is not limited to Chrome! The same issue arises with Firefox, Opera and MS Edge (I didn't check other browsers). I'm not sure if they all use the same display engine, but it would logical if they did and my tests appear to confirm this. None of the aforementioned software exhibits the same behaviour in Windows 10, so it seems more appropriate to file an Android bug report. If TheDJ concurs with my analysis then I'll do so ASAP.
Thanks to TheDJ for your observations re. bots, but are they really that stupid? (I'm not disagreeing with you, I'm just surprised) Could you point me to some documentation on this, especially with respect to the use of"display names"?
T3g5JZ50GLq, you could also replace it with a logo variant, which expects a light background. In your link to the Wayback Machine, the logo in the top left corner might be suitable. —andrybak (talk) 14:52, 27 March 2020 (UTC)
Duplicate IP address to mine? UPDATE: Answer, No. New question: Why is Wiki alerting me when edits from IP addresses similar to mine are posted, before I'm logged in?
Latest comment: 4 years ago6 comments4 people in discussion
There seems to be a duplicate IP address to mine; or mine's being used somehow also by another IP-number editor (vandal-type edits). I don't know how to confirm my own IP address from my machine but I get occasional messages when I come onto Wiki without being logged in. (My IP number as triggered/used by this imposter (?) is 12-digits -- four groups, three digits each; which per IP address doesn't necessarily look legit.) (Somehow there's only one recorded edit to this 12-digit IP number currently, and the edit is recent; but I've certainly encountered (a) similar situation(s) one or more time(s) before, too.) Does this outline a familiar problem at all, with a solution? Thanks! Swliv (talk) 23:04, 21 March 2020 (UTC)
Swliv, Four groups, 3 digits each sounds exactly right for an IP address. Each group can be 1 to 3 digits, from 0 to 255. There are a lot of reasons that this could happen. Mobile internet for one, or perhaps a shared internet such as at a university, business, library, or other similar places. Could also be someone else at the same house. SQLQuery me!23:27, 21 March 2020 (UTC)
UPDATE: SD0001, SQL and HiLo48, you gave me enough to motivate more work and I can now dismiss my first hypothesis, the 'duplicate' one. To start again from scratch: I've now identified two earlier cases in 2017 and 2019 respectively similar (but with different IP addresses) to the recent one that prompted the above. From all this and the Google search tip: No one that I can identify has a duplicate IP address to mine. The three 'false alerts' I'll now call them each matched my IP address for the first two trios of figures but the second two trios or doublets of numbers had NO matches (in sequential order or not) with my address. (I have only a public address as far as I can tell. I am not part of any private 'shielding' net that I know of.) So somehow it now seems Wikipedia is directing false alerts to me based on a 'resemblance'. It has occurred to me that another current device and other now-defunct devices of mine are or have been associated with my same user name; but the false alerts all appeared specifically when I was not logged in. The new question seems now to be why. If anyone has insight into the new puzzle, I'm again open to trying to solve it. Thanks much. Swliv (talk) 20:02, 27 March 2020 (UTC)
Getting text to display on mobile only
Latest comment: 4 years ago3 comments2 people in discussion
Apologies I've been asking a bunch of technical questions here recently, but one more: what do I wrap a span of text with so that the text displays only to mobile users? (I'm looking to link editors who visit tutorials on mobile to Wikipedia:Editing on mobile devices.) Sdkb (talk) 05:08, 27 March 2020 (UTC)
Thanks! I'll add a see also link to that template at Help:Mobile access so that the next editor doesn't need to come here. Feel free to revert if there are bean-spilling concerns. Sdkb (talk) 00:04, 28 March 2020 (UTC)
Why does this link show up?
Latest comment: 4 years ago10 comments4 people in discussion
Hello. If you open Template:Infobox power station in edit mode (not visual editor), you can see "Geothermal well" being listed under "Pages transcluded onto the current version of this page" (under the edit box). Any idea what part of the code is causing that? I can't seem to figure out. Rehman09:22, 26 March 2020 (UTC)
The doc page calls the template with qid = Q693330 and this triggers it in data22. Here is a minimal example where it stops if any part is removed: {{#invoke:WikidataIB |getValue |P527 |qid=Q693330 |fetchwikidata=ALL|onlysourced=no}} . The code produces: geothermal well
When the function in WikidataIB retrieves the value of the property has part(s) (P527) from Nesjavellir Power Station (Q693330), it finds "Q61846297". That has to be transformed into a readable English term, linked to an article, where possible. So the function looks at geothermal well (Q61846297) for a sitelink. If one is found, it uses it: no problem. But what if there is no sitelink (as in the case of Q61846297? Then it looks for a label (and finds "geothermal well" in this case). So can we link to an article using that label? There are three remaining possibilities: (1) the article of that name is a dab page, so we don't link; (2) the article of that name is a redirect, so we do link; (3) there is no article of that name, so we don't link. The only way to determine which of the three possibilities is correct is to examine the title object for that name, and unfortunately, that marks the page that the title object refers to as transcluded onto the page that calls it, whether the page that the title object refers to exists or not. So we either accept the erroneous transclusion, or we lose the functionality to link redirects. That would mean that we would lose two of the three links from calls like the one which returns occupation (P106) from Howard Carter (Q133682):
@RexxS: Thanks for the detailed explanation. Red links under the list of transcluded pages is often a sign that something is wrong and should be examined. The module could start with an #ifexist check on the target page and stop if the page doesn't exist. I don't know Lua but in templates, #ifexist does not affect the source page and only causes a WhatLinksHere entry (as a link and not transclusion) for the target page. That is less likely to cause concern. PrimeHunter (talk) 17:32, 26 March 2020 (UTC)
@PrimeHunter: The code (lines 630-639) already calls artitle = mw.title.new( label_text, 0 ) which returns a title object or nil if it doesn't exist in mainspace, so we know whether the article exists without any of the false links caused by #ifexist. However, you still have to check if artitle.redirectTarget exists, which is the cheapest way of distinguishing a dab link from a redirect at that point. It's examining that property that makes the false transclusion. I think you'll find that it's far better to have an false transclusion than a false file link, as bots look for the latter and report them. I've already been down that route, and believe me, the current algorithm is the result of lots of debate and amendment. Cheers --RexxS (talk) 18:21, 26 March 2020 (UTC)
@RexxS: I meant to only stop if no page exists. If it exists then it's OK to examine what it is. You say mw.title.new( label_text, 0 ) returns nil if it doesn't exist in mainspace. If that's how mw.title.new works then I don't see why you have to use artitle.redirectTarget when the page does not exist. But mw:Extension:Scribunto/Lua reference manual#mw.title.new says it only returns nil "If the text is not a valid title". I think this means it e.g. contains disallowed characters and not that there is no page by that name. As said, I don't know Lua but maybe it does have to either "link" or "transclude" the page to examine whether it exists. If we prefer to "transclude" then the documentation could mention it. I searched Module:WikidataIB for "transclu" while examining the original post. PrimeHunter (talk) 18:59, 26 March 2020 (UTC)
@PrimeHunter: I just checked and you're right about the meaning of "valid title". The code it affects is line 633 if artitle and artitle.redirectTarget .... I can change that to if artitle.exists and artitle.redirectTarget ... which will not evaluate the second part if the first is false, i.e. we don't get the false transclusion, but we do get the false file link by testing artitle.exists, so I don't think we're any better off, sorry. If you'd like me to make the change, it's just a moment's work, but I think we'll simply get the same complaints about file links instead of transclusions every time we do this:
@PrimeHunter: I've had a look around and I can't see any other relevant reports. I was relying on my memory for a couple of years ago when an earlier version of the code used the id and isRedirect object properties, and I had complaints about entries appearing in the "What links here" section of articles as file links. I've now found the discussion at Module talk:WikidataIB/Archive 3 #New parameter for getValue sought to avoid attempt to resolve redirects. This was ferret's last observation: Anomie appears to be correct in the strictest sense. isRedirect is a link, redirectTarget is a transclusion. Both appear in "What Links Here". I do not know if DABLINK stuff tells the two apart or not. If so, it is worth trying redirectTarget to find if a redirect works without tripping DABLINK. My reply now seems quite prescient: "Well, Anomie is insistent that redirectTarget doesn't create a link, so I'll happily refer any future complaints to him to sort out. The sandbox code does indeed only produce a transclusion, so I'm happy that it's an improvement, but I'd be willing to bet that the gnomes who look for links to dab pages won't see it that way. Complaints to Anomie, then. --RexxS (talk) 00:50, 28 March 2020 (UTC)
Instant Search Suggestion disappeared
Latest comment: 4 years ago6 comments2 people in discussion
Some time ago, my search box on the Wikipedia homepage stopped working (www.wiki.x.io), i.e. no drop-down suggested searches appear when I type in the search box. Perhaps this was the result of a change in the default account settings; perhaps it is an issue that can be resolved by wiping my history and cache? I do not know. If someone who is more technologically proficient than I can assist, it would be much appreciated. Ergo Sum04:17, 27 March 2020 (UTC)
@Ergo Sum: Please try purging your cache and trying again. If the issue persists, can you provide what web browser you use, and any errors that appear in the javascript console? DannyS712 (talk) 05:27, 28 March 2020 (UTC)
@DannyS712: I've just cleared my cache and no luck. I'm using Chrome. I see 6 errors in the console: cookie was set without "SameSite" attribute (x2), deprecated ResourceLoader (x3). Ergo Sum07:05, 28 March 2020 (UTC)
Latest comment: 4 years ago5 comments2 people in discussion
Seems that the Rdcheck tool is not working at the present time. Whenever a page is checked for incoming redirects, the search fails due to python coding errors. Steel1943 (talk) 20:32, 25 March 2020 (UTC)
Latest comment: 4 years ago6 comments3 people in discussion
Looks like Template:sfn has become badly broken. It is making red links and duplicate reference errors in tens of thousands of articles. Templates which use it in turn (like {{Listed building England}}) are also having problems. This seems to affect tens of thousands of articles. Is a fix coming, or is this some disruptive change that editors must manually repair in each affected article? -- Mikeblas (talk) 18:15, 28 March 2020 (UTC)
That discussion seems to be about problems where the sfn template isn't given a correct anchor for the reference. As far as I can tell, the Listed building England template correctly specifies a ref parameter. In the Harty article, for example, each of the cited buildings has the expected "CITEREFHistoric_England463505" (for example) anchor. The page still has six different "sfnp error" messages, and a "duplicate ref def" message, to boot. Maybe I'm missing some connection, but this issue seems far more than the change discussed at the Administrator Noticeboard mentions. -- Mikeblas (talk) 18:31, 28 March 2020 (UTC)
Mikeblas, it's all connected. The discussion started at User_talk:Trappist_the_monk#sfn broken?. I don't understand templates enough to understand what happened, I just know I was horrified last night when I saw the state of one of my articles. I got it cleaned up, but that's why when I saw the discussion, it caught my attention. Schazjmd(talk)18:46, 28 March 2020 (UTC)
The creator hasn't edited since 2016. I don't know the subject but I made everything optional except subunit(n), gene(n), locus(n) when SubunitCount > n-1.[8] Is that OK? PrimeHunter (talk) 12:08, 29 March 2020 (UTC)
Latest comment: 4 years ago3 comments3 people in discussion
(ColinFine from the Teahouse suggested I post this here) Hi. I would like to start writing user scripts for Wikipedia. I've read through the guide but I'm more of a visual learner so I was wondering if you could point me in the direction of any videos etc or any experienced user willing to walk me though it. Thanks, — RealFakeKimT16:50, 27 March 2020 (UTC)
Adding a bit that W3Schools is a good place to start, but it's also one of the worse sites to learn from overall. For basics, learn from Codecademy (it is only for basics imo) and then gradually use resources like freecodecamp.org (which is great) and Mozilla Developer Network (which is verbose). --qedk (t心c)13:38, 30 March 2020 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
The beta version of the Wikipedia app for Android can now help users add tags on Commons. These tags are called depicts. [9]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 31 March. It will be on non-Wikipedia wikis and some Wikipedias from 1 April. It will be on all wikis from 2 April (calendar).
Future changes
The video player will change to be simpler and more modern. The current beta feature will become the video player for everyone. The old player will be removed. [10]
There is a project to make templates easier to use. The next few weeks the developers will present ideas on the project page. You can watch that page if you are interested in giving feedback. [11]
Let's update all our COVID-19 data by bot instead of manually
Latest comment: 4 years ago21 comments11 people in discussion
Johns Hopkins University has created a public data set of all COVID-19 infections/deaths/recovered/active for all counties, states, and countries, updated and archived daily and sourced to reliable sources. While our editors have been doing an admirable job of updating all our numbers manually, the effort has not been 100% reliable or consistent, plus there's a pretty good chance you'll get an edit conflict when you try to edit the data because so many people are messing with it. I would like to propose that someone write a bot to automatically pull the data from the John Hopkins data source and use it to update tables such as Template:2019–20 coronavirus pandemic data and Template:2019–20 coronavirus pandemic data/United States medical cases by state. It could even be used to update the county lists in state outbreak articles, but that's probably lower priority. Any thoughts about this? Any volunteers? Kaldari (talk) 23:13, 26 March 2020 (UTC)
I can write python/pywikibot code that will do the updates, but it's important that it fetches the information from a source that the community trusts, and that doing bot updates is what the community wants. I can set this up tomorrow if that would help. Thanks. Mike Peel (talk) 23:36, 26 March 2020 (UTC)
I support that, but as you know, sometimes the JHU numbers lags behind compared to the reportings from individual countries/states/sovereignty as in the case for Template:2019–20 coronavirus pandemic data so I propose that the bot not only citing the JHU solely, but instead citing plurality of sources from different country like local sources like this one from Italy for example.—SquidHomme(talk)23:41, 26 March 2020 (UTC)
The license statement at the bottom of that Github readme is not compatible with CC by SA (the license is non-commercial). --Izno (talk) 23:42, 26 March 2020 (UTC)
I keep forgetting that some countries consider data to be copyrightable 🤮! I'll try to get in touch with the folks that produce the data to see if they will change the license. Kaldari (talk) 23:47, 26 March 2020 (UTC)
Yes unfortunately it says "copyright 2020 Johns Hopkins University, all rights reserved". I think in cases like this we allow for periodic usage within citations per fair use, but not systematically copying large parts of the database into Wikipedia ie. we are not competing with JHU by hosting their data here. It says at the top they acquired some of the data from public sources, so if we can determine those sources and use that data, JHU can not copyright PD data such as from the govt -- GreenC23:53, 26 March 2020 (UTC)
Different sources come to different approximations of the underlying actual number. OurWorldinData compares three major sources over time. Johns Hopkins, ECDC, and WHO.[12]Doc James (talk · contribs · email) 23:55, 26 March 2020 (UTC)
@GreenC: Raw data is not eligible for copyright in the United States and is therefore in the public domain. Regardless of their assertion, they do not have copyright and US law is clear that "facts that were discovered" are not sufficiently creative to qualify for copyright protection (see Feist v. Rural) no matter how much work went into compiling them. — Wug·a·po·des00:00, 27 March 2020 (UTC)
But do note that GreenC's is correct in saying the "systematically copying large parts of the database into Wikipedia" is not allowed. Copyright extends to the specific presentation format of the data, and so simply copy and pasting it is a violation of copyright law. So long as the data are rearranged or modified, there is no copyright infringement from simply using the underlying public domain data. Doing this to create a competing product is allowed and not infringing. — Wug·a·po·des00:10, 27 March 2020 (UTC)
Since both John Hopkins University and Wikipedia are based in the U.S., do we really need to worry about European database rights in this case? Kaldari (talk) 00:24, 27 March 2020 (UTC)
WMF Legal took a look at this and recommend seeing if Johns Hopkins will freely license their information. According to WMF Legal, while data can't be copyrighted in the U.S., choices in arrangement, selection, and presentation of the data can be copyrighted, which may apply here given what appears to be fairly significant human curation in putting the dataset together. I'll follow-up with my original plan of reaching out to Johns Hopkins directly and see if they would be willing to share the data under a free license. Kaldari (talk) 21:55, 27 March 2020 (UTC)
WugBot has been approved for trial to do something similar to this already. See Wikipedia:Bots/Requests for approval/WugBot 4. If there is consensus here, I can update the request and modify the bot to update more pages. Or, alternatively, we could probably write a lua module which parses the already available CSV data into a wikitable (I would even bet someone has already written a module like that). I don't have a strong opinion on the proposal, but am willing to help if there's consensus for it. (edit conflict) — Wug·a·po·des23:55, 26 March 2020 (UTC)
It would be nice to have a table were people can choice, do they want to see WHO, ECDC, or John Hopkins numbers. Maybe also through in WorldOMeter. Or the highest of the three. Doc James (talk · contribs · email) 23:58, 26 March 2020 (UTC)
It seems like Tedder has some sort of tool that they're using to create and maintain the case charts. I have no more information about it, since there's no BRFA or documentation. --AntiCompositeNumber (talk) 00:29, 28 March 2020 (UTC)
Hey, yeah, after doing Oregon manually and spending a lot of time with data and sources, I realized coronavirusapi.com was auditable and matched my data. So I ran against Florida and then expanded to all the other 'missing' states. I'm keeping it updated manually, and only updating pages that aren't being updated (or corrected) by others (example). It'd love it if someone replaced it with better data, this was the best I've been able to gather. tedder (talk) 03:30, 28 March 2020 (UTC)
I finally found a data source that is under an actual free license. Everything at https://covidtracking.com/ is under the Apache license (and they will soon be switching to CC0). They have an API and datasets on GitHub. However, they only cover data for the United States. Kaldari (talk) 22:09, 30 March 2020 (UTC)
Months
Latest comment: 4 years ago9 comments4 people in discussion
@PrimeHunter: thanks, but in source is only Mar 28. I said about dates at all. Please look at this template in ukwiki. If you drag the circle you will see that "Бер" was changed to "Mar"--Dimon2712 (talk) 17:03, 29 March 2020 (UTC)
Thanks for the hard work everyone. As a PSA, if anyone is looking for a new project, mw:Graphoid still uses Vega 2 despite the latest major version being Vega 5. As I've been developing the interactive maps, a lot feature requests are hard or impossible in Vega 2, and lots of people would appreciate work to update the Graphoid back end. See phab:T223026. — Wug·a·po·des23:57, 30 March 2020 (UTC)
Can you help with improving the Coronavirus epidemic maps?
Latest comment: 4 years ago6 comments4 people in discussion
Hi all
The maps on the Coronavirus pandemic articles need some technical help which has been documented on phabricator, I think that some of the tasks relate to templates, so please take a look.
If you're not able to assist with any of the tasks please still subscribe to it, it will help software engineers understand there is community support for fixing these issues.
As a dev, I can asssure you, we do not prioritize bugs based on number of subscribers or even look at number of subscribers. If you want to demonstrate support, award a token. We don't usually look at those either, but it has like a 5% chance of being noticed vs subscribers list which has like 0%. Bawolff (talk) 19:25, 28 March 2020 (UTC)
Hi @Snaevar: could you please write on the phabricator task how they could be acccomplished? It seems like the first issue is knowing which bit of software the bug is coming from. Thanks, John Cummings (talk) 15:22, 30 March 2020 (UTC)
There is no point in doing that. Like Aklapper said in the bug, this forum right here is to ask for individual maps or improvements to them, while phabricator is for things like missing features (not an exhaustive list). In this case, Template:Interactive COVID-19 maps/Cumulative confirmed cases goes as far as can be done for now, while the remaining points, apart from the last three, will only be possible once T223026 gets fixed.--Snaevar (talk) 01:38, 31 March 2020 (UTC)
Source code edit
Latest comment: 4 years ago3 comments2 people in discussion
Hi,
I am curious the reason for the 2010 wikitext editor (with editing toolbar on) in affecting the editing area, as the source code became styled unnecessarily. The only way for me to fix that is to dissable the editor. Are there anyone experiencing the same issue as me?--1233 ( T / C） 12:13, 31 March 2020 (UTC)
Latest comment: 4 years ago4 comments4 people in discussion
Hello to all the confined wikiistes on the Earth. The missing link is about this article in French fr:Réacteur_nucléaire_naturel_d'Oklo in which the inter-language link towards English is ineffective.
I insist on the fact that the article in English, Natural nuclear fission reactor, approxitively about the same subject, has got a link towards the equivalent in French that is effective.
I already sent this problem to your French equivalent here:
fr:Wikipédia:Le_Bistro/26_mars_2020#Lien_inter_langues_manquant_(suite)
, but I'm not able to understand the bla-bla (rabbiting) they write.
Thank you for your explainations (if I understand them, then I'll try to repair the thing) or for fixing the bug.Jojodesbatignoles (talk) 13:28, 31 March 2020 (UTC)
Latest comment: 4 years ago16 comments5 people in discussion
I have recently finished a user script which would help file movers when moving files. See the request here. From WP:File movers:
After moving the file, please replace all uses of the old file link with the new one.
The script is called FileMoverHelper, and can be found here. Here is a synopsis:
Get file move destination.
Move file to destination.
Remove {{Rename media}} template from destination (if there is one).
Find backlinks and redirect them to the new file name.
