Wikipedia:Village pump (technical)/Archive 81

• Village pump
  • Policy
  • Technical
  • Proposals (persistent)
  • Idea lab
  • WMF
  • Miscellaneous

Village pump (technical) archive  

This page contains discussions that have been archived from Village pump (technical).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.

---

< Older discussions · Archives: A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P, Q, R, S, T, U, V, W, X, Y, Z, AA, AB, AC, AD, AE, AF, AG, AH, AI, AJ, AK, AL, AM, AN, AO, AP, AQ, AR, AS, AT, AU, AV, AW, AX · 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217

The links table ("What links here") not getting updated?

If I'm reading Help:What links here and Manual:Job queue correctly, it appears that when a link in a template is changed, the links table gets updated in a while, via the job queue. That is, if page A transcludes template T, then when the template T is edited to add or remove page B, eventually the backlinks (what links to page B) will be updated, automatically, to include to exclude page A (respectively). Is this right?

Well, I moved Raghuvamsa to Raghuvamsa (dynasty) (and changed Raghuvamsa to redirect to something else), and also edited the Template:Ramayana to point to the new target Raghuvamsa (dynasty), both over five days ago. And yet, looking at Special:WhatLinksHere/Raghuvamsa shows a lot of pages that don't actually link to Raghuvamsa but merely transclude Template:Ramayana. Similarly, Special:WhatLinksHere/Raghuvamsa_(dynasty) doesn't show a lot of pages which do transclude Template:Ramayana and hence link to it. (A couple of days ago I manually went and made minor edits to quite a few pages that use the template so that the table got updated, but I've decided to ask about investigating the bug now.) What's going on? Is this a bug in the system or in my understanding of what's expected to happen? Or has, by some chance, the job queue just not yet got to it in five days? Shreevatsa (talk) 18:42, 11 October 2010 (UTC)

Specifically, these are some pages that are in Special:WhatLinksHere/Raghuvamsa just because they transclude the template (please don't edit these articles now, or the bug will go away): Sulochana (Ramayana), Shrutakirti, Mandavi, Lavanasura, etc. Shreevatsa (talk) 09:34, 12 October 2010 (UTC)

I have made similar observations recently, but I wasn't aware that there is supposed to be an update queue dealing with this. Hans Adler 09:53, 12 October 2010 (UTC)

• This case should be fixed, I gave the servers a polite kick in the pants to get them to clear up the mess. ΔT _{^{The only constant}} 12:56, 12 October 2010 (UTC)

Thanks! It's indeed been fixed. What had happened? Shreevatsa (talk) 13:06, 12 October 2010 (UTC)

Ryan Lane (one of the WMF devs) recently fixed an image scaler issue that was causing job queue issues. Ive kicked the pages related here, if this continues to happen we may want to hand this issue over to a dev for further investigation. ΔT _{^{The only constant}} 13:08, 12 October 2010 (UTC)

old Edit toolbar not working

i'm getting no response when clicking the buttons on the old Edit toolbar (bold, italcs, etc.) and inserting wiki markup code. is there a problem? --emerson7 18:49, 11 October 2010 (UTC)

Works for me. Do you have JavaScript disabled in your web browser? That would cause it to stop working. --Cybercobra (talk) 19:06, 11 October 2010 (UTC)

all was well until this morning. i'm having the problem in firefox and ie on three different computers. --emerson7 19:25, 11 October 2010 (UTC)

There is a null error at function getCaretScrollPosition on plugins.combined.min.js because $(this).data('wikiEditor-context') returns null if the old edit toolbar is used. So Vector expects the use of the new toolbar. Merlissimo 21:11, 11 October 2010 (UTC)

I cannot reproduce this. I have selected the old toolbar and tried with Firefox. This morning a new version of some of that code was deployed. Perhaps you still have a partial old version of the code that is causing this problem. Try shift reloading the page to bypass the browser cache. —TheDJ (talk • contribs) 21:30, 11 October 2010 (UTC)

tried it (shift reload) ....still no affect. --emerson7 22:08, 11 October 2010 (UTC)

Should be fixed with rev:74655. Merlissimo 14:40, 12 October 2010 (UTC)

resolved thanks! --emerson7 18:34, 12 October 2010 (UTC)

Wikipedia displaying funny?

Wikipedia is displaying funny for me...

• Instead of the normal categories "box" at the bottom of a page, there's just a list.
• Twinkle's gone
• Lupin's gone
• HotCat's gone
• WikEd's gone
• Instead of bing tabbed, my preferences is just one long list
• Image thumbnails don't display with the little box around them

I'm sure I'll find more soon... --- cymru lass ^{(hit me up)}⁄_{(background check)} 01:57, 12 October 2010 (UTC)

Also, when I watchlist a page, Wikipedia usually just puts the "you have watchlisted [blah blah blah]..." notice at the top of the page. Now it's actually taking me to a separate page that says that.... --- cymru lass ^{(hit me up)}⁄_{(background check)} 02:30, 12 October 2010 (UTC)

It sounds like you're not loading JavaScript/CSS. This could be for a variety of reasons. Try a different Web browser to see if that helps diagnose your issue. --MZMcBride (talk) 03:50, 12 October 2010 (UTC)

I just tried it in one of the worst browsers I've ever used, and it worked! But it's still not working in Google Chrome. How do I fix this? --- cymru lass ^{(hit me up)}⁄_{(background check)} 04:41, 12 October 2010 (UTC)

For what it's worth, HotCat and Twinkle both seem to be working fine for me (Chrome/Chromium 6.0.472.62, Ubuntu 10.04). I suspect MZMcBride's JavaScript thoughts may be correct. TFOWR 11:07, 12 October 2010 (UTC)

I also use Chrome and really like it, and it works for me; and I have tested all major versions between Chome 4 and current Chrome 7 (development version or beta version) in a test VM ; my most frequent default version is still Chrome 6 (official release with its default automatic updates). verdy_p (talk) 12:50, 12 October 2010 (UTC)

I love, love, love Chrome. And I was using it fine earlier, and now all of a sudden... Poof! --- cymru lass ^{(hit me up)}⁄_{(background check)} 14:40, 12 October 2010 (UTC)

Okay, so I've got Twinkle, Lupin and Igloo back, and HotCat's showing up again, tpp. But WikEd is still AWOL, which is very annoying, because it's the tool I use most. --- cymru lass ^{(hit me up)}⁄_{(background check)} 22:35, 12 October 2010 (UTC)

Why Wikipedia...

sometimes freezing? I mean it doesn't load sometimes funnily enough. Every time the problem emerges I look at the task manager, but i can't find any processes that disturbes wikipedia. I checked other sites, but they work fine. I tried with other browsers, but its the same problem. I have no clue what's the problem is. Thanks.-- ♫Greatorangepumpkin♫ T 07:35, 12 October 2010 (UTC)

Show/Hide links for collapsible/expandable box does NOT work in English Wikipedia

The Javascript or CSS customizations made to the [show]/[hide] links displayed in the title par of expandable boxes does not work at all, here in English Wikipedia.

It's simply impossible to expand boxes that are hidden by default (notably all those boxes with "auto" when there are more than one expandable boxes in the page), because the clickin on it will just perform a normal link action, using the target anchor "#" which brings us to the top of the page.

Please correct it. Your current code is clearly bogous on THIS wiki only (at least it does not work at all with standard-compliant browsers),.

Please, don't generate static CSS properties in the HTML that will cause the boxes to be hidden by default (in addition, such static markup will cause the content of these boxes to not be indexed by CSS-aware web indexers like Google, and this reduces a llot the visibility of important articles with many references). It's up to the javascript (when it is enabled in the vistor's browser) to hide an expandable box. In other words, use exactly the same technic that is used for displaying the default TOC's? And don't deviate from it.