I appeal for peer review in order to prevent any unnecessary disruption that a faulty script of this kind may cause. Kindly leave any comments on the talk page. Regards, —guywan (talk • contribs)20:11, 26 March 2020 (UTC)
Guywan, you should also match file names not in links such as images in infoboxes, this will also allow you to match the Image: namespace that might be used. I don't think people are including filenames as text on pages. Your regex should also match uppercase and lower case first letter as well as underscores instead of spaces and it should escape regex such as .s. I would just do something like new RegExp('[' + source.charAt(0).toUpperCase() + source.charAt(0).toLowerCase() + ']' + mw.util.escapeRegExp(source.slice(1)).replace(/[ _]/g,'[ _]')). You should use list=imageusage to get all image uses. – BrandonXLF (talk)04:24, 27 March 2020 (UTC)
Your input replacement and gallery regex should also account for a lowercase f in file: and there's not need to use `${destination}` just use destination. – BrandonXLF (talk)04:28, 27 March 2020 (UTC)
@BrandonXLF: Thanks for the help! I knew there would be some problems with the regular expressions. I don't think people are including filenames as text on pages. You're probably right about that; I'm nothing if not paranoid. Do you think the gallery regex is even necessary? —guywan (talk • contribs)18:02, 27 March 2020 (UTC)
Guywan, it's no longer needed. For part 4 of the script, you should use continue/iucontinue to replace all results. When the call is finished, if result.continue is present, you should make the call again using result.continue as the start of the new object you pass to mw.Api. If result.continue is not present, then there's no more calls that need to be made. You should also consider the edit rate limit when making the replacements. You can get the edit rate from meta=userinfo passing uiprop=ratelimit. It will be contained in query.userinfo.ratelimits.edit.type_of_user.hits and query.userinfo.ratelimits.edit.type_of_user.seconds. Store hits/seconds as editrate. You can then run an interval with an interval of 1000/editrate that will check if there's an edit to be made in the queue and it will make that edit else it will clear the interval. – BrandonXLF (talk)06:03, 28 March 2020 (UTC)
For admins, it doesn't return anything (data.query.userinfo.ratelimits is an empty object - just tried it out on the testwiki). So at present the script would give syntax error if an admin uses it. SD0001 (talk) 04:49, 29 March 2020 (UTC)
@Guywan: I'm not much of a scripting expert, but I know enough to bring up a concern. So, in your script, I see no references specifically for replacing file links (referencing "Find backlinks and redirect them to the new file name" in your items above) that may contain an "Image:" or "Media:" prefix instead of the traditional "File:" namespace prefix. Well, specifically, I don't see the words "Image" or "Media" anywhere in the script, which is why I assume that my aforementioned concern is valid. I have this concern because in regards to at least the "Image:" file link prefix, it is still being used on several pages' file links. Is this addressed in the script? Steel1943 (talk) 18:26, 27 March 2020 (UTC)
@Steel1943: Your concern is valid. I think that, working from BrandonXLF's feedback, I have addressed it. I did not consider the "Image:" prefix, because I made the assumption that editors would behave and not use it. I wasn't aware of a "Media:" prefix; where is it used? —guywan (talk • contribs)18:52, 27 March 2020 (UTC)
@Guywan: See the first section of Help:Files for a bit more information than I'm obviously providing about the "Image:" and "Media:" prefixes. Regarding the "Image:" prefix: maybe the script could be designed to find the applicable file links that contain the "Image:" prefix (and either retain the "Image:" prefix or replace the prefix with "File:", but from what I recall, doing the latter through any sort of automation is controversial) when doing the redirect replacements. And in regards to the "Media:" prefix: I don't recall ever seeing any currently active file links with the "Media:" prefix, but it's probably a "better safe than sorry" situation to consider links with the "Media:" prefix. (Also, per Help:Files, it looks as though and file links which use the "Media:" prefix should continue to use the "Media:" prefix even if a redirect bypassing occurs.) Steel1943 (talk) 19:59, 27 March 2020 (UTC)
@Guywan: For your script to load reliably, you need to make sure that the dependencies are loaded before the script code is run. You can do this by replacing $(() => with
@SD0001: Thanks for the help. I've never used mw.loader.using() in any of my scripts, and they seem to run fine. Could you clarify what it means to load reliably? —guywan (talk • contribs)21:54, 28 March 2020 (UTC)
Not a big deal. But all userscripts and gadgets do this. It just ensures, in the unlikely scenario where your script has loaded before the mediawiki.* dependencies have loaded, that your script waits for the dependencies to load before starting off. SD0001 (talk) 04:46, 29 March 2020 (UTC)
Latest comment: 4 years ago9 comments3 people in discussion
Could somebody who understands template magic look at Template:Closing. The {{#if:{{{admin|}}}...}} is obviously supposed to be switching between an admin close vs a NAC, but it always comes up NAC. -- RoySmith(talk)18:00, 31 March 2020 (UTC)
QEDK, It didn't work for me on this AfD. I get the "An editor is in the process of closing this discussion." version, with the link to WP:NAC. Maybe somebody took away my mop and didn't tell me? -- RoySmith(talk)18:34, 31 March 2020 (UTC)
@RoySmith: You didn't pass in anything in the |admin= parameter so it got parsed as empty making the #if-case false. Try something like {{closing|admin=yes}}. --qedk (t心c)18:38, 31 March 2020 (UTC)
QEDK, I'll try that the next time I use it, but I don't recall ever having this trouble before.I see Wugapodes made a recent change to the template, maybe that caused the change in behavior? I note the template documentation says, "This template takes no parameters."; maybe that's just wrong? -- RoySmith(talk)18:42, 31 March 2020 (UTC)
@RoySmith: Yes, that's the change which introduced the new parameter, the documentation wasn't updated after that change, previously it said "administator or editor", now that's dependent on the parameter. --qedk (t心c)18:47, 31 March 2020 (UTC)
Yes, that was the purpose of the change I made. I felt the previous wording of "admin or other editor" was redundant. If there's a need to specify that the closer is an administrator, it should state that unequivocally, otherwise it defaults to the catch-all term "editor". I guess I forgot to change the documentation. (edit conflict) — Wug·a·po·des18:49, 31 March 2020 (UTC)
Latest comment: 4 years ago15 comments5 people in discussion
Does anybody understand why the file in the infobox appears to be 90 degrees rotated? The original on Commons is not rotated.--Ymblanter (talk) 18:06, 31 March 2020 (UTC)
@Ymblanter: It looks OK for me, can you confirm if the issue still exists. For anyone else, no recent changes to the file, article or the infobox, and nothing else that I can check comes to mind. --qedk (t心c)18:32, 31 March 2020 (UTC)
Ymblanter, It looks correct to me, but I've seen something like this before. I vaguely remember it being cache related, so perhaps try all the usual suspects of emptying your browser cache, purging the page, etc? -- RoySmith(talk)18:38, 31 March 2020 (UTC)
Thanks. I have never visited this page before, so I do not quite understand where the cache problems could come from, but I can indeed try to wait for several days,. which is typical expiration time for image cache.--Ymblanter (talk) 18:43, 31 March 2020 (UTC)
Client-side (browser) caches can be cleared anytime (setting dependent on browser). Purging the server-side cache is done on a "as needed" basis, the purge action will do that instantly. --qedk (t心c)18:52, 31 March 2020 (UTC)
QEDK, Multi-level caching is 1) a good way to make things efficient and 2) a good way to make things confusing :-). I suspect the sequence of events is something like: 1) the server cached the wrong version. 2) Ymblanter viewed the page and now his browser cached the wrong version as well. 3) The server-side cache got invalidated and refreshed so it now had the correct version. 4) You and I looked at the page and saw it correctly, even though Ymblanter is continuing to see the stale version out of his browser cache. -- RoySmith(talk)18:57, 31 March 2020 (UTC)
The idea is that the server will be serving the image (now) with a different cache header, which would effectively tell the browser that the image stored in its cache is out-of-date and the browser would ideally re-download the new image (all modern browsers will atleast), so unless it's a particularly ancient browser with strict time-based cache headers, who knows. --qedk (t心c)19:06, 31 March 2020 (UTC)
QEDK, Well, that's certainly the theory. In practice, however, I see enough problems where the fix is to manually purge the cache, that I have to assume something is broken in the WikiMedia server-side cache management implementation. -- RoySmith(talk)19:28, 31 March 2020 (UTC)
My latest Firefox (macOS Catalina 10.15.2) shows fine as well. Wikimedia (Varnish? I think it's called) caching is indeed not perfect, just see category updating and you'll see how ineffective it can be. For the sake of posterity, I've now tried it on Chrome/Safari/Firefox on Windows/macOS via Responsive Design Mode and I am unable to replicate the issue. Calling it a night. --qedk (t心c)20:04, 31 March 2020 (UTC)
@Ymblanter: Seems likely it was caused by the Orientation field in the image's EXIF data saying it should be rotated 90 degrees. I re-uploaded the image with that field corrected. --Ahecht (TALK PAGE) 19:10, 31 March 2020 (UTC)
Ahecht, this problem happens if EXIF and XMP-EXIF are not in sync (which is something the editing/exporting software is responsible for). The thumbnailing server and the browser can then make different choices in deciding what the 'correct' orientation is. —TheDJ (talk • contribs) 21:41, 31 March 2020 (UTC)
Script for showing only the red lines in lists
Latest comment: 4 years ago12 comments5 people in discussion
A list can contain links to Wikipedia articles and then is it possible to make a script for filtering the list and to remove the lines not containing red links?
The script should be a gadget or a Greasemonkey script or maybe even a bookmarklet. What's the easiest way to implement such a filter and where should I ask for help for someone to create such a script?
Thanks.
— Ark25 (talk) 02:01, 31 March 2020 (UTC)
Ark25, I have two versions javascript:$('.mw-parser-output li').each(function(){if(!$(this).find('a.new').length)this.innerHTML=''}) and javascript:$('.mw-parser-output li').each(function(){if(!$(this).find('a.new').length)this.style.display='none'}). The first one makes the content of the list items without redlinks disapear whereas the second one makes list items without redlinks disapear. – BrandonXLF (talk)18:43, 31 March 2020 (UTC)
@BrandonXLF: These are really great scripts, thank you! The second is more useful for me because the result is more compact but both are great scripts.
Is there any chance you can make another script to remove elements not containing a certain string? Say for example the string is "ir" - then only the last 2 items in the list would remain. Thanks in advance. — Ark25 (talk) 22:45, 31 March 2020 (UTC)
Ark25, would something like javascript:$('.mw-parser-output li').css('display','');if(str = prompt('Please enter a string to search for:'))$('.mw-parser-output li').each(function(){if(!$(this).text().includes(str))this.style.display='none'}) work? – BrandonXLF (talk)23:28, 31 March 2020 (UTC)
BrandonXLF It works really great, it's amazing what JavaScript can do these days! Yesterday I've posted the same question on Stackoverflow too, can I/you post the script there too? Not sure why the script doesn't work in that particular page but it works very well here on Wikipedia. — Ark25 (talk) 23:40, 31 March 2020 (UTC)
Ark25, sure. The script I gave you is designed for Wikipedia only since there's parts of the UI on Wikipedia that aren't a part of the page content, a more general script would be javascript:(function(){for(var a=document.getElementsByTagName("li"),b=prompt("Please enter a string to search for:"),c=0;c<a.length;c++)a[c].style.display=a[c].innerText.includes(b)?"":"none"})(). – BrandonXLF (talk)23:56, 31 March 2020 (UTC)
Many things can also be done with regex in the normal source editor and toolbar. For the simple case with only a link, click "Advanced" in the toolbar and then the search icon to the right. Enter \[\[(.*)\]\] at "Search for" and {{subst:#ifexist:$1||[[$1]]}} at "Replace with:". Checkmark "Treat search string as a regular expression" and click "Replace all". Then save the page, or preview if that's enough for your purpose. ifexist is an expensive parser function so at most 500 are allowed at a time. PrimeHunter (talk) 02:36, 1 April 2020 (UTC)
Rhododendrites Just add a bookmark in your browser and instead of providing an URL (web link), put the script there. To add a bookmark, right click on the bookmark bar. If you don't see the bookmark bar then you probably don't have any bookmark in it so add the first one with (CTRL+D).
Also, this is what Google search on "how to make bookmarklet" returns:
In Chrome, click Bookmarks->Bookmark Manager.
You should see a new tab with the bookmarks and folders listed.
Select the “Bookmarks Tab” folder on the left.
Click the “Organize” link, then “Add Page” in the drop down.
You should see two input fields. ...
Paste the javascript code below into the second field.
Latest comment: 4 years ago8 comments4 people in discussion
The WP:RFC process has a Tragedy of the commons problem: we can accommodate a bad or unnecessary RFC here or there, but the more RFCs we have, and the lower the average value of the RFC, the less anyone wants to participate at all, or the less thoughtful their responses will be.
We are discussing some ways to improve this. One of the proposals is to limit the number of RFCs an editor can open in a month. The idea is to limit the "outlier" behavior, from the small handful of editors who create many RFCs, rather than to bother the ordinary editor (median RFCs started per year = 0). It seems to me that even among people who use the RFC process, it's unusual to create more than about three RFCs in a month. But I'd rather have "official records" instead of just telling people that I think most people won't start more than a couple, and that creating 10 in a month (which a now-banned editor did, a year or two ago) is a highly unusual and probably bad event.
I wonder whether anyone here could find out just how many RFCs each editor has started during a month (e.g., via database dumps). For some years, the RFCs have all been given a unique id number, and they're all listed at the RFC subpages. What I imagine would happen is that you search for the id number and then record the first username/link and the date after the RFC id (a few are signed with just a date, but it's not been common in recent years). The id numbers should prevent problems with double-counting a single RFC multiple times (e.g., if it's listed as history and science).
@WhatamIdoing: The RfC ID is only for currently-running RfCs (according to Legobot tracking {{rfc}} usage). If we track RfCs with 7-hexadigit hashes similar to Legobot, by ~2300 RfCs, we would reach the number of RfCs where the collision rate will be too high (people will start noticing, bot will be confused) to use the hashes for any realistic purpose. Simply put, even if you want the hashes as a tracking mechanism in the future, you have to change the bot hash function to produce longer hashes. --qedk (t心c)06:59, 31 March 2020 (UTC)
qedk, I thought that the id numbers basically started at zero and counted up in order. If there are only ~2300 numbers in use, then I think that would get approximately two years' worth of data. That would be a good starting point. (Also: this is a one-off to validate my impressions. I don't need daily updates forever or anything like that.) WhatamIdoing (talk) 16:56, 31 March 2020 (UTC)
@WhatamIdoing: The actual numbers would ideally be 16^7 (total range of the function: 0000000->FFFFFFF). At around ~2300, you'd probably start noticing collisions, where two RfCs have some probability of having matching hashes. Either way, two simple ways would be: a tracking bot to go after Legobot's |rfcid= while they are in-use. The other is to make Legobot itself do the job. --qedk (t心c)18:23, 31 March 2020 (UTC)
The rfcid numbers are not re-used. If desired, I can rig up a demonstration of why they must not be re-used, except when the second use is on the same page that the first use occurred.
Legobot maintains a permanent table of rfcids that it has issued in the past, and it uses this when generating a new rfcid to ensure that there is no collision (if you're interested, the table is described here, it's the first one of the five). So 16^7 are certainly possible, although as time goes by the process of generating a fresh rfcid will slow down as the collision rate increases. Some rfcs have had more than one rfcid: there are several reasons that this might happen, and the most obvious one is when a {{rfc}} tag is removed (perhaps by Legobot itself, due to thirty days having elapsed) and then re-added without the |rfcid= parameter (perhaps to extend the rfc). Another reason is so that an RfC may need to be removed from an inappropriate category - for example, if somebody uses {{rfc}} without any parameters at all, Legobot will put it into Wikipedia:Requests for comment/Unsorted, and the only way of getting it out again, without actually ending the RfC, is to remove the |rfcid= and simultaneously add a valid category, as I did here - Legobot followed that up with these edits. --Redrose64 🌹 (talk) 19:14, 31 March 2020 (UTC)
I tried finding the source code, seems you're better at web-sleuthing than me! Doesn't look like the table maintains record of revisions/filer/description, so that's a bit unfortunate. OTOH, I did get to see a good bot of 'ol programmer humour:
$page->addSection('HELP! PLEASE!','Something is very wrong. I\'m having trouble generating a new RFC id. Please help me. --~~~~');die();
@Redrose64:SELECTcount(rfc_id)FROMrfcWHERErfc_id=?; This counts the number of entries in the rfc_id column with one character? I don't understand it. --qedk (t心c)20:24, 31 March 2020 (UTC)
QEDK, in mysqli when a ? is added to a prepare statement, it is replaced later on when execute is called or when bind_param is called, it acts as a placeholder. In this case it's checking for a duplicate rfc_id that is the same as $tempid in the database because the ? is replaced with $tempid. – BrandonXLF (talk)04:32, 1 April 2020 (UTC)
@BrandonXLF: Thanks a ton! That clears it up a bit: generate random MD5 hash, truncate to 7 chars, and check for the hash in rfc_id, in case the function is colliding with existing output, it does that upto 5 times, eventually messaging Chris G asking for help. Neat-o, but should really use open addressing/chaining instead of retrying the hash function imo. --qedk (t心c)04:55, 1 April 2020 (UTC)
Button coding with Lua
Latest comment: 4 years ago2 comments2 people in discussion
Sdkb, Lua is not needed for this, all Lua does is create wikitext. What you would do show is "pressed" state is check if the current page name matches the target of the button and if it does you would add a class and use templatestyles to add CSS to make the button look pressed or you would just add inline styles (this is not recommended since there's more than one button). – BrandonXLF (talk)01:12, 2 April 2020 (UTC)
Lots of first level sections are not collapsible
Latest comment: 4 years ago19 comments4 people in discussion
SharabSalam, Can you give us more details? Are you looking at this with a normal web browser on your phone? Does the URL contain "en.wiki.x.io" or "en.m.wiki.x.io"? Or are you using the Wikipedia App, and if so, Android or IOS? And, specifically, which section head do you think should be collapsed? -- RoySmith(talk)14:08, 28 March 2020 (UTC)
This is happening in my phone. It is android and I am using Chrome. Lots of first second level sections in that article are not collapsible. The second third level sections have underlines which is not how a normal second third level section should look like. Here are some screenshots to illustrate the problem.
Screenshot AScreenshot B
Screenshot A is from the article. Notice that there is no downwards/upwards arrow and that there is an underline under the second third level section header which is not normal.
Screenshot B is from my sandbox. This is how sections should look like.
All of the sections in that article are like this except I think two sections at the end. One is "Notes" and the other is "History". This is annoying especially because there is a very long section that is not collapsible called "National law" and you have to scroll down for 15-20 seconds to reach the end of the section.--SharʿabSalam▼ (talk) 08:15, 29 March 2020 (UTC)
I see the same in Safari on an iPhone. A level 2 section has "==" and level 3 has "===". It's level 2 which isn't collapsible and level 3 which has underlines that aren't normally there. I have a collapse arrow at "Notes", "History" and "Zoophilic pornography" but the latter only collapses the first line with {{Main|Animal pornography}}. On the mobile version in desktop with Firefox I see the three navboxes in [14] so something odd is going on. I don't see them in preview and I never see navboxes on articles in the mobile version. They aren't supposed to be in the html in mobile so it's not just an unloaded CSS statement to hide them. A null edit didn't change anyhting. None of the unusual things appear in preview so it's difficult to investigate. PrimeHunter (talk) 13:13, 29 March 2020 (UTC)
SharabSalam, Wow, this is really weird. I made a copy of the page in my userspace and started hacking away at chunks. Eventually I got down to this diff causing the problem to appear or not. I'm totally mystified.
What this smells like is running into some kind of size limit rather than any actual broken markup. I'm not familiar with the code base, but I could imagine something like reading the next X kb of text to see if you can find the start of the next level-2 section, and that failing if the next section head is further away than that. But that's just pure speculation. -- RoySmith(talk)15:38, 29 March 2020 (UTC)
BTW, the maximum page size per WP:CHOKING is about 2 MB, so we're not even close to that. Also, I was reproducing this on my desktop by explicitly viewing the mobile version (i.e. en.m.wiki.x.io), so it's not specifically a mobile browser problem; it's happening on the server side. -- RoySmith(talk)15:47, 29 March 2020 (UTC)
The navboxes appear before the edit but not after when viewing the mobile version with Firefox on a desktop. After the edit they are also omitted in the html as expected. PrimeHunter (talk) 20:44, 29 March 2020 (UTC)
It has to do with size because no matter what I remove, as long as I remove enough of the page, the page parses properly on the mobile site. Both halves of the page work properly when split, 1 and 2. The edit that fixed the issue in RoySmith's userspace reduced the size of the page from 112867 bytes to 112834 bytes. – BrandonXLF (talk)22:54, 30 March 2020 (UTC)
It has to do with the number of files (at least partly, maybe it as to do with the size of the file links) because all RoySmith did was remove a file to fix the issue and I removed a different file here [15] and it also fixed the issue. This revision with a lot of files [16] has the same issue aswell. – BrandonXLF (talk)23:14, 30 March 2020 (UTC)
I was correct, MobileFrontend sets MFMobileFormatterOptions.maxImages to 1000 and since there's over 1000 images, the server doesn't apply the mobile formatting.– BrandonXLF (talk)01:15, 31 March 2020 (UTC)
RoySmith, I had this idea in my mind. I will see. By the way, I can't enter Phabricator. I always get internal server error. Only if I used another WiFi network or used a proxy I can enter. I don't know why. This has been the case since I joined Wikipedia.--SharʿabSalam▼ (talk) 22:57, 1 April 2020 (UTC)
SharabSalam, Well, I'm glad we found something that got you up and running. Weird about not being able to access phab. If you can give me some more details (URL you're accessing, error message you get, etc) I'll see if I can figure out what's up with that. -- RoySmith(talk)01:30, 2 April 2020 (UTC)
The message is like this:
Internal Server Error The server encountered an internal error or misconfiguration and was unable to complete your request. Please contact the server administrator at webmaster@wikimedia.org to inform them of the time this error occurred, and the actions you performed just before this error. More information about this error may be available in the server error log.
It's not a big deal. It seems that it only happens when I am using my WiFi network because when I use another WiFi network or a proxy I can enter that website. If I wanted to enter that website I just use a proxy.--SharʿabSalam▼ (talk) 03:12, 2 April 2020 (UTC)
Eliminating certain users' edits from your watchlist
Latest comment: 4 years ago8 comments4 people in discussion
The code is broken in various ways, so I'd recommend against using it. (Normally, JS code can be added to Special:MyPage/common.js.) Any code to remove particular listings would need to work somewhat differently depending on what settings one is using. --Yair rand (talk) 21:10, 1 April 2020 (UTC)
@The Rambling Man: Specifically the relevant settings are the "Use non-Javascript interface" option in Special:Preferences#mw-prefsection-watchlist and the "Group changes by page in recent changes and watchlist" in Special:Preferences#mw-prefsection-rc. Assuming both options are at the default state (unchecked), then, using Jorm's code (slightly modified below to use the correct style keyword and to run after load) should work:
(Replace "BobTheUser" with the appropriate username to be blocked.) This should work, so long as the watchlist JS itself isn't doing anything crazy, which it might. --Yair rand (talk) 22:30, 1 April 2020 (UTC)
Just FYI, using js to do this will literally just "delete the line" from the screen, if you are not using "grouped" option and "normal user" makes a change, then "hidden user" makes a change to a page, it won't be able to show the "normal" change on your WL - it will just be empty. — xaosfluxTalk14:02, 2 April 2020 (UTC)
Template not working
Latest comment: 4 years ago3 comments2 people in discussion
Latest comment: 4 years ago13 comments5 people in discussion
Hello, can someone create a regular expression rule for me please? I want to match all words which start with ر letter. I tried with English letters successfully, but it doesn't work for Arabic letters. Thanks! ⇒ AramTalk11:23, 2 April 2020 (UTC)
Where would the regex be used like a "insource:" search, "Search and replace" button in the edit window, AWB, etc? Is there an example word that starts with it? -- GreenC12:52, 2 April 2020 (UTC)
If javascript regexes handled unicode properly, then \b(ر\w*) would do it. I'm not a webdev, and a quick google and some experimentation with my own browser suggests they don't. You can try to match the word boundaries manually with something like (^|\s)(ر\S*) which is sufficient for your example, but that way is fraught with peril: it only matches words delimited by whitespace, and not (for example) punctuation; and if you're doing something like replacement instead of just searching, you also have to deal with replacing the whitespace character the first group matches. —Cryptic16:37, 2 April 2020 (UTC)
@Cryptic: Oops! I'm sorry! I had a mistake in the result above. Please look the result again. I want to match only ر letter (not the word.) ⇒ AramTalk17:33, 2 April 2020 (UTC)
This is drifting wayyyy out of scope. The technical section of the village pump is used to discuss technical issues about Wikipedia. Now you want me to help you with Google searches. No can do. --Redrose64 🌹 (talk) 20:04, 2 April 2020 (UTC)
FWIW, if I needed advanced help with a regex these days, stack overflow would be my go-to site. I actually consider myself a regex wizard, but this is the intersection of regex, javascript, and unicode, which is a scarry place. -- RoySmith(talk)20:12, 2 April 2020 (UTC)
@Redrose64: Thank you! I fixed the issue. But why are you angry at me? I just told you about flags. Anyway, I do apologize! Thank you all! ⇒ AramTalk20:40, 2 April 2020 (UTC)
Mass redirecting?
Latest comment: 4 years ago12 comments8 people in discussion
I have a long list of uncreated titles (here), all of which should be redirected to the page Maymun-Dizh (with the content being like this).
I wonder if such an automated task has been done previously? If not, I wonder what would the best way of doing this (e.g. bots)? --Z20:15, 31 March 2020 (UTC)
But 216 redirects seems excessive. I tried five random in quotes on Google. The first four had zero results. The last "Meimun-Dez" had three. PrimeHunter (talk) 20:28, 31 March 2020 (UTC)
Please do not mass create redirects to one article. There needs to be a reason to create a page beyond why not?. Johnuniq (talk) 22:18, 31 March 2020 (UTC)
No, don't do that. If the alternate spellings are genuine ones (or common mispellings), that's fine. But we don't create redirects for every possible phonetic spelling of a name. Black Kite (talk)22:25, 31 March 2020 (UTC)
Please just let search engines do their job of finding articles based on the search term, rather than create redirects for misspelled words. isaacl (talk) 04:46, 1 April 2020 (UTC)
Oh, these are not misspellings; all of them are valid spellings, because there are many different systems to transliterate words from some non-Latin alphabets (see e.g. Transliteration of Persian). Although there may be no hits in Google, they are still useful because a reader might type one of them in Search. --Z17:26, 2 April 2020 (UTC)
Representations using ASCII characters, that is, common transliterations (for example, Pele also redirects to Pelé while Kurt Goedel and Kurt Godel redirect to Kurt Gödel).
A transliteration with zero Google hits is not common, and you may have hundreds of them. I tried five random again. All five had zero hits. PrimeHunter (talk) 01:18, 3 April 2020 (UTC)
Here "oe" and "o" are both common transliterations for "ö". Similarly, there are common transliterations for many letters of some non-Latin scripts that their combination results in hundreds of forms for a single title, all of which are more or less valid. Since redirects had not been created on this scale I started a new discussion at Wikipedia_talk:Redirect#Redirects_from_alternative_transliterations and clarified the situation to see more comments. --Z14:12, 3 April 2020 (UTC)
problems with duplicate definitions from template-defined references and notes
Latest comment: 4 years ago8 comments3 people in discussion
Looks like there's a problem with templates that produce references when they're included into another article with {{excerpt}}.
As examples, I'm trying to figure out the three duplicate reference definitions given in the "2020 Democratic Party presidential primaries" article. They seem to come from the "NVresults" section of the "2020 Nevada Democratic caucuses" article. The note marked duplicate is defined there with {{efn}}, and the note is used in that article, too, with {{efn}}.