Note: the [hide]/[show] links in expandable boxes do work for example in French Wikipedia with all browsers using wellknown engines (Gecko/Mozilla/Firefox, Wekit/KHTML/Safari/Chromium/Chrome, Trident in all versions of Internet Explorer, or even exotic text-only browsers with very limited CSS support such as Lynx), including with alternate skins (Vector and Monobook), and with browsers whose javascript is disabled (in this case, expandable boxes are always shown and cannot be hidden at all). The default CSS state (without Javascript) should always be visible !

Your current solution also fails with WAI usability rules (users of screen readers that are CSS-aware, are unable to to expand these hidden boxes that are also hidden to them, as the link also does not work for them).

verdy_p (talk) 01:05, 12 October 2010 (UTC)

As a user of the screen reader JAWS, I recently noticed a change in the way hide/show links worked; they were previously normal links but they became same page links. However, the show/hide links work fine for me under the new arrangement. I can't comment on the rest of the message. Graham87 01:27, 12 October 2010 (UTC)

Try with Google Chrome: they act only as simple local "#" links that causes the browser to scroll up to the top of current page... verdy_p (talk) 01:45, 12 October 2010 (UTC)

Are you referring to this edit to MediaWiki:Common.js? Note that there are two kinds of collapsible boxes, with two different scripts using different techniques (and they both have small disadvantages/fails in matter of accessibility).

There is nothing wrong with using a href="#" itself on such trigger links for collapsible contents, it's even a best practice.

I did not say that this href attribute was wrong, just that it does not work or that it is incorrectly handled on user action, so that the normal link action takes places in priority, instead of the expected javascript event handler. And that those boxes are collapsed by default here (where they should only be collapsed by Javascript when it's active) and can't then be shown. verdy_p (talk) 10:10, 12 October 2010 (UTC)

But it should always be accompanied with return false; to prevent the bug you are reporting.

This is not what you are doing here (on other Wikipedia, and before the dif you have just shown here, the href contains a "javascript:collapseTable()", so there's no other possibility when interpreting the target of the link). But you are not returning false, but instead "killevt(evt)". But even if this was supposed to work, the collapseTable() call would have been called and would have shown the content, before branching to top of page with the "#" default link target, and this should remain in that state even if the page was scrolled up to the top.

It does not, so we can conclude that the target of the link specified by the href attributes still overrides the event handler that you have setup, or that it simply does not work and is incorrectly installed, or incompatible with all browsers because you are setting up incorrectly with an incorrect name javascript property like "onClick" vs. "onClick"...), or that it is installed on the wrong HTML element. verdy_p (talk) 10:23, 12 October 2010 (UTC)

As a member of the accessibility project, I have to say this is indeed an accessibility issue. Some browsers or assistive technologies may correct the bug because it's also their job to compensate mistakes of developers, just like JAWS 10 Graham87 is using. But it will fail on most configurations. See the W3C WCAG 2.0 technique: SCR35: Making actions keyboard accessible by using the onclick event of anchors and buttons.

There is also an important accessibility improvement that can be done on the other collapsible script (used on {{hidden}}). If a developer is interested, please contact the Wikipedia:WikiProject Accessibility. Also, when the devs will finally deploy JQuery to all skins we will finally be able to improve the user interaction of this collapsible box. Yours, Dodoïste (talk) 02:09, 12 October 2010 (UTC)

Much more likely is that something in one of your user scripts is breaking things. Try blanking User:Verdy p/monobook.js and/or User:Verdy p/vector.js and see if that fixes your problem. And please, tone down the vitriol. Anomie⚔ 03:06, 12 October 2010 (UTC)

Did you even tested what you suggested, or just investigated in it ? Certainly no, you are just formulating an assertion without any base here. And I was not send vitriol in my post, but YOU DID, that's why I reply now this way to you, ignorant and incredule !

Nothing's wrong in my custom javascripts (they don't interact and are completely unrelated to this issue). In fact I also have the same failure when not logged on (so when no using these customizations). Clearly the bug is not mine, but in the javascript used in this wiki (and that I don't experiment on the hundreds others where I may have frequently used similar imported customizations). verdy_p (talk) 10:01, 12 October 2010 (UTC)

I have used the DOM inspector included in Chrome, and I confirm that the custom javascript event handler that is supposed to set on the "onclick" property on the link is simply NOT installed at all (this property is still null, only the href="#" is there.) So this is clearly a bug of Commons.js specific to this wiki, or of one of its javascript library functions that do not properly setup this event handler ! In other words, this line does not work at all:

addHandler( ButtonLink, "click", new Function( "evt", "collapseTable(" + tableIndex + " ); return killEvt( evt );") );

Most probably, the bug is in addHandler(), why not simply doing this:

ButtonLink.onclick = "return collapseTable(" + tableIndex + " )";

and making sure that collapseTable() always returns false, instead of an unspecified value, and avoiding the instanciation of a new Function for just binding an unused "evt" parameter  ?

verdy_p (talk) 10:53, 12 October 2010 (UTC)

The working (tested) solution:

ButtonLink.setAttribute( "onclick", new Function( "evt", "return collapseTable(" + tableIndex + " ); return killEvt( evt );" );

I confirm this is a bug of addHandler()... verdy_p (talk) 11:16, 12 October 2010 (UTC)

I have just set the replacement code in my own User:Verdy p/common.js script (whose content is shared and imported by my vector.js and monobook.js). No more problem with all browsers (I've already tested 27 versions, on distinct versions of Windows, MacOSX, Linux, FreeBSD, Solaris, OpenVMS, and including old versions of Netscape, IE, Opera, Mozilla/Firefox, KHTML/Webkit/Safari/Chrome, Lynx... I can test up to about 135 combinations from my various VMs, plus a few commercial browsers). verdy_p (talk) 11:42, 12 October 2010 (UTC)

There's also a similar feature in Template:Collapsible list and in Template:Hidden that do work for me (with or without the correction for collapsible tables, because they don't attempt to set href="#" plus an onclick="Function(){...}" event handler, but directly a href="javacript:..."). verdy_p (talk) 11:49, 12 October 2010 (UTC)

I know now why addHandler() (or addClickHandler() which is a shortcut based on it) does not work, it does this:

/**
 * Add an event handler to an element
 *
 * @param Element element Element to add handler to
 * @param String attach Event to attach to
 * @param callable handler Event handler callback
 */
function addHandler( element, attach, handler ) {
    if( window.addEventListener ) {
        element.addEventListener( attach, handler, false );
    } else if( window.attachEvent ) {
        element.attachEvent( 'on' + attach, handler );
    }
}

(this code copied from wikibit.js, shared on all Wikimedia projects)

However, it depends on the test of whever the current window contains an addEventListerr() method, and assumes that the element also contains support for this method. However, browser windows DO have the possibility of adding handlers with this method, notably because the window has been surcharged by many other scripts when the page was loaded.

But this is not necessarily the case for all elements that are added later in the document. Such "attached" event handlers in fact only work for window events, not for events of elements in documents loaded in that window, and notably not for the dynamic elements that are created dynamically using document.createElement().

So element.addEventListener() just fails (no such method, or no implementation). This is a side effect of using the generic Javascript library, which adds many methods to the window, but does not override or extend the "createElement()" method in the loaded document, so that the created elements will also benefit from the compatible methods added to the window (event handlers for window or frame, i.e. containers of documents, are very different from event handlers on document elements, and this is expected because windows are not necessarily HTML elements and have much fewer properties, in addition to having more security restrictions to avoid XSS attacks).