I don't think there's anything wrong with these articles themselves, and the duplicate reference definition errors are the result of a bug. Is there something I'm missing? How can this be fixed? -- Mikeblas (talk) 15:50, 2 April 2020 (UTC)
User:PrimeHunter/sandbox7 displayed 'Cite error: The named reference "2020delegatesNVD" was defined multiple times with different content' when it contained [18]:
Module:Excerpt isn't used in these examples. {{Excerpt}} includes: {{#if:{{#lst:{{{article|{{{1}}}}}}|{{{fragment}}}}}|{{#lst:{{{article|{{{1}}}}}}|{{{fragment}}}}}|...}}. This tests whether the requested section produces a result and returns the result in that case. The test apparently causes MediaWiki to duplicate a reference in the output but in a way which triggers the "different content" error. That sounds odd to me. Ping Sophivorus who made {{Excerpt}} and added the test in [20]. PrimeHunter (talk) 15:00, 3 April 2020 (UTC)
Fixed. It's a pity because that fix loses us the functionality of detecting articles with broken excerpts, but this is the second issue caused by that #if, so I think it's better to remove it for now (eventually I should be able to restore the broken excerpt detection by improving the Module:Excerpt). Please let me know if you notice any other issues with excerpts, thanks! Sophivorus (talk) 17:44, 3 April 2020 (UTC)
Redirect not functioning properly
Latest comment: 4 years ago9 comments5 people in discussion
I've come across a redirect that appears not to be functioning properly, even though I can see nothing wrong with it from a technical standpoint. Yaw Kunadu was being erroneously detected as an uncategorized article, which isn't supposed to happen to redirects — and regardless of whether I type it into the search bar or click on it from a link, instead of actually taking me to the redirect target like a redirect is supposed to, it just takes me directly to the redirect itself as a page, and even then I just see the text "1. REDIRECT Yaw Konadu" in regular font size rather than the black arrow followed by a larger, bolded iteration of the target title that looking directly at a redirect is supposed to produce (and still does if I try going directly to any other redirect besides this one). I'm not seeing any obvious error in how the redirect is coded, however, so I'm not sure why any of this is happening.
I've already tried a temporary deletion and recreation, which didn't work — in the meantime, I've had to add it to Category:Temporary maintenance holdings to get it off the uncategorized articles list, but I wanted to ask if somebody with more technical skill can look into it. Thanks. Bearcat (talk) 16:38, 3 April 2020 (UTC)
@John of Reading: I am not really familiar with redirects, but rather than a mysterious "space" character, wasn't this just a matter of needing to insert the "#" character at the beginning of the line? Fabrickator (talk) 17:40, 3 April 2020 (UTC)
@John of Reading: Thanks for the response. I made the mistake of looking at the "rendered" version of the page, thus the "#", being interpreted as a numbered list, showed up as a "1". Non-breaking space? That makes sense, it's not just the non-breaking space that could do this, there are things like "zero-width space", etc. Some editing programs (e.g. "vim") have a way to display such "invisible" characters. Fabrickator (talk) 18:27, 3 April 2020 (UTC)
Latest comment: 4 years ago7 comments5 people in discussion
Is there any good way to find all the external links in articles to a particular website? The data is in the externallinks table, but the table is huge and not indexed in a way that's useful for this. Where I'm heading is trying to find eBay auction spam. -- RoySmith(talk)16:05, 3 April 2020 (UTC)
I do not fully trust the externallinks database to be complete in my experience it misses links. It is populated by a program that searches the wikitext and extracts what it thinks are URLs, but it ends up missing some things (it might miss URLs in some templates for example). Insource (Elasticsearch) does an excellent job, but is limited to 10,000 results. For 10k+ links that must be accurate parsing the XML dump is accurate and scales, but slow. -- GreenC21:35, 3 April 2020 (UTC)
You're probably more likely to miss links by trying to search the wikitext. If you can find a reproducible test case where a link doesn't make it into the externallinks table, you should file it as a task in Phabricator so it can be fixed. Anomie⚔21:52, 3 April 2020 (UTC)
Query suggestion:
USE enwiki_p;
SELECT el_to, page_namespace, page_title
FROM
externallinks
JOIN page ON el_from = page_id
WHERE el_index LIKE "%//gov.census.factfinder2.%"
Latest comment: 4 years ago27 comments7 people in discussion
Twice today I have been automatically logged out of my account whilst using WP:REFILL. It's a separate website used to fill in bare references, when you click the 'preview/save' button which brings you back to Wikipedia to review/save the edit. But when I click it and come back to Wikipedia I get logged out. Any idea why? GiantSnowman17:05, 21 March 2020 (UTC)
I am sorry this is happening to everyone. It hasn't done this to me (yet fingers crossed-I use Mozilla.) Please note the info in this thread above Wikipedia:Village pump (technical)#Requests for help regarding refill indicating that no one is running refill at this time. There is still the fact that refill2 was updated to refilla a couple weeks ago but I can't seem to find anyone who knows anything about that. Reflinks and citer are still available if things get too frustrating for you. MarnetteD|Talk20:39, 21 March 2020 (UTC)
It is frustrating, especially as I am using 2FA and have to waste a scratch code every time! I'll go back to Reflinks for now... GiantSnowman20:52, 22 March 2020 (UTC)
Giant and Moxy are you using the old "refill" or the updated "refill2" also called "refilla". I just stumbled on a thread that mentions the problem as being with "refill" - I assumed (and yes I know that is always fraught with danger) that everyone was using the newer version in my posts. If you are using the former you might try the newer version. If you are using the update then in the words of the immortal Emily Litella "never mind" MarnetteD|Talk07:43, 24 March 2020 (UTC)
Yeah that is the original version Giant. If you look at {{Bare URLs}} you will see the parenthetical (One such tool to fix up some types of bare URLs is reFill, see also its documentation) just click on the word refill and the tool will start working. One nice thing about the new tool is that it will convert bare urls in the external links section. Now it doesn't put them into cite templates but it does put them in brackets and add a title. One drawback is the first screen it goes to says "pending waiting for an available worker" sometimes it hangs up on that screen. If you see that message for more than 10 seconds just hit ctrl+F5 and that usually gets things moving. I should also mention that this need to click on the word "refill" to get the tool working is why I ask people to not put this bare url template in a "multiple issue" tag - the parenthetical does not show when this is done so you have to go in and move the template out of the MI so you can fix the bare urls in the article. Now I know that every persons interaction with these tools is different so parts of my explanation might not make sense. If that occurs let me know and I will try to explain it differently. MarnetteD|Talk16:12, 24 March 2020 (UTC)
The ReFill I get by clicking on the link through {{Bare URLs}} just attempts to 'fix' that template page. Where/how can I use that tool to input pages I come across? GiantSnowman17:47, 24 March 2020 (UTC)
Giant you need to add that template to whatever article you are working on and then click on the link. It will then go to work on that page. Here is one as an example Joint Navigation Warfare Center. I've never seen a start page for refill2 (like the original tool has) where you would cut and past the title of the article though there might be one. MarnetteD|Talk19:13, 24 March 2020 (UTC)
GiantSnowman, sure if you trust it. I hear it has lots of problems with non-google accounts. You should be using a 2FA app to login by typing in a 6 digit OTP generated by the app. —CYBERPOWER(Chat)17:16, 30 March 2020 (UTC)
Adding a template to a page and then clicking a link is too much unnecessary work - and based on Pinguinn's commnents I'm not using any form of ReFill until it's fixed. GiantSnowman12:39, 25 March 2020 (UTC)
I'm not sure how two clicks is too much work Giant. I've been doing it for over three years and it is a breeze. As I've mentioned before refill is not being maintained so it will not be fixed. Thanks for your help in working on these up until now though. MarnetteD|Talk05:13, 28 March 2020 (UTC)
Hi, I was wondering if someone would be interested to take a look at this. It might be helpful in one way of the other. Thank you for your time. Lotje (talk) 07:17, 30 March 2020 (UTC)
Latest comment: 4 years ago4 comments4 people in discussion
More recently, the issue was discussed at Talk:2019–20_coronavirus_pandemic#Length, but the navboxes at the bottom continue to crash (as of now, I see two #invoke:Navbox errors and three normal navboxes). Could we install some preventive feature, such as a warning on transclusion limit in the editing mode or prevention of saving, asking either to trim the text or the navboxes themselves? Brandmeistertalk08:58, 4 April 2020 (UTC)
The navbox is over 500k, so takes up a quarter of the allowable article transclusion size, and this will only grow as the crisis continues. Rather than split many articles, perhaps there is a better way of handling the navbox. Make a smaller one for the major articles, perhaps with a note and a link to the full navbox. — Jts1882 | talk09:39, 4 April 2020 (UTC)
Latest comment: 4 years ago15 comments9 people in discussion
Hi. I have no idea if I'm in the correct place, but would it be possible to run something through my account to tell me which Wikipedia shortcuts I use most often? I have decided that strings like WP:GNG probably won't mean much to a new editor, so I want to create a series of templates where I all I have to do is use braces instead and it comes out with something sensible (e.g. {{WP:GNG}} will produce our general notability guideline). (Please ping when responding as I use email rather than my watchlist.)--Launchballer17:14, 2 April 2020 (UTC)
That's a good point. I left some noinclude/includeonlys in WP:GNG but as it would require a number of if statements within the article, they would have to be at, for example, {{WP;GNG}}, which would effectively mean it would be in mainspace.--Launchballer19:32, 2 April 2020 (UTC)
Launchballer, you could also create a template at no shortcut name given (help) and have it be a switch statement that converts terms like GNG into more sensible terms, so usage would be {{WP|GNG}}. – BrandonXLF (talk)19:56, 2 April 2020 (UTC)
Sounds like a plan, except isn't there a limit on how many terms on how many terms a switch statement can take?--Launchballer20:12, 2 April 2020 (UTC)
@Launchballer: If there is one, you're not likely to hit it. I tested out one with over 2000 terms in my sandbox and it worked fine. However, if you're worried about creating a huge template, using something like {{WP/GNG}} would get around that (and you could then have {{WP}} contain {{ WP/{{{1}}} }} so that the {{WP|GNG}} format would work). --Ahecht (TALK PAGE) 20:25, 2 April 2020 (UTC)
@Ahecht and Launchballer: Please no, it's better to keep everything in one template to make it simpler to maintain. We can use Lua if we reach a limit with the switch statements or we can use nested switch statements, but I don't think we're going to be anywhere near that limit. – BrandonXLF (talk)21:54, 2 April 2020 (UTC)
Simple linking templates like some are suggesting above are usually deleted at WP:TFD (in Template space at least). To answer the original question, there is no tool I know of that can check to see whether you like to use the phrase WP:GNG, but you can perhaps try Special:Search. --Izno (talk) 00:04, 3 April 2020 (UTC)
Here's a related idea: I'd love to see a template where you can type a policy shortcut and, if that policy has a nutshell, automatically transcludes it into a tooltip, so that it'd be possible to read it without clicking. It'd look something like this: {{WP|BOLD}} creates WP:BOLD. Would that be possible? {{u|Sdkb}}talk18:14, 3 April 2020 (UTC)
@Guywan: That looks great! Given the number of shortcuts, though, I think to be useful it'd need to automatically fetch the {{nutshell}} from the linked page. One way I can think of to do that would be to wrap the nutshell in a fragment at Template:Nutshell and then excerpt that in the template. Is there a better way to do it? {{u|Sdkb}}talk19:16, 4 April 2020 (UTC)
@Sdkb: That wouldn't work. Template:Excerpt uses the #lst parser function, which has the following limitation: Section tags cannot themselves be transcluded in order to work on other pages.[21] A module could be written specifically to get nutshell text from a given page. Might be more trouble than it's worth, though. —guywan (talk • contribs)20:57, 4 April 2020 (UTC)
Latest comment: 4 years ago4 comments2 people in discussion
I have a problem. My wiki has a page "Windows tip" and 2 heading name "windows can't sleep" and "Windows wake from sleep". A search "windows sleep" only bring "Windows wake from sleep" then the results come from another page. How to list more than 1 result from a wiki page?
PS1: I can code a bit, so if this feature not available I can contribute. After read the code, I see the logic is somewhere at
Wikimedia uses CirrusSearch built on top of Elasticsearch, either way, you're mixing up the intent, it's for searching pages, the result is always meant to be one page, it's not something like a codesearch tool (like Hound) where you try to find instances of some snippet, it's designed to give you the correct page. --qedk (t愛c)14:00, 5 April 2020 (UTC)
Hi, can you suggest the best practice for me? Should I split the "big" page into multiple small pages?
BTW, I want to know where the logic is handled for searching since breaking 300+ pages to smaller ones need a lot of effort.
If you want sections in that manner, yes you should split the page, since (AFAIK) Elasticsearch indexes on page-content basis, so multiple hits on the same page will still be the same page, Elasticsearch is not intended to be solely for text-searching afterall - I don't know if you can reconfigure it for your purpose. Maybe Anomie or TheDJ will have more ideas about this. --qedk (t愛c)18:00, 5 April 2020 (UTC)
No categories
Latest comment: 4 years ago3 comments3 people in discussion
Hello. Is there a tool to find all articles that have no categories but they may be in a Hidden category? This page don't include pages that are categorizes only in hidden categories (I think).Xaris333 (talk) 14:33, 4 April 2020 (UTC)
Latest comment: 4 years ago10 comments6 people in discussion
Can anyone tell me if some files used by Monobook have changed recently? I have bad eyesight so Ive been using the site while zoomed in for the last five years or so. But I did it with CSS, not a browser zoom, and was able to make only the main text of the article magnified, while still having a normal sidebar and topbar. That for whatever reason no longer works, so I have to zoom the whole site, which means less content fits on the page. I might be literally the only one who's doing this but I want to try to find help if I can't figure it out on my own. Thank you —Soap—15:08, 4 April 2020 (UTC)
Soap It's possible the skin CSS changed; I know there's been some work in that area of late by multiple separate developers on multiple separate tickets. I'd just recommend checking to see whether the rules are more specific now or if the identifier you're using to add your CSS has changed. (It is also possible the rule in your user CSS isn't loading for some reason; a malformed rule somewhere earlier in those files would do it, or not getting the CSS from the server would also do it.) --Izno (talk) 16:26, 4 April 2020 (UTC)
Thanks, I did that, and it worked in the sense that it magnified the body text and not the sidebars. However it still looks nothing like what it used to .... the words are for some reason more densely packed than usual, and hard for me to read. Wiktionary and Commons are still looking normal for me, which is strange, because my Commons CSS isnt using any zoom at all, so I might be able to eventually figure out what's going on although it is beyond me right now. And if I cant figure it out, I guess I can get used to just using the browser zoom at 150%. —Soap—01:41, 5 April 2020 (UTC)
@Graham87:Thanks, but it didnt help .... when I did that and then restored the new CSS idea it just reverted to the densely typed text layout where the font is 150% but the letter and line spacing is not. Im not giving up, but Im probably going to have to just use the whole site at 150% for the time being. I dont see any recent changes to MediaWiki:Monobook.css, anyway , and so far I seem to be the only one with this specfic problem. It's possible some other file somewhere changed, but I wouldnt know where to look .... also I have used 2 PC's in the past few days and both had the same thing happen, so I know it isnt a browser problem either. Thank you all for your help, —Soap—03:39, 5 April 2020 (UTC)
@Soap: One last try! Perhaps you need to adjust line height aswell, like so:
Thank you, that worked. I dont know why, because I've been using the font-size parameter without the line-height parameter for years on this site, on other wikis, and on my own personal website, and it has always increased the line and letter spacing in tandem with the font size until now. With this new code, Wikipedia is once again easy for me to read without having a huge sidebar and images that dont fit where they're intended to. My watchlist, the edit window, and pages like Special:Contributions are still small, but I rarely use my watchlist and I think with my experience using CSS I can eventually figure out a workaround for the contribs and edit window page. I'm using WikEd, which apparently loads an external editor, so CSS may not help me there ... but it looks like I did some work with Javascript a few years back and could bring it back again. Also, the text was zoomed in in the edit window until a week ago, so maybe CSS can do it after all. Many thanks, —Soap—18:08, 5 April 2020 (UTC)
Latest comment: 4 years ago2 comments1 person in discussion
I have been keeping the stats updated on 2 state articles but I do not know how to fix the state county-by-county maps. I requested help on the COVID-19 WikiProject page but no one has responded yet. I also posted over on Commons to two editors who have worked on state maps but no response there yet either. Please, O Wikipedia Technical Mavens, could someone update these 2 maps? The last date they were touched was March 27th!!! I'll list the articles, the template name, and where to get the stats/information below:
REFILL hasn't worked for me in a long time. Is there an alternative tool? I only use it to turn bare URLs into a full citation which I then paste onto a page. Normal Op (talk) 15:23, 5 April 2020 (UTC)
Latest comment: 4 years ago2 comments2 people in discussion
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Problems
There was a problem with user pages not being shown properly on desktop. This was because of a bug. It will soon be fixed. [22]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 7 April. It will be on non-Wikipedia wikis and some Wikipedias from 8 April. It will be on all wikis from 9 April (calendar).
Future changes
MediaWiki will use a newer version of Unicode. Some characters that did not have an upper case equivalent before do now. Titles beginning with one of these characters will be moved. A list of these titles can be seen on Phabricator. The titles will be renamed by the user Maintenance script. This will start on 13 April 2020. You can rename them before this if you wish and the new title can be different from the one the script would rename it to. [23]
A note about this "Future changes" item for people who know about double redirects: I poked at a few of the pages listed in the phab ticket, and many of them are redirects. We may want to have a savvy person do these redirect moves here on en.WP. – Jonesey95 (talk) 20:26, 6 April 2020 (UTC)
We should be fine. I looked at them all on Saturday, and (after I fixed four) all should either move to the new title without issue or are redirects to the same place as the new title. Anomie⚔23:42, 6 April 2020 (UTC)
Latest comment: 4 years ago3 comments2 people in discussion
Was wondering if someone familiar with templates might mind taking a look at the ones like Template:2004 Summer Olympics Greece men's football team roster being used in this article. For some reason, "view.talk" links to the templates' pages and corresponding talk pages are being transcluded into the article body. The same links can be found on each template's page. I've seen "V.T.E" links in the syntax of navigation templates, etc., but I've never such links transcluded into the body of an article in this manner before. -- Marchjuly (talk) 00:07, 7 April 2020 (UTC)
System font (unchangeable sans-font in most of the modern gadgets)
ComicNeue
OpenDyslexie
I feel better readability and comfort of reading with proportional serif especially Garamond, Goudy Old Style, Times new Roman etc.
So I plea to add these fonts in output font settings.
I'm not sure where in Wikipedia's code the font options reside, nor what the potential benefits/downsides of introducing new options are. The folks at the technical pump might be able to point you to a place monitored by more knowledgeable editors. I tentatively support the idea of a serif font option, although I wouldn't want it to be the default. {{u|Sdkb}}talk00:52, 7 April 2020 (UTC)
@Sdkb: yes I meant as option; given that Wikipedia already provide options for font change, few more font styles should be introduced. Not telling of a drastic new feature. RIT RAJARSHI (talk) 07:05, 7 April 2020 (UTC)
Assuming that image is a screenshot of your own preferences, you've switched WebFonts on. If you go back to the defaults (i.e. uncheck the "Download fonts" box), then Wikipedia should display in whatever your browser has set as its display font. ‑ Iridescent09:27, 8 April 2020 (UTC)
@Nehme1499: not really, you can dump the names with something like AWB and paste them in - but it won't be dynamic for future category member changes, for that you would need phab:T3710 to happen. — xaosfluxTalk18:54, 7 April 2020 (UTC)
Latest comment: 4 years ago3 comments2 people in discussion
Hello, using {{Multiple image}} in articles causes the saved edit/latest revision don't show in the guest/not logged-in state, there is no problem when user/visitor is logged-in, I have done all the Wikipedia:Purge instructions in the guest/not logged-in state and nothing has changed, I have tested this using Firefox and GNOME Web, I had this problem in Visual pollution 8 days ago and reported it at Template_talk:Multiple_image#Does_not_show_in_an_article and User:andrybak did a dummy edit and problem was solved after it, but it is still exist in every newly edited article, the last one is Red Hat and even one dummy edit didn't solve it and I think making dummy edits is not a solution even if it really solve the problem. Thanks--Editor-1 (talk) 12:12, 7 April 2020 (UTC)
The owner of the script is still active (last edits few days ago) so I think Interface-admins may decline to act, but you can request so at their noticeboard if you think it's urgent. – Ammarpad (talk) 11:38, 7 April 2020 (UTC)
Sorry I didn't look at the script itself. It seems it simply imports the script on Lingzhi2 so shouldn't be a problem of itself. Nil Einne (talk) 06:40, 8 April 2020 (UTC)
Latest comment: 4 years ago9 comments3 people in discussion
Hello everyone!
Can someone help by guiding me a bit on the actions I should take? I'm an admin at SqQuote and recently we've started having our good and featured articles. How do I make them appear here? Also, how do I make it that they appear with a small badge on their homologous names on different languages according to their status, just like they do in Wikipedia? Basically any help regarding that special page and that process in general is appreciated so I can understand better how to use it.
@Klein Muçi: page "badges" are stored on wikidata, for example on sqwiki w:sq:Saturni is a "good article" and this can be seen on the record here: wikidata:Q193. Does sqq have any pages that actually have a badge, is so can you provide the wikidata Q number to check? — xaosfluxTalk12:03, 8 April 2020 (UTC)
@Xaosflux: Hahaha, I didn't want to bother you more than I've already done, that's why I wrote here but you found me again. haha :P We have 5 good articles, located here. What exactly do I need to do to them in Wikidata to show like they should? Can you provide one practical example on any of them, what I need to do, so I can follow that on the others? - Klein Muçi (talk) 12:13, 8 April 2020 (UTC)
I mean, if I'm not wrong, all we have to do in Wikipedia is to put the appropriate template on the article and the rest of the process is handed automatically in Wikidata. At least that's what I've always thought. Maybe I've been uninformed but as far as I know, not many of our SqWiki admins do ever go to make "extra work" at Wikidata. - Klein Muçi (talk) 12:20, 8 April 2020 (UTC)
@Klein Muçi: OK, so I only know half of this - but it is the part you originally asked for :) I manually updated the wikidata item for one of your pages, and it is now working: see here. As far as the "how to update wikidata" better part - maybe anyone else will chime in before I research it. — xaosfluxTalk13:15, 8 April 2020 (UTC)
@Xaosflux: Thank you! So you update that manually at the literal badge beside the article's title at the individual project. That's better than nothing. I hope I can find a way to do it automatically from inside the project itself but anyway, thank you again! :) - Klein Muçi (talk) 13:22, 8 April 2020 (UTC)
@Xaosflux and QEDK: Oh, I see. I'm surprised I didn't know about the existence of such bot. Maybe because I haven't dealt specifically with Wikidata much. I'll talk with Pasleim. Thank you! :) - Klein Muçi (talk) 14:40, 8 April 2020 (UTC)
Cannot click on reference links any longer
Latest comment: 4 years ago10 comments4 people in discussion
I am not sure why, but when I click on reference links when logged in, it no longer goes to the reference. It must be something I have engaged, but I cannot for the life of me figure out what it is.
Literally any page. I clicked on Random article and got: Glenn Switkes. The first reference is in the first sentence. I click on it when logged in and nothing happens. jps (talk) 17:22, 8 April 2020 (UTC)
Actually, that's not precisely true. My browser (Chrome) shows the link as a tooltip at the bottom. But there is no navigation to that part of the page any longer. It's as if there is something preventing my browser from following the link. jps (talk) 17:24, 8 April 2020 (UTC)
It sounds like you have ReferenceTooltips enabled (I think it's a beta feature). You can't click the reference itself after opening that tooltip; there is a link in the tooltip to take you to the reference. It does sound like there is an issue though that probably should be logged in Phabricator; the tooltip should be displaying at/near the reference itself. In the meantime you can probably turn the feature off. --Izno (talk) 17:27, 8 April 2020 (UTC)
Well, I did have ReferenceTooltips enabled, but I disabled it and the same problem persists, so it's not that. Good thought, though ! jps (talk) 17:30, 8 April 2020 (UTC)
Latest comment: 4 years ago4 comments3 people in discussion
It looks like categories added by authority control are unhidden for some reason in articles, as seen eg. in Richard Sorge where there's a stack of "Wikipedia articles with foo identifiers" categories. Could we make such categories hidden? Also, those categories currently contain an error "Lua error in Module:Pages_with_authority_control_identifiers at line 116: attempt to concatenate local 'txTOC' (a nil value)". Brandmeistertalk10:30, 9 April 2020 (UTC)
An WP:ITSTHURSDAY for those using the Timeless skin: The font size was recently increased from the standard 13px (as used in Vector, Monobook, Modern, CologneBlue) to a larger 15.2px: phab:T247325. To restore the normal font size, you can add:
Latest comment: 4 years ago1 comment1 person in discussion
Hey there, I wondered if someone can offer any help or advice about my recent troubles with Wikipedia.