The correction is simple (?? Not tested ??):

/**
 * Add an event handler to an element
 *
 * @param Element element Element to add handler to
 * @param String attach Event to attach to
 * @param callable handler Event handler callback
 */
function addHandler( element, attach, handler ) {
    if( element.addEventListener ) {
        element.addEventListener( attach, handler, false );
    } else if( window.attachEvent ) {
        element.attachEvent( 'on' + attach, handler );
    }
}

In other words, test the presence of the method on the element, because this is the instance on which you'll call the method, if it exists; testing the presence of the property on the window is not reliable.

verdy_p (talk) 12:17, 12 October 2010 (UTC)

Verdy, can you give an explicit example of page(s) and link(s) that are not working for you and state what browser version. The [show]/[hide] links I tested just now all appear to work correctly for me under the current version of Chrome. Dragons flight (talk) 12:22, 12 October 2010 (UTC)

All the horizontal navigation boxes at the bottom of articles (you may choose any one in those linked from the Wikipedia front page). (e.g. Joan Sutherland : the navigation box at the bottom of page, but all similar boxes are affected and are automatically collapsed but cannot be shown, in almost all browsers that I have tested, independantly of their version).

Note that the show/hide links are working within the vertical infoboxes at the top of articles, because they use different templates (Template:Hidden or more rarery Template:Collapsible list) that are simply generating href="javascript:...", and not "href="#" plus an "onclick" event handler for collapsible tables. verdy_p (talk) 12:41, 12 October 2010 (UTC)

The boxes at the bottoms of CFM International CFM56, Ernest_Augustus_I_of_Hanover, Treaty_of_Salynas, and Soyuz TMA-01M all work for me on the current Chrome. Are these not working for you? Dragons flight (talk) 12:49, 12 October 2010 (UTC)

No they don't work for me (if I disable my current correcting patch that I have just setup now in my User:Verdy_p/common.js script). I am also using with the official release Channel of Chrome (with its current automated updates). And not even when I am logged off (to make sure I use the default scripts of Wikipedia). When looking at the Chrome's DOM inspector (right click on the [show] link, choose "Inspect element" then in the right panel, open the "Properties" list and get the list of the first "HTMLElement"), you'll see that "onclick" is not set and is still null, even though the "href" property was correctly set to "#". So clicking on this link will just bring you to the top of current page... verdy_p (talk) 12:56, 12 October 2010 (UTC)

For the inspect element on the show link I see: <a id="collapseButton0" href="#">show</a> and Event Listeners -> click -> a#collapseButton0 -> listenerBody: function anonymous(evt) { collapseTable(0 ); return killEvt( evt ); } Dragons flight (talk) 13:36, 12 October 2010 (UTC)

I also see it, but it does not work. But you're not looking at the same place.

I was speaking at the "Properties" panel, in the properties of the HTMLAnchorElement, which says "onclick"="null".

Visibly, the null "onclick" property takes precedence to the list of event handlers of type "click", and is missing an implementation to start looking at the list of event listeners (which it would do if the HTML element was created as a container with a sandboxing window, something that the browser does not imply by default, because here there's no child window instanciated for the link content, which is just a text element decorated/styled by the anchor)... verdy_p (talk) 15:35, 13 October 2010 (UTC)

I also have Properties -> HTMLAnchorElement -> onclick = null, but the EventListener works fine and the box opens and closes as expected on the current Chrome (English version, Windows 7). It is true that setting onclick is a way to fix the problem, but either onclick or the EventListener should work. The system shouldn't require onclick in particular. Dragons flight (talk) 18:38, 13 October 2010 (UTC)

I have no idea how you changed your environment to extend some elements with that function and not the others, and it isn't normal, but Fixed in r74686. Will take a few months before it is deployed. —TheDJ (talk • contribs) 19:17, 12 October 2010 (UTC)

I repeat that I have not changed my environment. And this bug occurs even if I am logged off (so when I use the default Wikipedia settings). For some reason that I have not investivated a lot, this has to do with how addHandler() works, and how elements are created in the document using document.createElement() so that they don't inherit the behavior of their addEventHandler() method within some conditions that I have not investigated (or that could be caused by a security restriction).

However my correction (or workaround) for this issue is simple, and better matches what is documented by the W3C : don't use attached (and chained) event handlers, but only use the "onscript" property of document elements directly, without support for chained events (which were not created for document elements but for allowing document containers to filter the events before they are passed down to the contained document (there's absolutely no reason to use chained events on document elements as this breaches the separation between document containers and the documents itself, i.e. the wellknown document/view separation paradygm). Using chained events for this case is really overkill or a hack that is not supposed to work.

The only case where you'll want to use chained events on an element, is when the element itself is a container that must sandbox what they will display and manage in their internal layout (but even in this case, the "onclick" property of the element will handle their own container behavior in priority, before deciding to chain events in their own area, for the possible document that they may render within the element's area : this occurs for example with "frame" and "iframe" elements, or any kind of HTML that can embed and manage another document (for example to scroll it within the element's area. This is clearly not the case here as it is a simple anchor element containing plain-text that does not need to be controled by the anchor as if the text was a separate document to which the click could be forwarded by chaining. verdy_p (talk) 22:38, 12 October 2010 (UTC)

Well you are the only one reporting a problem. Also, there ARE plenty cases where we have multiple click handlers installed for a single element in pages (in Vector extensions, the clicktracking research tool and many gadgets), so this is the best way to handle this, in order to avoid onclickhandlers overwriting each other. This was a real problem in the past. The old methods was to use javascript:// links in the a href btw, but that is broken on Internet Explorer, which thinks you are leaving the current page if a javascript uri is executed (breaks the wikieditor). This reminds me btw that NavFrame still needs to be looked into. —TheDJ (talk • contribs) 00:12, 13 October 2010 (UTC)

I don't care baout these other cases, because I don't use any of those optional extensions. And in fact no extension will need to override the onclick event on hide/show links for installing their own handlers: either the whole show/hide feature is disabled completely (so all boxes will be shown by default and there won't be any such hide/show link generated), or it will be anabled and the hide/show feature should manage it completely, because it is generating these links itself within the page before hiding contents.

Your current problem is that you have installed by default a javascript library that overrides many methods within window, but not the objects created by document.createElement, so that they will also support the same methods (and notably the same addEventHndler() method): instead these elements are basic elements as they are created by the default native implementation in the browser. So it's certainly a limitation of the javascript libraries that are installed and that have forgotten this case, which starts appearing now possibly because new XSS attackes have been detected and the security model for elements vs. window/frame containers has been further restricted : may be this bug/limitation is already known by jQuery authors: document.createElement() possibly forgets to import a method because jQuery forgot to supplement the element prototype (in Chrome I see that documentcreateElement is "native", it's not overriden by a jQuery definititon). verdy_p (talk) 00:23, 13 October 2010 (UTC)

I don't care about your problem, because I don't use your browser.... </sarcasm and reversal of argumentation by editor>. Either figure it out and file a bugreport with detailed instructions on how to REPRODUCE the issue, otherwise people can't help you. You still haven't given anyone a proper path on how to reproduce your problem. —TheDJ (talk • contribs) 00:35, 13 October 2010 (UTC)

I DO have indicated how I get the problem. This issues occurs always, on all navigation boxes, on any pages. I also gave an example URL where it does not work, and confirmed that it also does affect the other URLs you suggested.

I have also indicated which workaround I use now for this issue. I also indicated with which browser at least I get the issue (Chrome, current release, French version), but I did not indicated my system, if this plays a role: Windows 7 (RTM) French version.

I have also said that this was NOT related to my customizations as it also occurs when I am logged off. And I have said that this only affected English Wikipedia (because this is the only wiki I use regularly that uses both href="#" and a chained event handler, instead of the method recommanded by the W3C which uses assignment to the onclick property.

Chained event handlers does not work on HTML elements (not even in "form" elements : it's up to the container window to manage this state when it really needs to control some events and bypass or change the UI events that elements are trying to bind on themselves...), and this will be strengthened in HTML5 and with newer browser versions, because of XSS security issues : chained event handlers are only meant for document containers (i.e. "window" objects, which are either a browser window client area, or a browser tab client arean, or a frame or iframe content area, where additional restrictions will be enforced by browsers to make sure that they are acting like javascript isolation sandboxes (documents should not be able to access to the property of its container window, unless they can provide a security token, or know a valid reference to it, transmitted via an external channel that explicitly gives it a back link to the window controlling it).