Since yesterday's (9th of April) evening all Wikipedia page previews (the ones that appear when you hover the mouse over a link to the article) disappered completely when using my tablet (the browser on the tablet is Chrome 55.0.2883.91).
The previews were there and worked properly only hours before that, I never had any troubles with them previously, and I am certain I didn't change anything whatsoever in my site settings, browser settings or else.
Now they refuse to pop-up on screen regardless of which language-based Wiki I use or if I'm switching between mobile and regular version.
Since I never turned them off using the cog icon, the option of turning the preview back on doesn't appear in the lower section of the page, as suggested in the FAQ.
Cache emptying didn't help. JS is turned on and the search field drop down menu with results works just fine.
@DragonOfWisdom: The page has not yet been reviewed the new page patrol. That's why it's noindexed. The category only contains pages that actually have __NOINDEX__ on them. In this case, the page is being noindexed by the underlying software. SD0001 (talk) 13:38, 10 April 2020 (UTC)
Specifying locations in articles?
Latest comment: 4 years ago3 comments3 people in discussion
I've been mulling over an annotation tool architecture where there's a server side that does all the hard work, and a lightweight client side in javascript that just handles the U/I. For example, the server side might come up with a suggestion that a specific word might be replaced with a different word. What I'm thinking about is how the two sides communicate the location in the text. One obvious, if simplistic way, would be a character offset from the beginning of the wikitext. Another might be some kind of jquery-ish specification for how to navigate a parsoid tree. The idea is that the location information is divorced from the presentation (very much unlike how diffs work now) so different kinds of clients could make use of the same interface.
I don't fully understand the question or the use case, but I can nevertheless point you at a couple places. https://phabricator.wikimedia.org/T149667 is a proposal / idea that hasn't been implemented for generalized annotations for different use cases. But, as far as Parsoid goes, Parsoid maps DOM nodes to wikitext ranges that generate them. So, when Parsoid analyzes the DOM and finds a DOM node with a specific kind of lint error, it sends that lint error with the wikitext range for that node to a database. So, on the front-end, JS code can use the wikitext offsets to highlight the substring in the editor. Not sure if that helps / answers User:RoySmith's question.SSastry (WMF) (talk) 17:34, 10 April 2020 (UTC)
Parsoid's effects on talk pages
Latest comment: 4 years ago1 comment1 person in discussion
mw:Talk pages project/replying isn't the visual editor, but it is currently using Parsoid, just like the visual editor. I've just finished looking at all (583) diffs at arwiki, frwiki, huwiki and nlwiki, where mw:Extension:DiscussionTools new Reply feature was offered as a Beta Feature there last Tuesday. In contrast to User:Hawkeye7's experience of 2% of talk pages here having some sort of error, the number of dirty diffs I found was maybe a little lower than that, and all the changes were small (example; almost all of them involved removing an extraneous space). I found zero serious problems. There's no guarantee that this wiki would have the same experience, but I'm taking it as good news overall. If you're interested in this project, please put Wikipedia:Talk pages project on your watchlist. If you're interested in trying it out, please do so (you will have to opt-in to the "Discussion tools" Beta Feature at one of those four wikis), and then ping me to tell me what you thought about it.
On the original subject, since our conversation about it back in January, I think there has only been one request to have the visual editor enabled in talk pages, and I don't think that's going anywhere. Whatamidoing (WMF) (talk) 18:43, 10 April 2020 (UTC)
I find it easier and less error-prone to add these templates without a date. The bot always adds a typo-free date for me. – Jonesey95 (talk) 13:33, 11 April 2020 (UTC)
Or even better, just subst the template - this even avoids the redirects, eg {{subst:linkrot}} gives the clean {{Cleanup bare URLs|date=April 2020}}. SD0001 (talk) 15:06, 11 April 2020 (UTC)
Question about notifications
Latest comment: 4 years ago12 comments6 people in discussion
In Preferences, I can choose to be notified "when someone links to a page I created from another page." Is there some way for me to receive some kind of notification when someone links to a page I did not create? For example, it would be interesting to be notified each time someone links to the Billboard article, because it's likely that the link should be changed to Billboard (magazine). Thanks! GoingBatty (talk) 19:05, 11 April 2020 (UTC)
@GreenC: That looks great! I'm using Windows 10, and I see your code requires a "Unix system that has GNU Awk 4.0+". Any recommendations for basic instructions on how to set this up on my Windows PC? Thanks! GoingBatty (talk) 20:42, 11 April 2020 (UTC)
GoingBatty, not easily with Win10. I should make it a web service anyone can subscribe to on Tools. But that is more time than I have right now. What I can offer is to install it on Tools (without the subscribe/unsubscribe form). There will be a "stop button" page so you can halt and restart it any time you want. If that is of interest let me know. -- GreenC21:26, 11 April 2020 (UTC)
awk is one of the most awkward programming languages ever invented (except possibly APL). I only ever used it if I couldn't get sed (possibly with other filters in the pipeline) to do what I wanted. --Redrose64 🌹 (talk) 21:57, 11 April 2020 (UTC)
Redrose64, Awk was a great tool in its day. It provided just enough power to write quick report generation type tools without having to dive into compiled languages. But, people pushed it beyond its comfort zone and you ended up with a mess. These days, anything people used to do with awk is probably better done with your choice of Python, PHP, Ruby, Go, or I guess even Perl. -- RoySmith(talk)22:19, 11 April 2020 (UTC)
I used awk extensively, from the open-source cygwin package. It wasn't awkward for me, just (refreshingly) different. Jmar67 (talk) 00:47, 12 April 2020 (UTC)
Awk has two side or faces. There is the CLI which is arcane knowledge. The people into that are creating super concise command-line mini-programs that feed through pipes, it can be quite elegant and satisfying. Then there is working with .awk script files with functions like a normal programming language. This is basically a simple version of Perl or JavaScript (both took syntax and ideas from awk) with a small standard library like match(), split() etc.. that is familiar to any programmer. I use awk because of the intuitive syntax, portable, very fast execution, very low memory footprint, stable (no garbage collection), no system libraries or environment (interpreter is a single executable), fast development speed. For more complexity I use Nim, which is basically Python but at C speed, more features and an easier syntax. -- GreenC01:18, 12 April 2020 (UTC)
@GreenC: If you can install it on Tools in a way that users would be able to define their own pages (like you have in bw.cfg), then yes please! Thank you! GoingBatty (talk) 23:39, 11 April 2020 (UTC)
@GoingBatty: to set this up I will need an email address, where to send the report to. I don't see a way to contact you can you email it to me? -- GreenC01:24, 12 April 2020 (UTC)
It seems like the issue got fixed? I almost had a mini-heart attack when I thought my changes didn't go through for a large article. Tenryuu 🐲 ( 💬 • 📝) 06:32, 12 April 2020 (UTC)
The chat on IRC #wikipedia shows that site ops were aware of the outage and tried to fix it as soon as possible. It occurred because of a problem at the Eqiad cluster, which is in Ashburn, Virginia.[26]--♦IanMacM♦(talk to me)06:41, 12 April 2020 (UTC)
template parameters vs ref tag arguments
Latest comment: 4 years ago6 comments2 people in discussion
Lots of templates generate refs which are transcluded by articles. This is messy, but it happens a lot. I wondered whether the mess could be lessened by putting such refs into groups named for the template where they are generated. This would need to be done in the template generating them and, in order to do this, that template would need to discover what its name is or to have that information passed to it.
Templates can discover their names with {{TEMPLATENAME}}, and information can be passed to them via invocation parameters. As far as I can see, however, neither of these can be used to set the group= parameter of refs generated in the template. I'm guessing that this is a scoping issue and that there is no solution or workaround.
If you assign a group based on the template name, then your references will look like [Template name here 1] rather than the more usual [1]. It would probably be better to just give the references names that are unlikely to collide with reference names in articles, as was already suggested to you at WP:VPP#Refs in templates. Anomie⚔22:35, 8 April 2020 (UTC)
Thanks for the info and advice. I understood it at VPP and I understand it here. Does anyone have a responsive answer? Wtmitchell(talk) (earlier Boracay Bill) 13:29, 10 April 2020 (UTC)
To elucidate a bit, I don't see non-cosmetic note names as a disqualifier of discussion at this point. Without getting down in the weeds, I'm wondering whether, if a simple template can't be used for this (I'm still holding onto hope that it can, pending info from here), I'm wondering whether it might be possible using WP:LUA. If it's not possible, it's not. If it is possible but has cosmetic problems, it might be useful to look at whether the cosmetic problems can be dealt with. Wtmitchell(talk) (earlier Boracay Bill) 17:25, 10 April 2020 (UTC)
On this page history, you will note that
three consecutive edits in February and March 2020 have been redacted out of the history. Okay, I understand that, but now look at this. First
select an edit a bit earlier (say this one).
Click on "Next edit →" and it works. Click on "Next edit →" again and now you get "You cannot view this diff because one or both of the revisions have been removed from the public archives."
But now click "Next edit →" again, and you get: "One revision of this difference (940771782) was not found" -- and no forward link this time. This is not helpful.
Since three edits were removed, I expected one of two things. (A) The "You cannot view this diff" screen might repeat twice, and then "Next edit →" would take me to this edit -- the next one that I am allowed to see.
Or better yet, (B) on the "You cannot view this diff" screen, instead of a non-useful "Next edit →" link it could have had "Next publicly available edit →", which would take me directly to the above one (or a "Next edit →" link that would have that behavior).
Latest comment: 4 years ago9 comments7 people in discussion
Hi. You know when you have a list and there is the option to bullet each line and saves you having to do it manually? I was wondering if there is a way of commanding a list to add a dash at the end of every line instead of having to add one to each line manually?† Encyclopædius10:57, 10 April 2020 (UTC)
@Encyclopædius: You can use the search and replace tool (in the source editor > Advanced tab, the rightmost icon) for this. Do a regular expression search for $ (which matches the end of line), replace with - (or whatever you want). This will add a dash at the end of every line. If you want to insert a dash at the end of all non-empty lines, search for (.)$ and replace with $1 - . SD0001 (talk) 12:53, 10 April 2020 (UTC)
SD0001, Am I the only one who does stuff like this by copying the wikitext to my desktop, fixing it up with emacs, and pasting the result back in? -- RoySmith(talk)15:29, 10 April 2020 (UTC)
@RoySmith: no even I do sometimes do that (I dunno if the search-and-replace tool supports all regex syntax like lookaheads and lookbehinds). Not an emacs users, I use VS Code, maybe someday I'll install emacs and see what's so hyped about it! SD0001 (talk) 07:51, 11 April 2020 (UTC)
I've been using wikEd, which doesn't require copying text and is useful just for syntax highlighting when I'm copy editing a section with more reference markup than prose. Its GUI rendered better when I used it with Firefox under Windows 7 than it renders now under Safari on my ancient iMac, but it's still decipherable and is quite sufficient for the regexing I want to do. Dhtwiki (talk) 06:07, 12 April 2020 (UTC)
Latest comment: 4 years ago5 comments4 people in discussion
I propose that MediaWiki:Tooltip-ca-talk (the tooltip that displays when you hover over the talk tab) be changed from "Discussion about the content page" to "Discuss improvements to the content page". "Discussion about" → "Discuss" switches the message to a more active voice mirroring the other tabs and makes it shorter. Adding "improvements to" could go a long way toward remedying the WP:NOTAFORUM issues pervasive on talk pages. Ideally, I'd also like to change "content page" to something more colloquial, but that would require the ability to adapt this message based on the namespace, which seems complicated. {{u|Sdkb}}talk17:46, 10 April 2020 (UTC)
@Ammarpad: Yeah, I plan to do that. I've submitted a bunch of edit requests recently, though, and been lightly reprimanded on a few of them to establish consensus before making the request, and given that this is a change that will (in an extremely minor way) affect every page on Wikipedia, I wanted to come here first to gather a few "yeah, that sounds like a good idea"s. Can you or anyone else here chime in to offer that? {{u|Sdkb}}talk06:32, 11 April 2020 (UTC)
How do bot edits get noted in page history? Specifically, I'm looking at this edit which doesn't appear to have been recognized as a bot edit by whatever automation tags old drafts for WP:G13. -- RoySmith(talk)14:34, 12 April 2020 (UTC)
I think the bot tag is added to an edit if done by a bot and if the bot requests it. Possibly this bot does not request it because it's really a change that editors should note (it removed a deleted image which anyone interested in the article should investigate). Johnuniq (talk) 02:33, 13 April 2020 (UTC)
No, "bot" is just a user group like "autoconfirmed" or "admin". All accounts in group "bot" have their edits tagged as bot. NixinovaTC02:49, 13 April 2020 (UTC)
What Johnuniq said is correct. However to use botflag, you don't necessarily need to be in bot group, you just need "bot" userright which is a separate thing. In some wikis there are human users with the right. There's also a minor version of it given to to admins here (markbotedits) which allows them to mark edits as "bot" but constrained to only rollback requests. Botflag is just like minor edit flag, even if you have it you can decide to use or not use it.– Ammarpad (talk) 05:49, 13 April 2020 (UTC)
@Nixinova: also note, that just because an account is in the bot group, thus has the bot permission, doesn't mean its edits to recent changes will be tagged as bot edits - the bot must also assert the edit as a bot edit for this to happen. — xaosfluxTalk12:59, 13 April 2020 (UTC)
To answer the original question for @RoySmith: - they don't. "bot" notations are only used in recent changes, and are not part of the revision history. This is a feature request open as phab:T13181. — xaosfluxTalk12:59, 13 April 2020 (UTC)
Xaosflux, That's quite surprising, thanks for the explanation. It certainly explains why I'm not seeing bot edits marked as such :-) But, the more general question is, "How can I tell if an edit was done by a bot?" Recent changes only goes back a limited amount of time. -- RoySmith(talk)14:29, 13 April 2020 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
You can now use the articletopic search word on all Wikipedias. It searches articles by topic. [28]
There was a problem with the Wikidata database last week. Some wikis went down for twenty minutes. Wikidata and other projects showed error messages. Interwiki links were not shown, some tools did not work and other problems. Some of this was fixed quickly. The developers are working on fixing the rest. [32][33][34]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from April 14. It will be on non-Wikipedia wikis and some Wikipedias from April 15. It will be on all wikis from April 16 (calendar).
Future changes
Some graphs have not worked on mobile. This will soon be fixed. [35]
The article tab on talk pages of redirects links to the target of the redirect. It could link to the redirect page itself instead. You can leave feedback on this.
For pages using syntax highlighting, the use of the deprecated <syntaxhighlight lang=""> tag, as well as the use of the deprecated enclose parameter, will add tracking categories.
This action has been automatically identified as harmful, and therefore disallowed. If you believe your action was constructive, please inform an administrator of what you were trying to do. A brief description of the abuse rule which your action matched is: Ağır vandalizm
When I search the uploaded text the string Ağır does not appear. There is also no user on trwiki with that name [36]. My guess is there is an IP number in the page associated with a block. Is there a cross-language abuse filter? -- GreenC19:30, 11 April 2020 (UTC)
Hi team, since yesterday, whenever I save an article it suddenly replaces all special characters such as < and > with their code &equaivalents;. This causes all references and some formats to become regular texts. I can undo it by going back to the temp save under article history but it is really complex. Can anyone look at my history and figure out this sudden problem? I did not change any settings yesterday or recently. I wonder if it is related to this functionality where I have to click radio boxes for many edits since a few months back. This also complicated WP editing by a lot. gidonb (talk) 18:38, 12 April 2020 (UTC)
I'm not 100% certain but under editing > editor the last three out of four options are checked, including this one: "Enable the editing toolbar. This is sometimes called the '2010 wikitext editor'." gidonb (talk) 19:08, 12 April 2020 (UTC)
Normally but since half a year or so in most saves, it throws me to screens where I have to select my version for each paragraph I edited. Not as a conflict remedy, just as an extra hassle. I think it said beta functionality at first so I was still hopeful it would go away. The first save creates an intermediate save, the second a final one. When all is good these are identical. A total waste of time but now entire articles get messed up between the saves. gidonb (talk) 19:22, 12 April 2020 (UTC)
From every instance where I've gotten the EC notification when it's obvious that I'm somehow EC'ing with myself, I've found that my edit is saved already, and I can click Cancel on the EC to not save the spurious characters. Ravensfire (talk) 21:40, 12 April 2020 (UTC)
Ruslik, good question. There was a double column beta-feature that looks exactly like my trouble. I disabled it along with the one other beta-feature that was enabled but doesn't do much. First save worked correctly. After a real edit conflict, this was also Ammarpad's suggestion! Thank you both!!! gidonb (talk) 19:47, 12 April 2020 (UTC)
{{Location map many}} can be used for this sort of map. I don't know about putting it in the infobox. You might just move the map out of the infobox to its own place on the page. – Jonesey95 (talk) 20:15, 13 April 2020 (UTC)
Last night I editeddemodex. Once I logged out my 2 edits were invisible (but present in the page wiki code; it's as if the code is not been taken into account when rendering the article page). There weren't even present in the revision history (the become present therein much later). Once logged in, they became visible again (including revision history). Article browse status depends on whence I browse the page and varies by time: e.g. presently
via my mobile phone internet connection (different carrier/isp, different IP address, different OS, etc...) said changes are visible -while logged out- including the revision history but an edit I did a bit later is at the stage in which the previous ones were before
via my desktop computer -when logged out- all three are still invisible (but the first two are listed in the revision history and all three are present in the wiki code).
via a secondary desktop computer sharing internet connection with the primary desktop one, the situation is the same with the one of the phone.
Just editedTiberias. When logged out my edit is present in the revision history and inside the page code but invisible when browsing the article.
Have tried multiple times flushing browser cache, computer reboot, modem/router reboot (causing inter alia change of wan IP address), different browsers, different OSes, a page purge, a null edit.
Is anyone else experiencing these issues? I feel as if either I am a total idiot and missing something pretty obvious or something very weird is going on with the wikipedia servers. PS Hope that this is the right place to report this issue; if not please point me to the proper one. Thanatos|talk|contributions21:18, 13 April 2020 (UTC)
Latest comment: 4 years ago4 comments3 people in discussion
I need to send the next Wikipedia:VisualEditor/Updates soon. Wikipedia:VisualEditor/Newsletter is about six years old, and I'm sure that some of those people are permanently gone. Others might not be editing here, but they're still around (at least occasionally) on other wikis. MassMessage sometimes gets unhappy with long lists, and it's presumably pointless to deliver a newsletter about editing to someone who isn't editing.
Would someone like to go through the list (ideally with a script that could be re-used for other/similar lists) and figure out who hasn't made any edits to any WP:SUL wiki during the last, say, two years? Then I could send them a note saying that I'm taking them off the list, and telling them how to re-subscribe if they want to. Whatamidoing (WMF) (talk) 20:01, 9 April 2020 (UTC)
Citing reprinted articles and news aggregator sites
Latest comment: 4 years ago2 comments2 people in discussion
There are two scenarios regarding citing reprinted articles using Template:Cite that trouble me.
Scenario 1, there is a book that compiled various newspaper articles that were published before (the word "reprint" here does not refer to the common usage which an entire book got printed again). The outcome of the citation would look like the following, in APA and MLA styles. How can I make either of the following two happen by using Template:Cite?
Krikrok, Mark. "Dream On." National Review. 2010: 35. Reprinted in Rhetoric with Readings. By John D. Bean. Boston: Pearson, 2012. 505-07. Print.
Piaget, J. (1988). Piaget’s theory. In K. Richardson (Ed.) Cognitive development (pp. 3-18). Chicago: Penguin. (Reprinted from Child Psychology, pp. 3-7, by T. Mussen, Ed., 1970, New York: Albany)
Scenario 2. A news article got "instantly republished" by news aggregator sites , for example, an article on Wall Street Journal got instantly republished on Yahoo! News the same day. (Assuming the WSJ article could not be accessed due to pay wall, I must cite the reprint on Yahoo! News). I don't even know how to cite this properly in MLA or APA, let alone how to deal with it in Template:Cite. Any thoughts? Lovewhatyoudo (talk) 14:54, 14 April 2020 (UTC)
Not really a technical question so next time ask at WT:CITE or at the template's talk page.
If the article to which you are adding these citations uses the style that you show here, then use that style there.
If the article is using {{cite}}, then you should also use {{cite}}. The template is not designed to cite more than one source at a time so for your first example:
{{cite |last=Krikrok |first=Mark |title=Dream On |magazine=National Review |date=December 2010 |page=35}}
Krikrok, Mark (December 2010), "Dream On", National Review, p. 35
and:
{{cite |last=Krikrok |first=Mark |chapter=Dream On |editor-last=Ramage |editor-first=John D. |editor-first2=John C. |editor-last2=Bean |editor-first3=June |editor-last3=Johnson |edition=9th |location=Boston |publisher=Pearson |date=2012 |pages=505–507 |title=Writing Arguments: A Rhetoric with Readings}}
Krikrok, Mark (2012), "Dream On", in Ramage, John D.; Bean, John C.; Johnson, June (eds.), Writing Arguments: A Rhetoric with Readings (9th ed.), Boston: Pearson, pp. 505–507
Latest comment: 4 years ago4 comments2 people in discussion
Hello everyone, Is there any tool or script to move many pages title (i.e. categories) quickly? Don't worry! I use it on our project (not here) Thanks! ⇒ AramTalk08:51, 14 April 2020 (UTC)
@Xaosflux: I want to change the categories names due to typo errors (i.e. We have 100 categories named "Category:Abc..." and i want to change their names all to "Category:Xyz..."). Is this example clear? Really?! Can AWB move many page names quickly? How can i do that? ⇒ AramTalk17:59, 14 April 2020 (UTC)
@Aram: keep in mind that moving the category page will not make the pages in that category be moved to the new category - so you will need something like AWB or other tooling to help you with that part, but the page move part isn't automatic. You can see phab:T117909 for discussion and some ideas about using AWB for that. There is no native/builtin function for this task - so you would need to use client-side tool like a script if you want to automate it. — xaosfluxTalk18:10, 14 April 2020 (UTC)
Tree chart template issue
Latest comment: 4 years ago3 comments2 people in discussion
Latest comment: 4 years ago6 comments2 people in discussion
Resolved
Hello all,
Last week my article on Śāriputra was promoted to GA status. However, I noticed that the foreign language versions of the article such as [38] dont feature the good article icon for the English version that is normally found (you can see the one for Russian is there, which is also a good article). Legobot failed to update the article with the icon when it was promoted so the icon and revision note was added manually [39] and [40], so I'm concerned i might have screwed something up. Does anyone know how this works? Wikiman5676 (talk) 16:23, 14 April 2020 (UTC)
@Xaosflux: I tried updating it manually but got the error message "Could not save due to an error. The save has failed.". Any idea what could be causing this? Wikiman5676 (talk) 01:53, 15 April 2020 (UTC)
Latest comment: 4 years ago5 comments2 people in discussion
It's about Module CS1 on my wiki.
I getting the following error:
Lua error in Module:Citation/CS1 at line 2223: attempt to index field 'keywords_lists' (a nil value)
When I clicked on the error to see further information, they read as follows:
Lua error in Module:Citation/CS1 at line 2223: attempt to index field 'keywords_lists' (a nil value).
Backtrace:
(tail call): ?
Module:Citation/CS1:2223: in function "citation0"
Module:Citation/CS1:3858: in function "chunk"
mw.lua:511: ?
(tail call): ?
[C]: in function "xpcall"
MWServer.lua:99: in function "handleCall"
MWServer.lua:313: in function "dispatch"
MWServer.lua:52: in function "execute"
mw_main.lua:7: in main chunk
[C]: ?
I don't know how to add the database table 'keyword list', please help.
I have resolved the first trouble and, the solution is as follows:
''--[[--------------------------< E X P O R T E D T A B L E S >------------------------------------------------
]]
keywords_lists = keywords_lists, '''--was missing in (Module:Citation/CS1/Configuration)'''
Now, I cannot save the following line in ''Module:Citation/CS1/Configuration''
''--[[--------------------------< L A N G U A G E S U P P O R T >----------------------------------------------
These tables and constants support various language-specific functionality.
]]
local this_wiki_code = mw.getContentLanguage():getCode() -- get this wiki's language code
local local_lang_cat_enable = false; -- set to true to categorize pages where |language=<local wiki's language>
--[[--------------------------< S C R I P T _ L A N G _ C O D E S >--------------------------------------------
This table is used to hold ISO 639-1 two-character language codes that apply only to |script-title= and |script-chapter=
]]
local script_lang_codes = {
'am', 'ar', 'be', 'bg', 'bn', 'bo', 'bs', 'dv', 'dz', 'el', 'fa', -- ISO 639-1 codes only for |script-title= and |script-chapter=
'gu', 'he', 'hi', 'hy', 'ja', 'ka', 'kk', 'km', 'kn', 'ko', 'ku',
'mk', 'ml', 'mn', 'mr', 'my', 'ne', 'ota', 'ps', 'ru', 'sd', 'si',
'sr', 'ta', 'tg', 'th', 'ug', 'uk', 'ur', 'uz', 'yi', 'zh'
};''
Whenever the line is included, the page Module:Citation/CS1/Configuration won't saved or published, error message:
''Internal Server Error
The server encountered an internal error or misconfiguration and was unable to complete your request.
Please contact the server administrator at webmaster@en.example.org to inform them of the time this error occurred, and the actions you performed just before this error.
More information about this error may be available in the server error log.
Additionally, a 500 Internal Server Error error was encountered while trying to use an ErrorDocument to handle the request.
Apache Server at en.example.org Port 443''<br />
As a result, the new Citation error is:
Lua error: bad argument #1 to 'fetchLanguageName' (string expected, got nil).
More information on the error, shows:
Lua error: bad argument #1 to 'fetchLanguageName' (string expected, got nil).
Backtrace:
(tail call): ?
[C]: in function "error"
MWServer.lua:81: ?
(tail call): ?
(tail call): ?
Module:Citation/CS1:1498: in function "language_parameter"
Module:Citation/CS1:3223: in function "citation0"
Module:Citation/CS1:3858: in function "chunk"
mw.lua:511: ?
(tail call): ?
[C]: in function "xpcall"
MWServer.lua:99: in function "handleCall"
MWServer.lua:313: in function "dispatch"
MWServer.lua:52: in function "execute"
mw_main.lua:7: in main chunk
[C]: ?
Does this have anything to do with the extension Bable, I don't have it here yet, I'm on share host.
Not going to tell me where I can have a look?
Elsewhere, people who have attempted to upgrade the module suite encounter problems similar to your first problem because they have only updated one or some of the necessary modules. Are you doing that? Has the cs1|2 module suite previously worked? If you are upgrading, upgrade all of the modules, not just those that you think might need upgrading.
Maybe I can be more helpful if you tell me where I can see what it is that you are doing.
Sure, I upgraded all but, some like 'Module:Citation/CS1/Configuration/sandbox' won't saved or published so, I let it be. And Module:Citation/CS1/Configuration only saved half the content, excluding the language section.
What can be done?
Oh! I see, I just check the error_log of en.example.org and found:
''[15-Apr-2020 09:39:41 America/Boise] PHP Warning: Error while sending QUERY packet. PID=740726 in /home1/example/public_html/en/w/includes/libs/rdbms/database/DatabaseMysqli.php on line 46''
What that means, maximum query allow to MySql database is reached.
I asked my host to increase it but they say it cannot be done otherwise it would affect all website hosted on the same server and advised for upgrade to VPS which I am not ready for.
They also advised on a second option: to reduce the query size to smaller chunk; please, how do I do that ?
If you are getting internal server errors, then you may have bigger problems than getting this module suite to work. I am not the person to help you fix the server errors or mis-configuration problems.
I do not know how to reduce query size. Perhaps someone at WP:VPT can help you with that.
The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
My attempts to edit the lede of the 2020 Delhi riots have repeatedly failed and I have received the message [Xou5NApAICEAAHRxKa4AAAEU] 2020-04-06 23:20:22: Fatal exception of type "Wikimedia\Rdbms\DBQueryError" in response. Any ideas whether this is an issue at my end, with wikipedia servers or with the particular article?
In case it matters, the edit I was attempting was really minor and changed just a few words of plain-text without fiddling with any of the templates, links or wiki-markup in the lede. And I tried the edit using both Firefox and Chrome browsers, with similar results. Abecedare (talk) 23:28, 6 April 2020 (UTC)
Ditto; I earlier even had a problem posting the above message on this page. Although the issue now seems to have somehow resolved itself for me <fingers crossed>. Abecedare (talk) 23:32, 6 April 2020 (UTC)
phab:T249565. A Wikidata database table was dropped, it's not clear how. This caused the database errors when viewing any page linked to a Wikidata item (hence why most talk pages were still working). The table has been recreated as blank, which resolved the page viewing issue. However pages will be missing the links to other languages in the sidebar, and there may be other problems. Work is underway to rebuild the table, or to restore from backup. the wub"?!"00:28, 7 April 2020 (UTC)
I'm not able to access many Wikipages, including the Main page and WP:Village pump (technical) due to a "database error." Huh? ~EDDY(talk/contribs)~ 23:09, 6 April 2020 (UTC)
I can load other articles but am getting the database error on the main coronavirus pandemic page. Using an Android smartphone in Northern California. Cullen328Let's discuss it23:14, 6 April 2020 (UTC)
Ditto. Happening on all mainspace edits for me. The long string seems to change value every edit, but the rest of the message is the same. Sam-2727 (talk) 23:16, 6 April 2020 (UTC)
(edit conflict) x2 IJBall asked me on my talk page to revert an edit because he was getting this error, but I am also getting it, so this appears to be site wide rather than just some users. And resolving edit conflicts is also not working how it should. Amaury • 23:20, 6 April 2020 (UTC)
Note that language links are not populating from wikidata. The data is still there in Wikidata, so if anyone links an article to one of another language it is resulting in duplicate items on Wikidata. Oddly the Wikipedia links may show on multiple Wikidata items this way. Possibly there is an issue at Wikidata as this action should be forbidden. SFB00:04, 7 April 2020 (UTC)
For those of you not following the phab ticket, there was just an update posted. The gist is that the deleted table is being restored from backup, that's expected to take another 12-24 hours, and until that's done, a whole bunch of wikidata-related stuff (which includes inter-language links) is going to be farshtunken. -- RoySmith(talk)01:26, 7 April 2020 (UTC)
Good ops practices: don't have your database dump cronjob run with write permissions on production master. Lesson learned. ;) Not to pick on y'all WMF devs too much. Thanks for your hard work; most operations that've been around for a while tend to accumulate crufty code like that. That's a really good case study in how a bunch of unrelated changes to different moving parts over the years can wind up randomly shooting you in the foot one day when a code path no one realized existed gets triggered. --47.146.63.87 (talk) 07:07, 8 April 2020 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
No pencil icon at section level in mobile view
Latest comment: 4 years ago9 comments4 people in discussion
The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
I am unable to edit at the section level on any page in mobile view (iPhone, iOS 13.3). Desktop view is OK. Jmar67 (talk) 01:32, 10 April 2020 (UTC)
Update: The pencil icon is not showing, but the field it represents is active and I do get the edit screen when I touch it. Possibly a font problem. Jmar67 (talk) 02:23, 10 April 2020 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Global watchlist - Update 6
Latest comment: 4 years ago1 comment1 person in discussion
If anyone is interested in reviewing the code and making suggestions regarding coding style, techniques, or something else, please let me know. See phab:T234507 for more.
Recent additions
The results for each site, as well as the list of sites without changes, can be collapsed; the list of sites without changes starts collapsed.
When the username associated with an edit has been hidden (either via revision deletion or suppression) this is now properly shown in the output, rather than showing "undefined".
Links to view changes open in new tabs.
When marking a site (or all sites) as seen, only changes made prior to the watchlist loading will be marked as seen, so that any changes since then are not missed.
There is now a help link in the upper right hand corner that links to m:User:DannyS712/Global watchlist, though it doesn't have the normal help icon.
Translations
Many thanks to those that have contributed translations for the script's messages. Translation takes place at m:User:DannyS712/Global watchlist/Translate, and help is appreciated.
Newly translated messages are added to the script when I notice them; if you have translated messages that aren't showing up, please let me know at m:User talk:DannyS712.
Next release
Version 9.0.0 was just finalized for the development and stable branches of the script.
Latest comment: 4 years ago3 comments3 people in discussion
When leaving user warnings on IPv6 talk pages, I feel this is nearly useless as on some providers the IPv6 hops every couple of hours or even minutes. I think there is little to no downside in combining IPv6 talk pages to a single /64 range. In the few cases users share a /64 range and edit Wikipedia, you'd get a shared IP page. It also would be nice if Twinkle rollback reverted IPv6 edits on the basis of the /64, vandalism by a IPv6 switching withing its /64 range often gets left in. The upside to combining the talk pages for a /64 is that the IP user is more likely to see the warning and recent change patrol and admins are more likely to see prior warnings to the IPv6.--Eostrix (talk) 13:50, 16 April 2020 (UTC)
I've tried purging the page (in fact, I tried everything in Wikipedia:Purge), but the data for today still doesn't show up. Just to rule out local caching, I tried on a different device (mobile vs desktop), connected to a different ISP, with a different browser (Firefox/Android vs Edge/Windows), and I still get an out-of-date page. --NetRolller3D16:55, 16 April 2020 (UTC)
Logging in or out makes no difference here - both logged in and logged out, the template page is up to date, the main article isn't. --NetRolller3D17:05, 16 April 2020 (UTC)
Broken WikiProject templates - looking for a guilty userscript
Latest comment: 4 years ago3 comments3 people in discussion
Is there a script that suggests that WikiProject banners be added to talk pages? I ask because I monitor Wikipedia:Database reports/Broken WikiProject templates, a report that is updated weekly. This report lists transclusions of non-existent templates whose names begin "wikiproject", case-insensitive. In the past, it typically had just a few entries - rarely more than about ten, often just five or six. This last few weeks, the amount has shot up:
27 February 2020 - 5
5 March 2020 - 4
12 March 2020 - 6
19 March 2020 - 24
26 March 2020 - 19
2 April 2020 - 14
9 April 2020 - 35
16 April 2020 - 41
23 April 2020 - 25
In older versions, these non-existent WikiProject banners were normally due to the activities of about three people, who I shall not name. But the "usual suspects" seem to be behaving themselves these days, and in the last five versions of the report, the non-existent WikiProject banners have almost always been added by a wide variety of different people, just one talk page per person, but anything up to five bad banners per talk page (which might be in the main Talk: or in Draft talk:). These people are probably newbies, using some kind of article creation script - is there one which adds Wikiproject banners to talk pages? --Redrose64 🌹 (talk) 23:01, 16 April 2020 (UTC)
The AfC helper script has an option to add WikiProjects when accepting a draft. I don't think it's the culprit here, though, looking at the pages. At least one of them used Kephir's rater, which (I confirmed myself) has an extra dialog box if you put in a nonexistent WikiProject. Lots of music-related projects on the current revision of the database report - education-related, maybe? Enterprisey (talk!) 00:11, 17 April 2020 (UTC)
Gadget proposal: find archived sections easily
Latest comment: 4 years ago9 comments6 people in discussion
Per a discussion on my talk page, I am proposing that the script User:SD0001/find-archived-section be made a gadget. When you follow a broken section link to an archived discussion, the script searches the archives and tries to find the section, and gives you the correct link. I feel that this would be useful for everyone.
Strong support, really solid gadget. (When this closes, feel free to ping me if you need someone to actually edit the gadget pages.) Enterprisey (talk!) 15:26, 16 April 2020 (UTC)
ClueBot (in its archiving role) does it today for the pages it archives, but I know that functionality has brought the bot to its knees at least once before. --Izno (talk) 16:28, 16 April 2020 (UTC)
Also, the bot cannot fix links made on other wikis and off-wiki such as on phabricator, gerrit, toolserver, github, etc etc. SD0001 (talk) 03:51, 17 April 2020 (UTC)
And that's to say nothing of the links in edit summaries, log entries, Special:Notifications, old versions of pages, etc. SD0001 (talk) 07:44, 17 April 2020 (UTC)
Strong support. Granted, I'll still be editing links to redirect to the archived location, but searching for it becomes a lot easier. —Tenryuu 🐲 ( 💬 • 📝 ) 04:01, 17 April 2020 (UTC)
Searching variations of insource:/washingtonpost\.com\/\// returns error "An error has occurred while searching: Regular expression syntax error at unknown: unknown"
Latest comment: 4 years ago9 comments4 people in discussion
I realize I made a mistake and with my bot added extra slashes that will break URLs on the washingtonpost.com domain to a few pages. Trying to find these pages and fix them I tried searching with insource:/washingtonpost\.com\/\// (using regex because it has characters not normally searchable) to find these. However, this gets the error "An error has occurred while searching: Regular expression syntax error at unknown: unknown". I have no idea what is wrong. How can I fix this search and find the pages to revert? Thanks a ton, DemonDays64 (talk) 20:44, 15 April 2020 (UTC)(please ping on reply)
That also works and gives the same results. It only needs square brackets before the final / to avoid //. You can also end with (\/)/ or many other things to break up //. PrimeHunter (talk) 01:54, 16 April 2020 (UTC)
@PrimeHunter: oh dang...I forgot to refresh the code on the tabs running the bot! I will use that search to find the pages now and do a manual JWB run later today to fix it. Thanks for telling me! DemonDays64 (talk) 15:51, 17 April 2020 (UTC)
I have now run the bot on 47 pages, using the search insource:/https:\/\/(www\.)?washingtonpost\.com\/\/./ using JWB annotated, Colin M's fork of JWB that allows generating page lists with regex search, and the settings as:
That looks like a bug that should be reported if there isn't a report already. There's no good reason why that should fail; looks like there's buggy code that assumes the regex never ends with / and barfs when it does. --47.146.63.87 (talk) 08:24, 16 April 2020 (UTC)
Pages transcluded onto the current version of this page
Latest comment: 4 years ago4 comments3 people in discussion
I find it annoying that the list of "Pages transcluded onto the current version of this page" beneath the edit box shows up when editing an entire existing page yet not when it's a section or a nonexistent page. Is there a way to see it when previewing an edit to a section or a nonexistent page? Nardog (talk) 02:40, 18 April 2020 (UTC)
Huh? It does show up when you preview a section or a non-existing page. Maybe you have to click a triangle next to the heading to expand it. It doesn't show up on the first edit click at a section since the information isn't known at that time. You have to preview first. PrimeHunter (talk) 04:05, 18 April 2020 (UTC)
But it doesn't show up on preview for those of us using the AJAX preview ("Show previews without reloading the page") preference. AJAX previews are faster and really ought to be the default. SD0001 (talk) 05:09, 18 April 2020 (UTC)
Ah yes, only in the live preview does it not show up. (I tested with JS off before posting this but apparently I forgot to include a template!)
But when previewing an edit to an entire existing page, it does show up even with the live preview on, so it doesn't seem like a limitation on the live preview's part. Nardog (talk) 06:12, 18 April 2020 (UTC)
Image cropping/clipping
Latest comment: 4 years ago2 comments2 people in discussion
Latest comment: 4 years ago2 comments2 people in discussion
Hi! ^_^ , very nice to meet you. ^_^ . I put an entry into Portal:Current Events which summed up said that Hong Kong activist Jimmy Sham warned police to expect "more resistance than ever". OK, that was in source, I now go to the source and he is not quoted anymore as saying that but instead warning police of possible "violent protests". My question is, anyone who reads that may rightfully think I was lying but I was not, the first time I read the source he was quoted as that, but it seems to be the editors of the newspaper changed the information. What to do in such cases?. Thank you and I hope to have got across with my mediocre English ^_^ -- :love: --CoryGlee (talk) 20:29, 18 April 2020 (UTC)
Special header for a page when incoming from a redirect?
Latest comment: 4 years ago2 comments2 people in discussion
Is there any facet in the Wikimedia software to display a header for a page when that page is coming from a redirect from a different specific term?
The case in point here is that while "Wuhan virus" is a searchable term for the Severe acute respiratory syndrome coronavirus 2, it is highly layman and considered derogatory. While Wuhan virus does redirect to that page, it would be nice to be able to tell the user only on that situation that they have reached the right page but we're not calling it by "Wuhan virus" due to the inappropriateness of the term. But I could see this applying in similar situations where a redirect term is something that would not be mentioned or not with the weight of other lede material, but is still a layman's search term. --Masem (t) 03:50, 19 April 2020 (UTC)
Latest comment: 4 years ago4 comments3 people in discussion
{{Annual readership}} appears to be broken on any page that uses it (including the template page itself). Instead of the normal graph, only a broken image icon is displayed. I'm unsure how to begin diagnosing this one, so I thought I'd post about it here. Thanks, –Deacon Vorbis (carbon • videos) 17:49, 17 April 2020 (UTC)
I purged it and it works now, it's probably something server-side that will be fixed sometime, I don't know, it's been not working since forever. Purge the template, then purge the article, huzzah it works for 1 to 2 weeks and we're back to it being a placeholder ? icon. --qedk (t愛c)17:57, 17 April 2020 (UTC)
There was a non-printing character at the decimal point in 5.5 in the second ref. I fixed that, but the two refs should still probably be combined. MB19:49, 19 April 2020 (UTC)
Latest comment: 4 years ago4 comments3 people in discussion
Resolved
At User:Doug Weller/common.js when I open the page to edit I get "https://en.wikipedia.Org/w/index.php? title=User:Doug_Weller/common.js&actic at line 818: Uncaught Error: Another textSelection API was already registered". Maybe connected, I often get "JavaScript Error script error when I load a Wikipedia page - both of these in boxes at the right-hand top of the page. I don't think that the bottom two scripts are installed correctly although the top one works, eg it marks the use of lulu.com at Girolamo Cavalcabo. Thanks. Doug Wellertalk09:35, 20 April 2020 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Tech News
The next issue of Tech News will be sent out on 4 May 2020.
Recent changes
The small wiki toolkits is to help smaller wikis that need technical skills. They can learn and share technical skills. [42]
Over-qualified CSS selectors in Wikimedia skins have been removed. div#content is now .mw-body. div.portal is now .portal. div#footer is now #footer. This is so the skins can use HTML5 elements. If your gadgets or user styles used them you will have to update them. [43]
Changes later this week
There is no new MediaWiki version this week.
Future changes
Some things on the wikis might look weird or not work in Internet Explorer 8 in the future. Internet Explorer 8 was replaced in 2011. [44]
The new version of MediaWiki will be on test wikis and MediaWiki.org from 28 April next week. It will be on non-Wikipedia wikis and some Wikipedias from 29 April next week. It will be on all wikis from 30 April next week (calendar).
.I went to the linked page, cut the bold text below the image there (reading פרויקטים בביצוע אד הוק בניה, עבודות גובה ושיקום מבנים, and pasting it here worked OK; maybe it's peculiar to wikilink-labeling), tried to paste it in to label the wikilink there, and got unexpected results in the edit window. I see similar unexpected results doing that here. The text I'm trying to paste in is Hebrew, with R->L direction. I'm not literate in Hebrew, I'm aware of the direction difference, but the results I get here are still unexpected. Just thought I would mention it. I'm using Win10 and Google Chrome if that matters. Wtmitchell(talk) (earlier Boracay Bill) 12:08, 21 April 2020 (UTC)
"unexpected results" is vague. Your url above is rendered because there is a newline before ]. Just remove it. Browsers may assume that numbers after right-to-left text is part of that text and should also be displayed right-to-left. You can insert a left-to-right mark‎ before the numbers. With ‎:
My Firefox renders "14" in the right place with ‎ but to the left of the link without it. When it reaches "January" it knows the right-to-left text has ended. It's only to the left in the rendered page for me. In the source text I see it in the right place but browsers may vary. The produced html is in the original order so it all depends on the browser. PrimeHunter (talk) 13:22, 21 April 2020 (UTC)
Word counter
Latest comment: 4 years ago3 comments2 people in discussion
If anyone wants to test out a new tool, I made a new word counter in ~3 hours: https://wordcount.toolforge.org
Upcoming features include character count (maybe?), API (maybe??) Honestly though, I made this tool only because I saw that ArbCom currently recommends using a commercial advert-filled website and I just wanted to get some OSS into the mix. This pandemic has left me with too much free time. Facepalm --qedk (t愛c)20:54, 20 April 2020 (UTC)
Editors who would like to do word counting on-wiki, and don't mind a little JavaScript in their lives, can import User:Dr pda/prosesize. I think there is a similar script as well, but I can't put my finger on it at the moment. – Jonesey95 (talk) 23:16, 20 April 2020 (UTC)
Just adding that prosesize is pretty useful for on-wiki pages, but it counts entire pages on the site only, so there's that. --qedk (t愛c)14:59, 21 April 2020 (UTC)
Change Graph:Lines default axis, labels, legend, and grid color to #54595d grey for color-inversion friendliness
Latest comment: 4 years ago4 comments3 people in discussion
The team is planning some upcoming changes. Please review the proposed design and share your thoughts on the talk page. The team will test features such as:
an easy way to mention another editor ("pinging"),
a rich-text visual editing option, and
other features identified through user testing or recommended by editors.
I think this would be a nice beta feature to have. Can it be placed under "Beta" so I can try it out on enwiki? Aasim19:02, 13 April 2020 (UTC)
You can try it on the wikis where it's already in a beta. enwiki is an unlikely candidate for betas because well, enwiki. @Whatamidoing (WMF): Does this new tool have some feature in place that prevents or fixes edit conflicts? How does it deal with them? --qedk (t愛c)13:27, 14 April 2020 (UTC)
User:QEDK, the Reply tool auto-resolves most of edit conflicts (the magic is in mw:Parsoid, apparently). If it can't resolve it, you get an error message and the opportunity to copy your message (so you can either reload the page and try again, or open your normal wikitext editor and reply the old-fashioned way). I've seen edit conflicts that were auto-resolved; I've not seen complaints on any of these wikis about it failing (sample size =~500 real-world replies). If anyone wants to test this, remember that you need two accounts to trigger an edit conflict with yourself. MediaWiki overwrites your own edits rather than triggering the edit conflict tools.
@Whatamidoing (WMF): That sounds good! There's a couple of issues I've come across: 1) sometimes I just get an empty box saying "OK" when clicking "Reply", not sure why. This has happened a few times, refreshing the page fixes it. 2) The "Reply" is disabled as soon as you hit "Reply" on one message, this is not ideal because sometimes you misclick and want to switch to a different editor, ideally you want a prompt asking you to switch replying to that if the editor has started typing a reply to someone else. For info, I am using Safari 13.0.4 on macOS Catalina 10.15.2. --qedk (t愛c)19:09, 14 April 2020 (UTC)
is deliberate, because it can only handle one Reply at a time. The red "Cancel" button will bring the other Reply buttons back (first copy any text you have already typed and want to keep, of course). Whatamidoing (WMF) (talk) 22:35, 14 April 2020 (UTC)
I played with it a little bit. Would I be correct to deduce that it looks for a line (or paragraph, if top-level text) that contains both (a) a link to a User: or User talk: namespace (can be cross-wiki or redlink); and (b) a time stamp in the exact format of the wiki (case-sensitive, same language and timezone, etc.)?
Pelagic, it looks for a timestamp (exactly that wiki's default format) plus any link to the user's account (user page, user talk, or contributions). It must be one of those pages and not a subpage or redirect. (I believe that in MediaWiki, "a line" is the same as "a paragraph".)
The problem of how to detect signatures in templates such as {{u}} and {{unsigned}} is phab:T250516. If you have opinions or advice (e.g., whether it's better to fix the templates or the extension), then please share there. Whatamidoing (WMF) (talk) 15:46, 21 April 2020 (UTC)
""ntsamr"-pattern spambot filter"
Latest comment: 4 years ago4 comments3 people in discussion
I noticed this hidden edit filter whilst patrolling the AbuseLog and was wondering what exactly it is. What does the "ntsamr" pattern constitute, i.e. what does this filter look out for in an edit?
Nothing to say about me really. If it's private there might not be much I can publicly say about exactly how it works, but think spambot user pages. The filter is not very reliable for various reasons, although it's not too bad. You might see some examples if you look at the logs of the deleted user pages of the blocked accounts; one such example is this another this. -- zzuuzz(talk)11:16, 19 April 2020 (UTC)
Hello! I have a question that may sound a bit silly. I already checked the FAQ and I don't think I saw it there so...
When we delete a certain page, why do "have" to keep its talk page? Those get "lost" and the way I see it, only remain as a burden to the servers (albeit an insignificantly small one). Not many users, to say the least, will ever search for a talk page of a non-existent page. Are there strong reasons we keep them? Apart from the edge cases where users want to go and discuss about something related to the deletion process. Is there a way to track these kind of "friendless" talk pages? Is there a way to delete them simultaneously when we are deleting a certain page? What's the overall general logic in keeping talk pages in these situations?
I'm nearly sure something like this has been answered before and there must be reasons I'm overlooking in my logic but I couldn't find it at the FAQ page so curiosity got the best of me. - Klein Muçi (talk) 14:54, 20 April 2020 (UTC)
User:Klein Muçi may have been thinking about user talk pages. Even if we delete a user page, we almost never delete the associated user talk page, even if the user asks us to. We keep it to maintain the user's history. See WP:DELTALK. -- MelanieN (talk) 15:29, 20 April 2020 (UTC)
@Xaosflux: Yes, I know of those cases and that policy because I've researched a bit this subject because of my curiosity. But I was referring mostly to the classical, unenhanced way of deleting pages. When you try to move something, you get options that allow you to decide if you also want to move its sub/talk pages. When you delete pages, there are no facilities like these. Wouldn't it be easier to have the same options when deleting pages? I believe there will be thousands, if not more, articles', categories', templates' and what not pages deleted whose talk pages still remain in the project without any specific need. This is certainly the case at SqWiki when I'm an admin. Many admins there don't even know that the talk pages still remain after deleting those kind of pages and those that do know, don't delete them anyway because they think they are needed because of something they don't know yet, because otherwise they'd have the options to delete them together with the page, like the move options.
@MelanieN: No, I know you keep them even when users get retired. Keeping that in mind, I thought maybe a reason for that exists in other namespaces too. I'm asking because, as I said, we have years (more than a decade) of a habit of not deleting talk pages in SqWiki because of the reasons specified above. And since no one really knew the reason why we were doing that (assuming they did know we weren't deleting them, because most of them thought the deletion happened simultaneously) I thought of asking here. And if no general reason really exists for that, I'm starting to think we might need to do a massive cleanup at our community. - Klein Muçi (talk) 15:44, 20 April 2020 (UTC)
@Xaosflux: I mean, maybe? I'm not sure what namespaces should I search for. Giving it a quick thought, I think there is not one namespace that wouldn't be affected by that problem. And it's the first time I'm seeing this quarry tool so I can't say much about it. Besides, I'm in a dilemma if I can automatize the process or not. Maybe there are unassociated pages that are needed to be like that (like the examples you mentioned above)? (Besides x2, how deep should the quarry go? Considering that subpages also have talkpages.) As I said, I haven't researched much the actual situation in SqWiki because I thought we had a reason for keeping them, since that has been the case for over 10 years now. It's only now I'm considering the said cleanup and I was wondering if you've ever dealt with a situation like this before here, but looks like you haven't. [Text written before an edit conflict.] - Klein Muçi (talk) 16:27, 20 April 2020 (UTC)
Basically, there is no way to find these on-wiki, so you have to result to a database analysis, which is what quarry does. The primary example of an appropriate talk page without a matching page would be "user talk". If you do a database review, you could exclude pages with "/" in their titles to eliminate most subpages. — xaosfluxTalk17:29, 20 April 2020 (UTC)
@Klein Muçi: Regarding "the classical, unenhanced way of deleting pages" and "When you delete pages, there are no facilities like these." - in fact, there are. When an admin clicks the "Delete" tab, selects a reason and clicks Delete page, they then get presented with this screen, note the full-width pink box. This box is suppressed if there is no talk page, or when deleting a talk page or a User: page. --Redrose64 🌹 (talk) 10:21, 21 April 2020 (UTC)
@Redrose64: I see... Actually, we don't have that system message in our wiki. We have it set up differently. Check for yourself here. And even like this there are still some extra steps compared to what happens with move pages. But you offered me a solution for the future. If I ever am able to do that clean up I've mentioned below, I'll set that system message up like in here, hoping we won't get in a similar situation in the future again. Thank you! - Klein Muçi (talk) 11:40, 21 April 2020 (UTC)
@Stryn: Hey, that's actually pretty good! Is there a way to see the whole list? The limit of 100 results doesn't really do it for me in this case, as you've may have already thought of. Other than that, I think this is perfect for this situation, assuming the list is correct (I tried it at Sq.Wikiquote and it found out only 15 pages. The project is really young but still the low number baffled me.) - Klein Muçi (talk) 17:16, 20 April 2020 (UTC)
Did you try downloading the data from the query I ran, #44156? You should be able to do easy text filters to remove user talk and pages with "/". — xaosfluxTalk18:21, 20 April 2020 (UTC)
@Xaosflux: No, I didn't. I've never worked with that tool before and this looked like an easier solution from the tech side at first glance. But I'll give both a try later then.
@Xaosflux: Haha, yes but I meant learning a bit what to do with it without disturbing you. In what format am I supposed to download it from? And where am I supposed to feed it? Maybe AWB? Or am I off track with the logic here? - Klein Muçi (talk) 18:48, 20 April 2020 (UTC)
It really depends on what you want to do with it, you probably shoudn't just feed it in to some sort of delte-o-matic. — xaosfluxTalk19:29, 20 April 2020 (UTC)
@Xaosflux: I'm seeing the downloaded file now. Was it supposed to show user talk pages? Because I'm seeing a lot of those at that file. I thought you had done your quarry in such a way as to remove those? Or have I misunderstood you? - Klein Muçi (talk) 00:27, 21 April 2020 (UTC)
The namespace is in there, if you want to do large bulk things, you're going to have to use a text editor, maybe a spreadsheet program, to filter and get the data in just the way you need it - using those is a bit beyond the scope of VPT though! Someone may have some recommended utilities, I usually just use shell scripts or text editors. — xaosfluxTalk01:14, 21 April 2020 (UTC)
@Xaosflux: Oh, so it's not a mistake. How about subpages? Are those counted too? Just want to know with what information I'm dealing with. I'll try to use Excel for sorting/filtering data.
And this query is somewhat 'simple' it does include subpages (they have a "/" in them) but it doesn't exclude them if there is a "root" companion page for example. — xaosfluxTalk01:34, 21 April 2020 (UTC)
@SD0001: thank you! I had never seen this tool before so all this is new info for me. Can you run one for SqWiki too? So I have it, just in case. I did get more responses than I hoped for in this question and I'm delighted for that. Now if only I'd get some suggestions on how to handle the situation practically regarding the high number of orphaned talk pages that we have... - Klein Muçi (talk) 16:22, 21 April 2020 (UTC)
Save edit in Visual Editor, it's asking to save again?
Latest comment: 4 years ago5 comments2 people in discussion
I've seen this three times over the past week or two. I edit a page in the Visual Editor, save my changes, and then I'm back to VE asking me to save. For example, I just made this edit. Yet, I'm looking right now at a "Publish changes" dialog, asking me to save those exact same changes. I just clicked "Publish changes" again, to see what would happen, and it looks like it's just hung. I've got the "Review your changes" barber-pole across the top, with the strips moving to indicate operation in progress. Anybody else been seeing anything like this? -- RoySmith(talk)21:53, 18 April 2020 (UTC)
Just had it happen again, on this edit. I hit "Publish changes", the edit got saved, and I've got another dialog asking me to save. Clicking "Review your changes" shows no changes. Clicking "Publish changes" hangs. -- RoySmith(talk)14:55, 19 April 2020 (UTC)
RoySmith, can you tell me your web browser and OS? (I get something like this in Firefox on macOS, but only at Wikivoyage.) Also, the next time this happens, could you please open the page history in another tab, and see whether it has posted your changes already? (Please ping me.) Whatamidoing (WMF) (talk) 15:52, 21 April 2020 (UTC)
This is being tracked at T250620; could you update that with your observations? But, to answer your question, I'm on Chrome / MacOS, and the other editor who was able to reproduce it, the same. And, yes, every time I've checked the history in another window, the history recorded it had been saved. -- RoySmith(talk)16:28, 21 April 2020 (UTC)
Diff: One intermediate revision by 12 users not shown
Latest comment: 4 years ago6 comments3 people in discussion
Can we get the Diff program to produce a more reasonable message than "One intermediate revision by 12 users not shown" when an intervening revdel is involved, as seen in this diff? (Possibly related discussions in archives:
4, 36, and 83.) Mathglot (talk) 05:19, 21 April 2020 (UTC)
Probably the second archive link is not related to this. But for the first archive link, I cannot reproduce the issue using what the OP described. (Open the the page history and compare the revisions that the OP described). It may have been fixed or maybe there was never a problem there. The issue in the last archive link is a different issue and it has been fixed since. Also what do you think the "x intermediate revision" message should say?– Ammarpad (talk) 10:26, 21 April 2020 (UTC)
Problem reproduced on IE 11, Opera 67, Chrome 80, and Firefox 69. I'm not sure what it should say, but it shouldn't be a logical impossibility. Maybe, "83 revisions by 12 users not shown"? If it doesn't have that information, then maybe, "Many revisions by...". Mathglot (talk) 11:55, 21 April 2020 (UTC)
uselang=qqx shows the message is MediaWiki:Diff-multi-otherusers called with (diff-multi-otherusers: 1, 12). Here at enwiki we could test whether $1 < $2 and add a possible explanation like "One intermediate revision by 12 users not shown. Some revisions may be hidden." But we don't know whether there are hidden revisions when $1 ≥ $2, and it could be confusing to only display the added text when $1 < $2. We don't have access to the real number of revisions without a MediaWiki change which should be requested at Phabricator:. PrimeHunter (talk) 13:55, 21 April 2020 (UTC)
I found that the real count is intentionally hidden to tackle unrelated issue. There were concerns of leaking "statistics of deleted revisions" too, so the real count for such cases is unlikely to be restored. Also note, what I cannot reproduce is the issue that you linked from this archive. I think I made that clear. – Ammarpad (talk) 14:46, 21 April 2020 (UTC)
*http://example.com/home (optional space between asterisk and url)
[http://example.com/home
<ref>http://example.com/home
url=http://example.com/home (optional space between "url" and "=" and also between "=" and the url)
However, it does not work on bare-URL references with a refname.
I am not sure what sort of thing to do here to guarantee I will capture all somewhat reasonably formatted refs. I need to be able to get all of these, at least:
I think I can handle most of the refs with just a lookbehind for <ref name=.*> (demo at https://regexr.com/530uv).
However, I would like to ask for a review of this — would this cause any problems, and are there any common formats of bare URL references this would not get?
BTW here is what the regex would be with the new lookbehind:
(Another regex I worked on for a little was <ref name="[^"]*">http:\/\/example\.com (see https://regexr.com/530up), but I don't actually see any reason it is better than the other and it only works on refs with quotes.)
Latest comment: 4 years ago12 comments9 people in discussion
I discovered this via the help desk. Apparently, for U.S. counties, if you search Google for how many coronavirus cases in a particular county, the sidebar pulls data from Wikipedia. That's par for the course. However, for some reason for counties in Wisconsin, it is not pulling the number of cases in the county as reported in our article, instead it is pulling from the column for cases per 100,000 people, then dropping the decimal. So, for example, La Crosse County, which has 25 cases, or 21.2 cases per 100,000 people, shows up on Google as having 212 cases. Milwaukee County, which has 2,191 cases or 229.6 per 100,000 people, shows up on Google as having 2,296 cases, and so on. As far as I can tell, that's the only state where Google is doing that.
For comparison, I looked at 2020 coronavirus pandemic in Michigan, which also lists how many cases per 100,000 people. The Wisconsin chart does have some columns the Michigan chart doesn't have, but overall they are similar. However, Google is correctly pulling the Michigan data from the cases column instead of the cases per 100,000 population.
We can't, it's the scraper they have built that is bugged. I don't know how we can leave them a note, especially that they don't even have the small "Suggest an edit" text at the bottom of COVID-related snippets. --qedk (t愛c)20:58, 20 April 2020 (UTC)
Thanks, TheDJ :) @ONUnicorn, as noted above, I'm pretty sure this is an issue with how Google's parser is reading the table, so I've given them a heads up about this. The Google team working on this feature is subscribed to the above mailing list, so it's a good way to directly report any issues you're seeing. I've also been monitoring various template and WikiProject talk pages for feedback/bug reports and sending out to the list on a regular basis, so pinging me works, too! MPinchuk (WMF) (talk) 13:38, 21 April 2020 (UTC)
Wisconsin Corona Virus cases incorrect on page that automatically comes up when googled
The counts by county for Wisconsin when corona virus cases is googled is so incorrect. When I go into the core they are correct but the page that comes up directly when googled is so off it is scary — Preceding unsigned comment added by 2601:40d:8001:ca60:8d21:8ed6:8014:b87e (talk) 18:39, 21 April 2020 (UTC)
Are you talking about the "Cases overview" that comes up in the Google Knowledge Graph when you search wisconsin coronavirus cases? If so we have no control over Google, and, although they always claim their information has come from us, it quite often has either come from someone else, or has come from us but is out-of-date. If you want to complain about Knowledge Graph, click the "About this data" link underneath it and a "Send Feedback" link will pop up. ‑ Iridescent18:46, 21 April 2020 (UTC)
Change to CS1 (cite xxx) templates is showing warning messages to some editors using custom scripts
Latest comment: 4 years ago43 comments8 people in discussion
The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section.A summary of the conclusions reached follows.
If you have the useful User:Ucucha/HarvErrors script installed in your common.js file, you may notice an increase in the number of brown "warning" messages that the script is generating in articles. This is due to a change in the Citation Style 1 (cite book, cite web, etc.) templates that adds |ref=harv to those templates by default in order to fix short citation links to those full citations. There was discussion of this change at Help Talk:Citation Style 1, a good (though sometimes quite active) forum to follow if you have an interest in citation formatting and templates.
No action is required on your part, or in the articles in which more messages are appearing. If the brown warning messages bother you, there is advice at the top of User:Ucucha/HarvErrors offering a few options to suppress them. (P.S. Please don't shoot the messenger.) – Jonesey95 (talk) 18:40, 19 April 2020 (UTC)
This is what I see now in "Selected works" and "Further reading" sections that use {{cite book}}, {{cite news}}, etc. But if I remove the script that points to the errors, I lose the errors that I want to catch, namely when a long citation hasn't been used as a source. Can someone point to the change that has caused this? Someone has made ref=harv the default. Pinging Dudley and Gerda who have also complained about it. SarahSV(talk)18:33, 21 April 2020 (UTC)
It's a little unclear to me why the solution to a script malfunctioning is to fork the script. Why isn't the original script being edited so that it stops bugging up? Jo-Jo Eumerus (talk) 18:35, 21 April 2020 (UTC)
Jo-Jo Eumerus, sorry, I can't answer that. They've said we have to ask for a new script to be written. But this change (that ref=harv is now the default) didn't have consensus. It was discussed in 2018 and again recently, this last time by just a few people. There should be wider consensus for changes that affect so many people. SarahSV(talk)18:38, 21 April 2020 (UTC)
@Jo-Jo Eumerus: because user scripts have security risks, and you can't edit someone else's user script. Maybe interface admin can, but mortals can't. So the solution, if you're not happy with the way a script works, is to make a new one. Or have the user update the script if they are still around and willing. If people don't posses the skills to make their own fork, WP:SCRIPTREQ exists to draw on the knowledge of those who can. Headbomb {t · c · p · b}18:51, 21 April 2020 (UTC)
Jo-Jo Eumerus: The script is not malfunctioning. It is showing the same no-action-necessary (but occasionally useful) warning (not error) messages that it has always shown, and it is now showing those warnings for an expanded set of templates. That change, like many changes, has resulted in disappointment from some editors who, like most of us, find change difficult. Instructions have been posted in various places telling editors how to modify the script's output, or how to use different scripts. – Jonesey95 (talk) 18:53, 21 April 2020 (UTC)
Also, the bit about consensus is obviously false. There was pretty much unanimous consent that this is an improvement to the ease of use of templates and fixes tens of thousands of articles with broken citation anchors. What we have now is 3-4 users discontent that a custom script they use doesn't work exactly the same as it did before, because the script makes assumptions that it shouldn't make, and they refuse to request a new one. Headbomb {t · c · p · b}18:54, 21 April 2020 (UTC)
(edit conflict)Sorry, I wasn't aiming this question at you but at the people who wrote the forks. I don't think the change to the reference templates is a big deal, myself, it just needs a bot to clean out the now-unneeded |ref=harv and some repair done to User:Ucucha/HarvErrors. The problem I have is that it's more work to ask people to use new scripts rather than simply adjusting the old ones. @Headbomb:, the thing is that Interface Administrators can edit the user scripts and this isn't really a style disagreement but a bug. Jo-Jo Eumerus (talk) 18:55, 21 April 2020 (UTC)
This is not a bug in the script. As I said in the discussion where this back-and-forth should be happening, per WP:TALKFORK: The error scripts are still useful for the task described above. If you have full citation templates in a "Works cited" section, and one of them should be but is not connected to an {{sfn}} template, the unused one will show a brown warning message. See this version of Coniston railway station (England) for an example. Only one full citation in the Sources section shows the brown warning message. In this case, it is because the sfn template that is trying to use it omits the year. – Jonesey95 (talk) 19:05, 21 April 2020 (UTC)
There is consensus, a minority that refuses to update their scripts does not overrule consensus to provide core functionality to CS1 templates. Headbomb {t · c · p · b}18:59, 21 April 2020 (UTC)
Please link to the consensus. So far you've provided two links: one to a discussion in 2018 and a recent one in which very few people participated, where you requested the change. Which discussion produced the consensus? SarahSV(talk)19:06, 21 April 2020 (UTC)
Not answered, which is why I started discussing it here. The number of people on that other page is limited, and you were involved in making this change and therefore are defending it. SarahSV(talk)19:25, 21 April 2020 (UTC)
Trappist the monk, you've suggested two scripts to help with this, but both remove the error messages we need. This change means we no longer see when citations are listed but not used, or we have to put up with big brown error messages in FR and Selected works sections that use {{cite book}}, etc. If this can't be sorted out somehow, then we need to discuss this at AN. I recently tried to add a publisher's name that has one italicized word in it, and the template produced an error message because apparently italics are not allowed in "publisher="? But why not? It seems central decisions are being made that cause editors to lose a lot of time trying to fix or work around them. I don't know what to suggest, but there needs to be some sense of gaining wider consensus before making changes that might cause a problem. SarahSV(talk)19:37, 21 April 2020 (UTC)
This change means we no longer see when citations are listed but not used, or we have to put up with big brown error messages in FR and Selected works sections that use {{cite book}}, etc. So then keep using User:Ucucha/HarvErrors.js, which apparently does display the warnings you want to see. You can't simultaneously want messages and not want them as the same time. Headbomb {t · c · p · b}19:41, 21 April 2020 (UTC)
I want to see error messages for unused sources when I am editing an article which uses anchors. When I am following the style of an article which does not use anchors, I leave out ref=harv from cite book so that I will not get false error messages. Now that I can no longer use ref=harv there is no way User:Ucucha/HarvErrors.js or any other script can distinguish between the two cases. Dudley Miles (talk) 20:12, 21 April 2020 (UTC)
Again, wrong. CS1 and CS2 behave the same way, so however you 'ignored' errors with CS2 can be used with CS1. Typically done via the addition of |ref=none. Headbomb {t · c · p · b}20:16, 21 April 2020 (UTC)
I do not know why you keep raising CS1 and CS2. I just asked whether they work the same way and you explained they did not. A script which relied on distinguishing sources depending on whether or not they had ref=none would not help to find unused sources in the countless articles written relying on ref=harv to indicate an anchor. It would give vast numbers of false warnings. Dudley Miles (talk) 20:28, 21 April 2020 (UTC)
I keep raising this because you keep complaining that CS1 (e.g. {{cite book}}) is problematic while simultaneously claiming that CS2 (e.g. {{citation}}) is not when they now both work exactly the same. This has been explained to you multiple times, including in the sentence right above your reply where you claimed i said that CS1 and CS2 worked differently. It's really hard to have a conversation when you read "CS1 and CS2 behave the same way" to mean "you explained they did not". Headbomb {t · c · p · b}20:30, 21 April 2020 (UTC)
I do not think that you explained that CS1 means cite book and CS2 means citation. You just assumed that I knew and I assumed they are different protocols for which applied to both templates. Dudley Miles (talk) 20:45, 21 April 2020 (UTC)
I don't understand why Headbomb doesn't understand what is being said. There is some kind of miscommunication. The issue is that the script we used no longer works. It gives us error messages where there are no errors. If we disable it, we lose the error messages we want. Trappist the monk has come up with two suggested scripts but they had the same problem. The issue is that we need to be able to remove ref=harv. It was a benefit to have a set of templates that allowed that functionality. It made things easy and fast. What you saw as a flaw, we saw as an advantage. SarahSV(talk)21:05, 21 April 2020 (UTC)
And I don't understand what you don't understand about using |ref=none to suppress anchors, rather than "removing |ref=harv" which was a solution that only worked with CS1 templates. You couldn't remove |ref=harv in {{citation}} to suppress anchors, so I really don't know why you are so distressed by CS1 and CS2 behaving exactly the same way as each other. Headbomb {t · c · p · b}21:08, 21 April 2020 (UTC)
Your reply is an example of what I mean. This has already been explained to you. {{Citation}} isn't used as much. Most people use the templates you've just changed or caused to change. We can't go around adding ref=none to all these articles. SarahSV(talk)21:13, 21 April 2020 (UTC)
You keep suggesting a new script, but as I pointed out above, it would not help. A script based on ref=none would be a good solution if it had been used from the start, but changing to it in the middle is not going to help with error checking in the many existing articles which relied ref=harv to distinguish between sources which are intended as anchors and those which are not. It may just give false error warnings. Dudley Miles (talk) 21:34, 21 April 2020 (UTC)
Giving that you're not happy with how the current script works, and that you want a different behaviour, I fail to see how a new script that worked like you wanted would "not help". You've got a path forward. If you don't want to take it, that's on you. Headbomb {t · c · p · b}21:39, 21 April 2020 (UTC)
So that's why I am suddenly getting HarvRef errors on articles where I don't even use HarvRefs. Stuff like this explains why I manually formatted citations for years, and regret now having switched to citation templates. Please fix this. SandyGeorgia (Talk) 21:51, 21 April 2020 (UTC)
If you went to WP:SCRIPTREQ to explain what is a "good warning" and what is a "bad warning" then you could have a script that keeps the "good warnings" and throws away the "bad ones". Headbomb {t · c · p · b}22:30, 21 April 2020 (UTC)
I just tried it. That script turns off at least one of the good warnings. If you use a long citation but not a corresponding short one, there's no warning. It's important to be able to check that kind of thing when doing source reviews. Doing it manually can be very time-consuming. SarahSV(talk)22:05, 21 April 2020 (UTC)
I really don't have time to deal with another citation/bot issue right now, and seriously regret having moved from manual citations to citation templates, because of the never-ending vagaries of bots and citation templates. I am unwatching this discussion. If a solution is found, Sarah has my permission to edit my userspace to implement whatever solution she finds that works. If she does that, I will do the appropriate ctrl-whatever to reload. I assume that whatever solves this problem for Sarah will solve it for me. SandyGeorgia (Talk) 22:08, 21 April 2020 (UTC)
@SandyGeorgia: if you used "manual citations", you'd have no warnings and no error flagging whatsoever. So if the warnings annoy you, then use User:Svick/HarvErrors.js and then you'd have errors, without warnings. Or use no script, and have nothing flagged. Headbomb {t · c · p · b}22:33, 21 April 2020 (UTC)
I think we are talking past each other a little bit here. SlimVirgin, at this point there is a lot of discussion to go through, would you be able to summarize the cases in which you would like to see warnings? Kees08 (Talk)00:17, 22 April 2020 (UTC)
It would be better to close this thread and continue the conversation at the original venue, per WP:TALKFORK, or at WP:SCRIPTREQ if a new script is desired. Multiple editors have offered to help with a new script request at WP:SCRIPTREQ. Citation Style 1 templates are discussed and decisions are made at Help Talk:Citation Style 1 and archived in the talk page archives for that page. Any discussions that are held here about CS1 templates will effectively be lost. – Jonesey95 (talk) 00:53, 22 April 2020 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Loading gadgets on another Wikipedia
Latest comment: 4 years ago9 comments5 people in discussion
English Wikipedia has the gadget "(D)Reference Tooltips: hover over inline citations to see reference information without moving away from the article text (does not work if "Navigation popups" is enabled above)" and it's very useful. Our MediaWiki:Gadgets-definition sets this up using the line
That's for English Wikipedia (where I was an intadmin, until that was unbundled from sysop a couple of years back); but I'm asking about Wicipedia Cymraeg, for which the list of intadmins is much shorter. I rather think that if Llywelyn2000 (talk·contribs) wanted cy.wp to have gadgets, he would already have set up something at cy:MediaWiki:Gadgets-definition. Hence my desire to put code into my personal common.js --Redrose64 🌹 (talk) 20:35, 21 April 2020 (UTC)
As you mentioned there regarding dependencies, I think it will be resolved once you add above code.Please check the last point mentioned here.
2)Paste the contents of this file in your wiki just by changing 'en' to 'cy' in the url. The content model also should not change. I think this also requires higher privileges.
4)Now, since gadgets have not been run on your wiki, I think something needs to be done related to ResourceLoader which I have not got info from anywhere. My suggestion is to try the above solution and if it goes wrong, I think it can easily be resolved by rolling back. Sorry if I have been too detail. Adithyak1997 (talk) 13:28, 21 April 2020 (UTC)
You have totally misunderstood me. Wicipedia Cymraeg is not "my wiki", nor am I an administrator there. I'm just an ordinary Joe who occasionally edits there, but when doing so misses the rich variety of gadgets provided here at English Wikipedia (which is my home wiki). So my question at VPT was carefully constructed in such a way that it should have been clear that what I want to do is import the en.wp gadget for my own personal use at cy.wp. You should also have noticed that I had already worked out which files were necessary. --Redrose64 🌹 (talk) 18:40, 21 April 2020 (UTC)
Is there a way to find (free) files with more than one (not deleted) version in the history?
Latest comment: 4 years ago3 comments2 people in discussion
It was renamed by a maintenance script not bot. From the migration plans, it seems even if you recreate it now, the script will move it again. But more importantly, the title will be completely inaccessible when the Unicode upgrade take place, so there's really no point in recreating it since it'll soon not be useful anymore. – Ammarpad (talk) 17:07, 22 April 2020 (UTC)
Special:Search/ƀ already goes directly to Ƀ, and {{uc:ƀ}} produces Ƀ. A few links at Special:WhatLinksHere/ƀ could be updated but it doesn't seem important. A page at ƀ becomes inaccessible after the link starts going to Ƀ, so the moves without leaving a redirect are deliberately made before this. PrimeHunter (talk) 18:46, 22 April 2020 (UTC)
Random Link to a mathematics page
Latest comment: 4 years ago5 comments4 people in discussion
I would like to have a random link to a mathematics page, however most Mathematics pages aren't pages in the Category:Mathematics but in one of its subcategories. So http://wiki.x.io/wiki/Special:RandomInCategory/Mathematics doesn't work. Do you have any idea on how I could generate a link that leads to an article recursively?
PS: I am very new to wikipedia (joined yesterday) and I am glad for any advice you can give me :D
You can use the subcategory as the parameter, or am I missing something? It won't check recursively if that's what you're asking. --qedk (t愛c)08:16, 22 April 2020 (UTC)
There's no way to generate such a link using just wiki-markup. But you can employ the services of User:AnomieBOT/RandomPage for this. Copy paste one of the sample usages to a page in your userspace, put |categories = {{SUBCATS|Mathematics|10}}. The bot should edit the page periodically, adding a link to a random article. SD0001 (talk) 12:05, 22 April 2020 (UTC)
Thank you for your response. To clarify: I would like to have a link that I can put into my browser, similar to the Special:RandomInCategory link, so every time I open a new tab, I get to see a mathematics page and learn something new and interesting or edit an article I at least know something about.
Latest comment: 4 years ago11 comments7 people in discussion
Just noting that I'm still losing work through edit conflicts, despite having activated two-column edit conflict in beta. Anybody else experiencing this since it was last mentioned here] on 6th April? Nick Moyes (talk) 14:08, 10 April 2020 (UTC)
@AntiCompositeNumber: My content loss occurred whilst replying at the Teahouse. (It also occurred on a userpage on Czech Wikipedia, but there I assumed I didn't have the two column edit conflict tool enabled.) Nick Moyes (talk) 07:03, 12 April 2020 (UTC)
@Max Klemm (WMDE): Thank you for this update. There were times when there seemed to be no interface at all, just a complete loss of edits, so, in the short term, I'll try and remember to copy my replies before posting them at potentially busy fora. (Just in case!) Cheers, Nick Moyes (talk) 10:15, 14 April 2020 (UTC)
@Max Klemm (WMDE): I believe I had the same experience as Nick Moyes a couple of times on the Teahouse or Help Desk, with no conflict resolution facility (I think***) – just an edit window for the whole page (as opposed to just the section I was editing) with my changes missing. I was able to use the browser's (FFox) back button, accept the dialog about resending, and get back to the edit window with my edits in it. I then copied it to a text editor, restarted an edit on the section, pasted my changes back in, and saved successfully. ***It's entirely possible that the page I got had additional language on it telling me there was a conflict and what to do about it – I was too busy panicking and trying to recover to look very carefully. —[AlanM1 (talk)]—02:17, 15 April 2020 (UTC)
Latest comment: 4 years ago1 comment1 person in discussion
So, I'm trying to get the word count java-script thing, and I have to clear my cache, but I've tried many ways but it won't let me. Any advice?? TuorEladar (talk) 22:47, 22 April 2020 (UTC)
New revision of audio clip won't load?
Latest comment: 4 years ago4 comments2 people in discussion
I uploaded the different portion of a song at File:We Are Sex Bob-Omb clip.ogg, but it loads the same revision (but just trimmed down). Actually, I reverted back and forth, but the current 14-sec one plays something different from the one I intended. --George Ho (talk) 21:56, 22 April 2020 (UTC)
Still loads the same clip, which is not the one I intended. The 14-second one is the one I wanted, but somehow the 21-second version is trimmed to 14 seconds. --George Ho (talk) 01:29, 23 April 2020 (UTC)
On mobile web and mobile app, I can hear the clip that I intended to use. At first, it was strange. The audio was the same while logged-out. Then it turns out that I had to clear the cache of one of my browsers. Now I was able to hear the audio. --George Ho (talk) 07:19, 23 April 2020 (UTC)
Request: Could someone make a template which links to editing the page in Visual Editor?
Latest comment: 4 years ago4 comments2 people in discussion
Hi all
I'm currently working on Wikiproject Covid-19 and collating information from UN staff members who are experts on different aspects on COVID-19 e.g agriculture, maternal health , education etc. One thing that I'm finding really difficult is that this information is best displayed in tables but VE is disabled in the Wikipedia: namespace. My understanding is this is because all Wikiprojects on English Wikipedia are in the Wikipedia: namespace which is shared with many other functions and some of these break with VE.
It is not a realistic solution to ask experts and new users to learn how to edit tables in source editor and having tables easily editable would be extremely helpful in many Wikiprojects. I would like a template that I can put at the top of these pages in Wikiprojects which would say something like:
'Click here to edit this page with Visual Editor' and the link would simply add ?veaction=edit to the end of the URL
We already have {{VEFriendly}}. I think the main reason VisualEditor is not linked by default in Wikipedia space is that it's often used for discussions, like this page. PrimeHunter (talk) 10:50, 23 April 2020 (UTC)
I haven't looked at your quarry queries (should have time later today), but five quick comments based only on what's here: try logging_logindex instead of logging; pasting into the Toolforge sql optimizer would have suggested that; you need parens around AND (pr_expiry = 'infinity' OR pr_expiry IS NULL); this will entirely omit pages without a corresponding log entry (like if the page was moved, instead of being protected at its current title) - try a LEFT JOIN logging_logindex and allow log_type IS NULL as an alternative to 'protect'; WP:RAQ should be your first impulse for this sort of question, not WP:VPT. —Cryptic15:46, 23 April 2020 (UTC)
Is there a way to add other cite template(s) to the drop-down menu in the editing box?
Latest comment: 4 years ago3 comments2 people in discussion
This is more curiosity than anything else, but I could see it being useful. I know the "Templates" drop-down (next to "Named references") has the choices "cite web, book, news, journal". If I wanted to add one - like "cite AV Media", for example, is that possible? JimKaatFan (talk) 19:28, 22 April 2020 (UTC)
Thanks. I wasn't looking to force changes on the whole world, but I was wondering if there's a way to modify that dropdown just for myself. Can I change those settings you linked so that the changes only apply to me when I'm logged in? Or is that a technical impossibility? JimKaatFan (talk) 15:52, 23 April 2020 (UTC)
Xtools edit count link for IP contribs
Latest comment: 4 years ago3 comments2 people in discussion
Why doesn't the contribs page for IPs show the Xtools Edit count page (example)? Also, the "Edit summary search" and "Articles created" links are missing that registered users have.
Xtools stats are useful for analyzing problematic IPs, so the link would be useful to have there. Is there a technical barrier? --Pudeo (talk) 13:44, 23 April 2020 (UTC)
Suggestion: display certain hatnotes only if the user has been redirected from certain targets
Latest comment: 4 years ago11 comments8 people in discussion
nb: I made this suggestion at the redirection template talk page (discussion here) and on phub ( phab:T232278) but no one seems to be entirely sure where the right place for it is. Following another suggestion I'm trying it here, too.
I think we should display certain hatnotes only if the user has been redirected from certain targets. For example:
The Blue Album is an album by Weezer. It contains a song called "The World Has Turned and Left Me Here". There is also an episode of a TV show titled "The World Has Turned and Left Me Here".
If you go to the article, it currently displays this hatnote: The World Has Turned and Left Me Here" redirects here. For the Vampire Diaries episode, see The World Has Turned and Left Me Here (The Vampire Diaries).
I think this hatnote should only be displayed if the user has searched for "The World Has Turned and Left Me Here" and been redirected to Weezer (Blue Album). My logic is this:
If you're redirected to the Blue Album article because you were looking for an article about the TV episode, the hatnote is useful. It tells you that you're in the wrong place and gives you a link to the right article.
If you arrive at the Blue Album article any other way (most likely more than 99% of cases), the hatnote isn't useful - you almost certainly don't care that there is an episode of a TV show with the same title as a song on this album. It's just clutter.
It would be great if there were a way to display certain hatnotes only if the user has been redirected from certain targets. Popcornfud (talk) 15:33, 12 April 2020 (UTC)
I'd oppose this. Trying to second-guess what readers will want to see is rarely a fruitful exercise; to stick with your original example, I can easily imagine a fan looking for information on the album finding it interesting that the name of one of its songs was also used as the title of a TV show. Since no reader in the history of Wikipedia has ever found a hatnote disruptive (we're talking about one line at the top of an article), this proposal would seem to cause potential disruption to readers for no obvious benefit.
There's a high number of pages where we undoubtedly would need to display the hatnote even to those who typed in the article title verbatim (e.g. if someone searches for "The Sirens and Ulysses" there's a decent chance they're actually searching for "Ulysses and the Sirens" and vice versa). Thus, if we did go down this route, the display hatnote=yes or display hatnote=no would need to be set manually for every single redirect on Wikipedia. To put that in perspective, as of December 2019 when the report last ran there were 8,966,504 mainspace redirects; I'm guessing you'll have difficulty persuading anyone to review nine million pages for little if any obvious benefit. ‑ Iridescent15:55, 12 April 2020 (UTC)
Iridescent, I find your first paragraph unconvincing. no reader in the history of Wikipedia has ever found a hatnote disruptive (we're talking about one line at the top of an article - well, I find it disruptive! And in the case of the Weezer article it adds two lines about two separate articles, not one. I really do think it adds a lot of clutter that helps only a vanishingly small minority of users. Adding a big lump of text to the top of the article is not a small cost from a usability/readability perspective.
I can easily imagine a fan looking for information on the album finding it interesting This is neither here nor there. If encyclopaedic information is worth including in the article then it should be included in the article proper, not delivered in a sideways fashion via a hatnote. And as a general rule, per WP:POPCULTURE, we don't typically consider references to things in other media encyclopaedically notable anyway.
However, regarding the second paragraph: I have no experience of managing the technical side of this sort of thing so it would be up to other editors to figure out if what I'm proposing could realistically be implemented at all. If the implications of that are unmanageable then the idea is indeed dead on arrival, oh well.
I'm wondering, though, if we could add an optional flag on redirects to only display a hatnote on certain conditions, so that we wouldn't have to change any existing ones - it would just be added as an exception or override where useful. Popcornfud (talk) 16:25, 12 April 2020 (UTC)
Knowing the article from whence one came is not possible in wikitext/Lua. This would be a Javascript addition that would need to be made and I would accordingly oppose it as Not Worth It. --Izno (talk) 16:28, 12 April 2020 (UTC)
I see. I assumed we had that information in a way that could be used as it says "redirected from <wherever>" when you get redirected. Obviously even if it could still be used this might easily still be Not Worth It. Popcornfud (talk) 16:42, 12 April 2020 (UTC)
Even if it were technically possible, I wouldn't be in favor of it. Just as a reader wikisurfing, hatnotes have led me to surprising and interesting connections that I wasn't aware of. As an editor who uses Special:Random a lot, it would not be helpful if the version of the article that I see is not the same version all readers see. Schazjmd(talk)16:46, 12 April 2020 (UTC)
I think this could be done by adding a hash to the end of the redirect (e.g. #redirect [[Page title#hatnote-1]]) (which would need the hatnote template to support an ID parameter) then using css's :active pseudoclass to make it visible. NixinovaTC02:44, 13 April 2020 (UTC)
I like this idea a lot. I don't at all buy the "hatnotes can provide random interesting information" argument — that's not what they're supposed to be for, and they're occupying extremely valuable space at the top of the article. And it wouldn't be necessary to go through all 9 million redirects — just introduce a parameter and let it be adopted as needed. As to the technical side of things, I'm not going to be able to productively contribute to that, but if it's not possible now, perhaps we could tuck this away as a good idea to come back to once Wikipedia's coding is in an improved state. {{u|Sdkb}}talk05:45, 13 April 2020 (UTC)
Having hatnotes that people will not see when looking at the page directly seems like a potential vandalism time bomb where vandals can place offensive text that will only show up when someone reaches the page from a particular redirect. 24.151.50.175 (talk) 15:54, 20 April 2020 (UTC)
Hmm, that's a really good point. If this moves forward, something will have to be done to address that, but I'm not sure what. {{u|Sdkb}}talk07:10, 24 April 2020 (UTC)
Latest comment: 4 years ago1 comment1 person in discussion
Hi. I would like to ask for some technical help regarding an issue/bug with the Template:Graph:Chart. In a nutshell, when using the date format for the x-axis in combination with the line graph type, everything (almost, see below) goes fine; however in combination with the rect graph type (i.e. bar charts) the dates are not formatted properly and are shown in Unix date format. Furthermore, the date format – for any graph type, line, rect, etc – appears to be incompatible with the choice of xMin and xMax values, again for the x-axis. I would suggest that it's better to direct helpers to the existing discussion in the template talk page, where there are also graphical examples of the issues, and per WP:MULTI. Thanks! --Ritchie92 (talk) 08:56, 24 April 2020 (UTC)
Mystery of the Mixing Sailors
Latest comment: 4 years ago4 comments4 people in discussion
Today, I came across a nonsensical draft whilst patrolling new pages titled Draft:Mixing’Sailors, containing the following text:
1. A rich mummy, usually green or bald.
2. To overbreak toes, i.e. to put a cat into a fish or into a dog.
3. To stab an elephant with a nostril.
Mystified by who would think up such a thing, I looked up the text on the Internet and found the same thing pasted on many other wikis and user-editable sites, as well as elsewhere on Wikipedia. Also, the page title Draft:Mixing Sailors, without the apostrophe, has actually been protected from creation using the title blacklist. Does anybody know who is spamming this stuff across the Web, and for what reason? Passengerpigeon (talk)13:28, 18 April 2020 (UTC)
It's old and appears to always be added by French IP addresses or new accounts. User talk:81.185.159.77 is from 2009. It's called "The ideal joke" at User talk:79.85.113.185. It was globally blacklisted in 2011.[46] It's rare enough to just be one odd person. User:Croobatch posted it and also made Oil and clown colors with only content: "Bill and the maps are sticking well together because they appear to be made of swollen iron and greasy spokes." They got a checkuserblock. User:Gdodjolsoljkfvg got one the same minute by the same checkuser. They created Gsthae with tempo! with: "Moosh drives like a horn because his tie is not more than a crablike queefish. Cats dubbed with strings, a horror diet clowned about a forshipmore then skies linked a dribbloo." I don't expect to find sense in any of this. PrimeHunter (talk) 14:57, 18 April 2020 (UTC)
The lesson, as always: Wikipedia editors will always do something more outlandish than you can imagine. P.S. I know Moosh. She does drive like a horn. – Jonesey95 (talk) 15:35, 18 April 2020 (UTC)
Safari (iPadOS) screws up when rendering Wikipedia (desktop)
Latest comment: 4 years ago2 comments1 person in discussion
I had noticed this issue before, but for some reasons, I didn’t bother to ask about this. Anyways, the issue is this:
Letters are misaligned, so sometimes words will overlap
If there’s a section title which should fit in one line in the navigation box, but has two words, Safari renders it in 2 lines
The issue gets resolved when reloading the page
Device: iPad (2017)
OS: latest iPadOS (no matter what I update it to, the bug stays)
When it began: I’ve forgotten (it’s been so long)
Has anyone else had to deal with this? RedBulbBlueBlood9911 (talk) 10:35, 24 April 2020 (UTC)
Post-posting comment: I noticed that this issue is only on Wikipedia, and that it usually happens between an ordinary word and a linked word. By the way, I don’t think different browsers can fix this (Apple demands that all browsers use WebKit, and I think I noticed this issue on Chrome on an iPhone 7) - RedBulbBlueBlood9911 (talk) 10:40, 24 April 2020 (UTC)
Removing pages from a category
Latest comment: 4 years ago5 comments5 people in discussion
The category is added by a template. If you want to remove empty categories then try a null edit of the category page which is in the category. PrimeHunter (talk)
An error icon appears near the citation. When you click it, it shows an error which is occured on that cite with the close button. --Ijoe2003 (talk) 13:24, 24 April 2020 (UTC)
Request, copy Template:Portal_navigation from Meta to English Wikipedia, 'Lua error in package.lua at line 80: module 'Module:Is rtl' not found.'
Latest comment: 4 years ago5 comments2 people in discussion
Hi all
I'm working on usability for a Wikiproject and want to use https://meta.wikimedia.org/wiki/Template:Portal_navigation however I don't understand how to copy it into English Wikipedia, I tried to copy it to Wikidata and it didn't work and someone had to fix it. If anyone knows how to copy it across I'd really appreciate it.
Update: I copied it across to Wikipedia but I'm getting a big red error message 'Script error: No such module "Portal navigation".' any help would be really appreciated. John Cummings (talk) 13:14, 24 April 2020 (UTC)
@Arbil44: "notifications" usually refers to the two icons to the right of your username at top of Wikipedia pages. Are you referring to notification mails? If so, do you have "HTML" or "Plain text" at "Email format:" at Special:Preferences#mw-prefsection-echo? "Plain text" sends a url as a plain text string, e.g. https://en.wiki.x.io/wiki/James_Gordon_(British_Army_officer,_died_1783). Your own mail software or webmail service may turn it into a clickable link and may omit an ending part in parentheses if it guesses it isn't part of the url. I don't know how much "HTML" encodes but maybe it will send a full clickable link so your mail software doesn't have to guess where the url ends. PrimeHunter (talk) 13:12, 24 April 2020 (UTC)
Thank you PrimeHunter. Your kind response is a bit too technical for me to fully understand, but I have several pages on my watch list and the links go straight to the relevant pages with no problem, but not the two I mentioned. Here is the latest notification which has caused me to ask the question now. I have emboldened the bit which isn't live in my email, but I suspect will go 'blue' here! Dear Arbil44,
The Wikipedia page James Gordon (British Army officer, died 1783) has been changed on 20 April 2020 by Necrothesp, see https://en.wiki.x.io/wiki/James_Gordon_(British_Army_officer,_died_1783)for the current revision.
Thank you again. Did as you suggested and it is already showing "HTML", so what now?! Keep it simple for me to understand please! Anne (talk) 13:47, 24 April 2020 (UTC)
@Arbil44: Watchlist mails are enabled at Special:Preferences#mw-prefsection-personal and not Special:Preferences#mw-prefsection-echo. They are not a part of the Echo notification system so the Echo preference has no effect on them. They always come as plain text (phab:T15303 from 2008 has a HTML request). It appears there is currently no solution for you, apart from changing to mail software which interprets the url string differently and includes a part in parentheses when it makes a link. But it's impractical to change mail software just for this. My own mail service (Danish webmail so not practical for others) includes the parentheses. There are also mail systems which don't make a clickable link for any url so you are better off than some. PrimeHunter (talk) 14:32, 24 April 2020 (UTC)
Thank you PrimeHunter. I appreciate knowing the situation. It was annoying me and I supposed it would be fixable, but now I know it isn't, then I will stop being irritated (or at least try)! Much appreciate your assistance, thank you. Anne (talk) 14:51, 24 April 2020 (UTC)
@Arbil44: It's not fixable for you. It could be improved in MediaWiki for some users but hasn't been done. Here are some technical details you are free to ignore since they don't help you. Wikipedia sends a mail like this:
The Wikipedia page Wikipedia:Village pump (proposals) has been changed
on 24 April 2020 by Shnizzedy, see
https://en.wiki.x.io/wiki/Wikipedia:Village_pump_(proposals) for the
current revision.
To view this change, see
https://en.wiki.x.io/w/index.php?title=Wikipedia:Village_pump_(proposals)&diff=next&oldid=952863338
For all changes since your last visit, see
https://en.wiki.x.io/w/index.php?title=Wikipedia:Village_pump_(proposals)&diff=0&oldid=952863338
Note that none of it is clickable. It's all plain text. When mail software displays the mail, it may automatically change some of the text to clickable links according to an algorithm chosen by the software. The algorithm in your current mail software apparently says that a link ends if parentheses are reached, so https://en.wiki.x.io/wiki/Wikipedia:Village_pump_(proposals) becomes https://en.wiki.x.io/wiki/Wikipedia:Village_pump_(proposals) instead of https://en.wiki.x.io/wiki/Wikipedia:Village_pump_(proposals). Many mail services do this. The usual complaint we get is not about watchlists but that a link doesn't work if a user mails the url. In that situation there is a workaround which works in most mail software: Replace ( by %28 and ) by %29: https://en.wiki.x.io/wiki/Wikipedia:Village_pump_%28proposals%29. Our software doesn't do this when it sends plain text mails about watchlist changes, and we have no HTML option for this. In HTML you specify exactly where a url starts and ends so there is no ambiguity. Another frequent problem is url's ending in a period like Martin Luther King Jr. which has url https://en.wiki.x.io/wiki/Martin_Luther_King_Jr. Nearly all software will think that the period is ending a sentence and is not part of the url. This includes MediaWiki which produces: https://en.wiki.x.io/wiki/Martin_Luther_King_Jr. It works out OK in this case because we made a redirect at Martin Luther King Jr without the period. PrimeHunter (talk) 15:06, 24 April 2020 (UTC)
Well, I certainly did you the courtesy of reading, rather than ignoring, your response, but I followed you only up to a point! Anyway, if I have to change my software/provider/whatever in order to get clickable links, well that would be a bit unnecessary. Anyway, you have been most helpful in explaining that I must just put up with the situation. One small irony is that I put this page on my watchlist, while this discussion was taking place, and the emailed link to this thread does not take me to this thread! Don't worry about explaining that since this matter is now 'resolved'! Anne (talk) 15:15, 24 April 2020 (UTC)
HELP! with medical cases chart
Latest comment: 4 years ago3 comments2 people in discussion
Oh THANK YOU. Sheesh, I don't understand this present chart as well as the previous style, where everything was clearly labeled, row by row...... I wish the documentation or tamplet page itself explained that as clearly as you just did. Shearonink (talk) 18:48, 24 April 2020 (UTC)
Making a personal fork of Wikipedia's Featured and Good articles.
Latest comment: 4 years ago3 comments3 people in discussion
Hi. I wanted to make a personal fork of the English Wikpedia's Featured and Good Articles (over 40,000 total). Uaing this ExtremeTech.com article as a guide, I downloaded a "Dump" file of the English Wikipedia (over 6,000,000 articles).
My questions are:
1. Is there an easy way to download only the Featured and Good articles (along with the necessary Template pages) rather than a dump of the whole encyclopedia? Or:
2. Is there an easy and efficient way to auto-delete articles; so that after setting it up, I can just delete the articles I don't want to keep? Such as a script or bot I can use, or point me in the direction of how to code one?
What you want to do is not terribly difficult, but it does require some programming skills. There is an API (Application Programming Interface which lets you do anything you can do via the web front end. There are client libraries which let you access this via a wide variety of programming languages. The basic gist would be to (have your program) look at Category:Good articles and Category:Featured articles and iterate over the entries, downloading the wikitext for each page and saving it locally. Figuring out which templates you need for each pages is only a little bit more complicated. You'll need to parse the wikitext and (possibly recursively) iterate over the transcluded templates, and download those via the api. Again, there's parsers available for a wide variety of programming languages. For somebody experienced in programming and familiar with the API and parser libraries, you could probably knock this off in an afternoon. If you're an experienced programmer but have never used these libraries, probably a couple of days to get used to how they work. -- RoySmith(talk)21:26, 24 April 2020 (UTC)
Deleting talk page history
Latest comment: 4 years ago4 comments3 people in discussion
Magnolia677 That is a consequence of archiving the history and text by moving it rather than copy-pasting. It is a valid, though rarer, form of archiving. You can see it in the first revision of that page. --Izno (talk) 22:07, 24 April 2020 (UTC)
Latest comment: 4 years ago4 comments2 people in discussion
I am not sure if this is the correct place for me to post this.
I am working on a tool that allows the review and answering of protected page edit requests. You can view the tool here. My question is how do we create an edit form so that reviewers do not have to switch between different tabs? Has someone already created one? And I am talking about the edit form that is kind of like the traditional Wikipedia edit form.Aasim23:21, 24 April 2020 (UTC)
@Izno:I have taken a look at the tables (and even have the editprotectedhelper script) installed. I am working on this both so I can learn more about MW gadgets/APIs and so other editors can have an additional tool to use on Wikipedia. Yes, we have all these categories, but it takes a ton of clicking to go to the article, edit the article, go back to the talk page, go back to the list, etc. I hope to make this a little faster with this script. My question was: How do we make an "editform"-ish because I have tried making an editform-ish but it did not work. I think I figured it out. Thanks for the suggestion, though! I will definitely consider looking at the EPH script, but I think for now I will just work on my second tool :) Have you tried checking here? Aasim08:33, 25 April 2020 (UTC)
Max page size
Latest comment: 4 years ago3 comments3 people in discussion
I don't think there is a maximum size for pages, but there are limitations on use of templates and processing time. One is the output of templates cannot exceed to 2MB (which is why pages with 500 plus citations using cite templates get into trouble). However, my understanding is that if it is simple text on the page it can be much bigger. — Jts1882 | talk08:37, 25 April 2020 (UTC)
Latest comment: 4 years ago4 comments2 people in discussion
Disclaimer: I'm asking on behalf of another wiki, so I understand if help may be limited in that regard. I am also a newbie with Javascript, so I only understand the barest of basics. I also apologise for the external links as I was unable to get the code to function properly on here.
On another wiki that I edit, StrategyWiki, there are two templates: Template:Control selector and Template:Control. Control selector takes in a string where each item is appended with a comma and renders it as a dropdown menu that functions as the switch, while Control works as the output and displays different content for the item selected in Control selector. You can observe how this works over on this test page: select an item from the dropdown menu on the right and observe how the icon to its left changes.
Control selector is dependent on Javascript in the wiki's Mediawiki:common.js file. Below is the code excerpted from the file:
Code excerpt from SW's MediaWiki:common.js
===Control Selector template===
*/
// Created by [[User:Prod]] with help from [[User:DrBob]]
// jQueryfied by [[User:Skizzerz]]
function selectControlSet( ) {
$('span.controlOpt').each(function() {
if ($(this).hasClass('control' + $('#control_selector_select').val()))
$(this).show();
else
$(this).hide();
});
}
function createControlSelector( ) {
var controlDiv = $('#control_selector_inner');
if (controlDiv.length == 0) return false;
var ControlSelector = $('<select class="ControlSet" id="control_selector_select" />');
ControlSelector.change(selectControlSet);
var sysTexts = controlDiv.text().split(',');
for (var i=0; i < Math.min(10, sysTexts.length); i++) {
ControlSelector.append($('<option>', {value: i}).text($.trim(sysTexts[i])));
}
controlDiv.empty().append(ControlSelector);
controlDiv.parent().css('display', 'block');
}
$( createControlSelector );
My question: is there a way for the dropdown menu item in Control selector to be remembered on future page visits? Using the test page above as a reference, is it possible to select the second (SNES) or third (GBA) item in the drop down menu, refresh the page, and have it stay on that dropdown menu item instead of reverting back to the first (PSX)? Doing some more research on the topic has led me to believe that user cookies may be involved somehow (this Stack Overflow thread mentions using jquery).
Awesome Aasim, that seems like it could be what we would use; thanks! Unfortunately I'm not much of a tech wiz, so I'm not sure how we'd incorporate GetItem and SetItem into the code. Would we add that somewhere around var sysTexts? —Tenryuu 🐲 ( 💬 • 📝 ) 18:01, 25 April 2020 (UTC)
A lot of the examples I've seen involved typing text and not so much selecting a dropdown menu item, so I'm not sure how much of that was applicable. :P —Tenryuu 🐲 ( 💬 • 📝 ) 18:03, 25 April 2020 (UTC)
WikiProject assessment category names
Latest comment: 4 years ago2 comments1 person in discussion
Latest comment: 4 years ago1 comment1 person in discussion
Please view question here and reply there, as this is where the helpee is looking for help. (I've also advised them of this forum's existence in case they need it in the future). Thank you and best regards, --Gryllida (talk) 05:28, 26 April 2020 (UTC)
Do rcats (redirect categories) influence the search engine?
Latest comment: 4 years ago7 comments5 people in discussion
I have long known, or believed to know, that the redirect categories will influence how redirects are treated both in the search results and in the the drop-down list of suggestions in the search box. So that, all other things being equal, a redirect tagged with {{R from misspelling}} will for example be ranked lower than one tagged with {{R from alternative name}}. However, I've just realised that I don't know how I've learned that, and I don't find it in the documentation at mw:Extension:CirrusSearch. Is this really a thing or have I been deceiving myself? – Uanfala (talk)21:54, 25 April 2020 (UTC)
No, "relevance" also depends on the popularity of the page, apparently determined by the number of incoming links. E.g, a search for Donald brings up Donald Trump at the top (after Donald itself), while all other lesser-known Donalds are further down the list. SD0001 (talk) 14:46, 26 April 2020 (UTC)
I believe it used to be the case that redirects tagged as unprintworthy (directly or indirectly) were excluded from the search suggestions (possibly for anything other than an exact match) but this is no longer the case. Redirects that are unprintworthy will generally have fewer incoming links and so naturally de-prioritised, but phab:T24251 is asking for a way to exclude some pages from the search results/search suggestions (it's been around since 2012 and triaged as low priority though so don't hold your breath). It has been a long-standing principle at RfD though that we don't delete redirects solely to avoid them appearing in the search suggestions box if they are otherwise useful, as the suggestions are only shown to a subset of users (mainly those with javascript enabled who use the internal search engine, the official Android app links to the target directly) and it is not knowable what proportion of users this is. Thryduulf (talk) 15:45, 26 April 2020 (UTC)
Suppressing reference group naming
Latest comment: 4 years ago3 comments2 people in discussion
Hi, I asked this question earlier at the Teahouse, and after some help from AlanM1, I thought about coming here and asking it instead. I occasionally edit on other wikis, and there in this one page where I want to insert the reflist in its body. AlanM1 did this over at the Teahouse and it works:
However, when I copy this exact same code over to the other wiki, the superscript is rendered as "lower-roman 1" and et cetera. What does Wikipedia have in its code that allows it to do this? —Tenryuu 🐲 ( 💬 • 📝 ) 21:15, 26 April 2020 (UTC)
Comment: One possibly "helpful comment to this problem" is copied here => *So now we have two options: either fix Facebook's parenthesis bug (for those who think that it is a bug), or have a bot (with consensus for approval) mass create redirects of the form "Foo (bar" for every existing article or redirect of the form "Foo (bar)", redirecting to either "Foo (bar)" (if it is an article), or the same target that "Foo (bar)" redirects to (if it is a redirect), avoiding double redirects. Also, if "Foo (bar)" is a redirect, then the bot would tag "Foo (bar" with "{{R avoided double redirect|Foo (bar)}}". Which option do you prefer, fixing Facebook or having a bot mass create redirects of the form "Foo (bar"? If the former, then the redirect should be deleted. If the latter, then the redirect should be kept. GeoffreyT2000 (talk) 17:17, 26 April 2020 (UTC)
Currently Template:! does not display wikilinks, bolding, or italics when you click on the images containing a popup text. The popup text is currently displayed as a caption, but the wikimarkup does not show up.
Sample images with transparent parts
Here is an example example adapted from Template:Photomontage. (Click on the images: Notice that "Four", "Stars", and "liquid" are not wikilinked--even though the text for the pipe link on "liquid" shows up correctly, it does not wikilink. Also notice that "dices" and "blue" lack italics and "outlines" lack bolding.)
I understand that the tooltip iteslf cannot be wikified, but wouldn't it be great if the wiki markup would show up on the individual file pages for PNG transparency demonstration 1, White Stars 3, White Stars 2, and Beakerblue?
{{sidebar|image={{Photomontage
| photo2a = PNG transparency demonstration 1.png{{!}}[[Four]] transparent ''dices''
| photo2b = White Stars 3.svg{{!}}Three white [[stars]] with black '''outlines'''
| photo3a = White Stars 2.svg{{!}}Two white [[stars]] with black '''outlines'''
| photo3b = Beakerblue.svg{{!}}A beaker full of a ''blue'' [[Liquid water|liquid]]
| size = 350
| spacing = 8
| color = transparent
| border = 0
| text = Sample images with transparent parts
| text_background = transparent
}}}}
@Epiphyllumlover: this seems like a good idea for sure, but also, I'm not sure it's so much a community proposal as it is a technical issue with the software preventing this from working - WP:VPT might be a better venue. Happy to be corrected if I'm missing something though! I've also slightly reformatted your post just so it takes up less space, I hope you don't mind :) Best, Naypta ☺ | ✉ talk page | 14:14, 26 April 2020 (UTC)
At first I attempted to post this at Phabricator, but I lack an email on my account so I am not able to. Additionally, I'm not sure if it has broader applications past the montage template. I have no experience with it, and wondered if people here can comment about this. At some level it could be a technical issue, but the policy aspect of it could be "exactly how much wiki markup do we want to allow?" and "should this just be for photomotage tooltip texts, or are their others?" I personally want to allow bolding, italics, and wikilinks--but maybe someone here has other ideas for what should be allowed in addition to that. Also I am not sure if Template:! is used on other templates besides photomotage, and of allowing wikimarkup could possibly harm them if wikimarkup is enabled. Thank you for fixing my post. For now intend to wait and see if anyone else comments here, and then post at VPT. (Yes, you are right, I should have posted there instead.)--Epiphyllumlover (talk) 14:37, 26 April 2020 (UTC)
@Epiphyllumlover: makes sense! My personal take would be that there's no particular policy reason not to allow all the standard wikitext in image captions like the ones you've shown, but there may be technical reasons why that would be difficult to do - I've not looked at the code for it myself, so couldn't say. If it's a question of community endorsement, though, my !vote is with allowing it - wikitext isn't restricted in that way anywhere else, so I don't see why it should be there. Naypta ☺ | ✉ talk page | 15:12, 26 April 2020 (UTC)
@Epiphyllumlover: thanks for bringing this up — I agree with Naypta in that I don't see any policy reason to disallow this (I actually was just wondering about this the other day, as part of implementing {{WP}}, which produces tooltips like this: WP:BOLD). I'd support moving this to the technical pump if no one objects, since I think technical hurdles are more likely to be the thing blocking its implementation than community consensus. {{u|Sdkb}}talk17:56, 26 April 2020 (UTC)
Not directly relevant to the issue at hand here, but is "dices" really regarded as correct? Maybe I'm just an old dinosaur, but my understanding is that the singular is "die" or "dice", but the plural is always "dice". Phil Bridger (talk) 18:21, 26 April 2020 (UTC)
Whilst this is completely irrelevant to the matter at hand, die in this context is a noun, not a verb, so it doesn't have a third person present. The plural of die is dice, but dice and dices are also in use internationally, and it really doesn't matter - everyone understands either way, it just ends up getting hugely prescriptivist when it comes to the differences. Naypta ☺ | ✉ talk page | 19:26, 26 April 2020 (UTC)
I bow to your superior knowledge. Of course "dice" is often used as the singular, but I have never heard "dices" being used as the plural. Phil Bridger (talk) 19:32, 26 April 2020 (UTC)
(after edit conflict) It is a noun, and is clearly used as such in the top left image shown, so it doesn't have a third person present form which only verbs do. The only verb that has a third person present form of "dices" is the verb "to dice", which is something completely different. Phil Bridger (talk) 19:29, 26 April 2020 (UTC)
@Epiphyllumlover: This is nothing whatsoever to do with Template:! and everything to do with how the MediaWiki software deals with images. When you write | photo2a = PNG transparency demonstration 1.png{{!}}[[Four]] transparent ''dices'', you are merely passing PNG transparency demonstration 1.png|[[Four]] transparent ''dices'' to the [[File: ... ]] syntax. That treats [[Four]] transparent ''dices'' as alt text. You can see that if you can hover over each image, or examine the page source. Alt text becomes an attribute of the html image tag, so cannot contain wikimarkup (or any other markup), and the software removes it. There is no html markup that specifically creates tooltips, although some browsers may render the 'title' attribute as a tooltip. Optionally, you can create a custom tooltip using CSS, as described at https://www.w3schools.com/css/css_tooltip.asp. In any case, you would really need to modify Module:Photo montage either to accept an extra parameter (like |tooltip1=, etc.) per image, or to explicitly extract unnamed parameters and use them as tooltip text. --RexxS (talk) 20:08, 26 April 2020 (UTC)
Thanks RexxS for the pointers there! I've done some digging and MMV, the media viewer used on Wikimedia sites, does support a separate caption - the code for it is here. At the moment, as the container for the thumbnail is empty, line 356 is called, so the title property of the anchor tag is made to be the caption of the image. So, if we want to address this, the way I see it, we'd need to go about it by firstly modifying the template for embedding images such that it adds the HTMLified wikitext to the anchor, perhaps as data-caption-wikitext or the like, and then just modify that line so that it becomes return $link.data('captionWikitext') || $link.prop( 'title' ) || undefined;. Seems eminently doable, what're people's thoughts? Naypta ☺ | ✉ talk page | 20:46, 26 April 2020 (UTC)
@RexxS:, you noted that it "cannot contain wikimarkup"--This seems to partly contradict what I see in the example above. The phrase "liquid water" shows up as "liquid" in both the tooltip and the caption for the image when you click on it. That it doesn't display with all three words along with square brackets looks like wikimarkup to me. Are you able to explain more of why it shows up as liquid? (I still don't know why.) Also, thank you for showing how to make tooltips with CSS; I did not know that. Naypta, thank you for researching this. --Epiphyllumlover (talk) 20:53, 26 April 2020 (UTC)
It seems as best as I can tell that Sdkb wants wikilinking and other markup to also show up in all tooltips themselves, and not just in the captions of images. This must be possible because WP:Reference tooltips do it all the time. I am not sure whether or not Naypta's solution to modify the template will also do this, but I tentatively think it won't. Is it better to separate or combine these two ideas for adding wikimarkup? (I suppose one would need to modify the code for generating tooltips in general instead of just the template for embedding images).--Epiphyllumlover (talk) 21:47, 26 April 2020 (UTC)
@Naypta and Epiphyllumlover: When I click on one of the images, I am taken to the file description page because I don't enable Media Viewer by default. It's quite possible to arrange the image parameters to pass other information to Media Viewer, but I was talking about the rendered html when we view the page here or on the file description page. Apologies that I hadn't realised you were talking about Media Viewer in your posts. "WP image pages" means something different to me.
As for the question about liquid: if I enable Media Viewer, the caption I see is "A beaker full of a blue liquid", which is exactly what I would expect when the MediaWiki software strips out the wikimarkup from A beaker full of a ''blue'' [[Liquid water|liquid]] in preparation for using it as alt text in the original page. That seems to be what gets passed to Media Viewer, and is identical to the alt text for fourth image on this page. In case that isn't clear, the wikimarkup stripped out consists of the four apostrophes, the four square brackets, the pipe character, and the Liquid water link target. Removing those leaves the seven words used as the caption in Media Viewer (as well as the alt text on this page). the same goes for the other three images. Is that any clearer? --RexxS (talk) 22:14, 26 April 2020 (UTC)
I didn't know it was called "Media Viewer"-- but now I do thanks to you, and yes the media viewer captions were what I was trying to write about earlier. As a tool tip, I see "A beaker full of a blue liquid" and the same "A beaker full of a blue liquid" shows up in media viewer. So it renders the pipe link coherently, except it doesn't actually wikilink or do bolding and italics. Even weirder, the Template:H:title template is completely unable to render any markup, even the square brackets:
Sorry for the talking at cross purposes. I think it probably bears writing down some core definitions and facts here - it looks like we're all on the same page about these now anyway, but I think it's a good idea to write them down for anyone else who comes across this conversation and is confused.
Tooltips are the text that appear when you hover over an image. They cannot be formatted specially, according to web standards.
Media Viewer (mmv) is the popup image viewer that opens up as a modal dialog over the page when you look at an image, assuming you have it turned on (which is the current default). It is technically possible for it to display HTML content (formatted wikitext), and would require relatively trivial changes to do so I think.
File pages are a key part of MediaWiki, and changing them is possible, but might require significant changes in the software.
As I understand it, the request Epiphyllumlover is making is for the Media Viewer captions, normally autogenerated from the titles of the links surrounding the images on enwiki, to support wikitext. This is possible as I highlighted above by changing the template for images on the wiki, and changing very slightly a short bit of code in mmv here. If there's consensus that this would be a good idea, I'm happy to take this little project on myself. Thoughts? Naypta ☺ | ✉ talk page | 22:54, 26 April 2020 (UTC)
Why does Wikipedia keep showing edit conflicts on my own edits?
Latest comment: 4 years ago3 comments3 people in discussion
Why does Wikipedia keep showing edit conflicts on my own edits?
I'll make an edit to a section or a page > send Publish Changes, receive an Edit Conflict page > and have to cut & paste from that to a New Edit page > sometimes it works.
No others contributing to the page at the time.
Fustrating!
Zulu Papa 5 * (talk) —Preceding undated comment added 18:32, 26 April 2020 (UTC)
Might you be somehow hitting the "Publish" button twice? It happens to me if I don't wait long enough for the submit to happen. Perhaps your browser is bouncing the click, or your keyboard or something is intervening there, or it's just user error? Elizium23 (talk) 23:21, 26 April 2020 (UTC)
Latest comment: 4 years ago4 comments2 people in discussion
Resolved
Hi there! I have a global.css file. It does change how things look on meta and commons for example but not on en.wiki or fi.wiki. Neither of those non-compliant wikis has a common.css either. (The fi one used to however for what it's worth.) I wonder what the problem is. --Palosirkka (talk) 09:09, 28 April 2020 (UTC)
Works for me, on both enwiki and fiwiki. What browser do you have? Did you try to do purge if it's some problem with cache? Stryn (talk) 15:20, 28 April 2020 (UTC)
Interesting. I'm using Firefox ESR. I tried purging a page I then edited and no change. Did you mean that or some other page? Palosirkka (talk) 10:17, 29 April 2020 (UTC)
@ImSonyR9: Remove that and try pasting {{subst:lusc|User:Qwertyytrewqqwerty/DisamAssist.js}} this. I also recommend using Twinkle via the Gadgets interface rather than loading it via the userscript (it's basically the same but whatever). --qedk (t愛c)14:35, 29 April 2020 (UTC)
redirect mystery
Latest comment: 4 years ago3 comments2 people in discussion
If I follow any link and then return using the browser's "back" facility, that message is no longer shown when I have returned. It also disappears if I simply refresh the page (F5 in my browser) - perhaps you did that by accident. --Redrose64 🌹 (talk) 19:56, 29 April 2020 (UTC)
Reminder: Technical maintenance planed
Latest comment: 4 years ago3 comments3 people in discussion
This is a reminder of the message sent on Monday 27th April.
A maintenance operation will be performed on Thursday 30th April at 05:00 AM UTC.
It impacts all wikis and is supposed to last a few minutes.
During this time, new translations may fail, and Notifications may not be delivered. For more details about the operation and on all impacted services, please check on Phabricator.
A banner will be displayed 30 minutes before the operation.
Please help making your community aware of this maintenance operation.
Latest comment: 4 years ago2 comments1 person in discussion
Greetings,
I am experiencing two separate and rather frustrating issues with my watchlist. I asked on the HD a few days ago and was directed here.
1) Any page I edit is automatically added to my watchlist despite the fact that "Add pages and files I edit to my watchlist" is unchecked in preferences.
2) Any page in my watchlist is subsequently removed from my watchlist permanently, the next time I edit the page.
Latest comment: 4 years ago2 comments2 people in discussion
I'm not sure whether this belongs here since it's not a Wikipedia technical issue, but anyway... has anyone had issues using the Internet Archive specifically to archive tweets lately? I tried earlier today [49] for a tweet which is still live, and for some reason the archived copy says the tweet does not exist. Is there a gratis alternative? 209.166.108.199 (talk) 04:39, 30 April 2020 (UTC)
This is because of the new Twitter interface, which prevents the Wayback Machine from displaying pages correctly because the Wayback Machine doesn't handle JavaScript well. The easiest alternatives are to change your user agent and use the old Save Page Now so that it gets saved using one of the old interfaces, or to use archive.today. (For example, if you use a Firefox or Chrome add-on to change your user agent to that of Internet Explorer 11, or use curl or wget, viewing https://web.archive.org/save/https://twitter.com/HowieCarrShow will save the page using the previous Twitter interface.) Jc86035 (talk) 04:51, 30 April 2020 (UTC)
Problem with Open Street Map integration
Latest comment: 4 years ago6 comments4 people in discussion
I'm not sure what the official name of the tool is, but I'm using the "Map all coordinatates using Open Street Map" link on Category:Wikipedia requested photographs in the Bronx. That gets you to here. The problem is that when you mouse over the links in the left-hand nav bar, the map automatically scrolls to the coordinates of each link. This is fine as long as you're at the default zoom. But, go to the link above, then click "+" a few times to zoom in several levels. Now, move the mouse over the nav bar; the map scroll around crazily to track whichever link you happen to be over at the moment. Uber-annoying. Does anybody know any way to stop it from doing that? -- RoySmith(talk)21:33, 26 April 2020 (UTC)
That tool is maintained by DB111, Kolossos, and Plenz. If I remember correctly, that behavior is part of the map library, and I don't remember how configurable it is. It might be useful to convert the tool to a more modern library, but that takes work. --AntiCompositeNumber (talk) 16:28, 28 April 2020 (UTC)
I'm not a core maintainer of the tool, so I can't help. But I wrote a similar tool with subcategory capabilities, even though it won't help in your case, because it cannot handle talk pages. --DB111 (talk) 18:24, 28 April 2020 (UTC)
I hadn't heard about your tool, but I've already found it useful. It is more stable and doesn't jerk around. I wish there was a way to generate permalinks to settings besides location based categories. For example, it would be nice if one could wikilink to any single point directly with it set to view images from Wikimedia Commons. If someone could develop this, maybe Wikimedia Commons could switch to entirely using this tool when linking to coordinates.--Epiphyllumlover (talk) 20:56, 29 April 2020 (UTC)
Resolved. :) --Tenryuu 🐲 ( 💬 • 📝) 20:15, 22 March 2020 (UTC)
The new version of MediaWiki will be on test wikis and MediaWiki.org from 24 March. It will be on non-Wikipedia wikis and some Wikipedias from 25 March. It will be on all wikis from 26 March (calendar).
Resolved
.PNG)
to the left of "Advanced". Did you clear cache with only F5, with Ctrl+F5, or clear the entire browser cache? See Wikipedia:Bypass your cache. PrimeHunter (talk) 17:41, 22 March 2020 (UTC)
Works for me @RoySmith: check now, you may need to clear your cache. — xaosflux Talk 16:36, 26 March 2020 (UTC)
), and whether it happens on other websites of reliable sources. -- Ohc ¡digame! 16:38, 26 March 2020 (UTC)
You are invited to join the discussion at Talk:Visual pollution#Old revision is shown for logged out users. —andrybak (talk) 16:33, 31 March 2020 (UTC)
Fixed Right now, I can't think of anything else that needs to be done, except for the script to be tested by a file mover. —guywan (talk • contribs) 18:05, 31 March 2020 (UTC)
--qedk (t 心 c) 20:04, 31 March 2020 (UTC)
Done @Dank and Johnboddie: the page has been restored. — xaosflux Talk 15:38, 15 April 2020 (UTC)
Moved from Wikipedia talk:Village pump (technical)
The small wiki toolkits is to help smaller wikis that need technical skills. They can learn and share technical skills. [42]
Facepalm --qedk (t 愛 c) 20:54, 20 April 2020 (UTC)
these pages: the main project page, Updates, Replying, and User testing.
.jpg){kind=link}
Resolved