You have not demonstrated that a chained event handler was needed for an HTML anchor element ; stick on using the "onclick" property here (as recommanded by the W3C), and this will work as intended : chained events are overkill here, and are in fact an heritage from old browsers, when they mixed the document/view separation model exposing in documents too many things from the window container (and when they were full of XSS security issues, that are increasingly being closed in browsers ; in addition, chained events are much slower than static "onclick" event bindings on elements, because they override the default native (and fast!) methods implemented by browsers in "window" container which are effectively dispatching the UI events to the appropriate element, using their exposed "onXXX" properties ; however if the document belongs to another security domain, the container will not be able to perform this dispatch itself, and will transmit it, after translation to the document's coordinates and filtering of the event, to the document that will perform this dispatch : for such cases, the browser will instanciate a separate invisible window (with a null "parent" window) for its client area, where it loads cross-site documents, so in all cases this will always be a window and not a document or one of its elements that will perform all event dispatches using chaining).

I understand that you may have difficulties to understand the concepts, because it is very technical. But I really understand that addHandler() on an element will not work the way it is currently implemented (in the first branch of the if()), and that only the "else" branch will work reliably.

I have made another test, by modifying what you have patched recently in wikibits.js: I have dropped the first branch in the if, and then it works (on all browsers I have tested, i.e. more than 100 browser/system/version combinations, including embedded browsers in smartphones, or set top boxes, or a Philips LCD TV with Internet browser based on embedded-Opera). verdy_p (talk) 15:08, 13 October 2010 (UTC)

Status changers

Wikipedia:WikiProject_User_scripts/Scripts#Personal lists two status changers, and another is described at User:Goyston/Status Changer Installation. Do any of them actually work? Are they vector compatible? Are there others? I've had no luck. Rd232 ^talk 11:20, 12 October 2010 (UTC)

You could try user:xenocidic/statusChanger2.js. –xeno^talk 12:37, 12 October 2010 (UTC)

Thanks, that has the advantage of simplicity, and it actually works. What I liked about User:Topaz/statuschanger.js (one of the options at User:Goyston/Status Changer Installation) was that it allegedly "auto-updates when you log in and out." That seems like a massive advantage. But it wouldn't work for me; is anyone using it - perhaps I did something wrong? Rd232 ^talk 13:38, 12 October 2010 (UTC)

You actually log out? ;> –xeno^talk 13:40, 12 October 2010 (UTC)

Well no... :) Ideally there'd be some sort of activity measure, eg "change status to Out after X mins since last edit". Rd232 ^talk 14:03, 12 October 2010 (UTC)

There used to be a bot that would do this, but the developers blocked it because it was making altogether too many edits. A javascript that kept track of local activity would be cool, and maybe queried if you wanted the status changed when you closed the browser... –xeno^talk 14:05, 12 October 2010 (UTC)

I imagined something inbrowser... and if Javascript could reliably do that, great. Any volunteers? User:TheDJ/Qui does something with cookies. (Incidentally, couldn't MediaWiki quite easily be changed to display a userpage notice showing last edit date and total edits in last 30 days? Instant sense of user's activity level, without a bot or script.) Rd232 ^talk 15:53, 12 October 2010 (UTC)

See bugzilla:14384. –xeno^talk 15:55, 12 October 2010 (UTC)

Which concludes essentially with Extension:OnlineStatus. Should we propose installing it and/or developing it so it could be? Rd232 ^talk 16:01, 12 October 2010 (UTC)

I've not bothered with status display for some time, but I wouldn't object. –xeno^talk 16:04, 12 October 2010 (UTC)

The bot was blocked for auto updating status, and if you created an autoupdating JS that would do the same, I suspect that developers will eventually block that script as well. That's why I never added anything like that to Qui. Qui uses cookies to cache your status btw, preventing too many requests on the API servers and thus keeping load as low as possible. Basically I designed it to be as un obtrusive for the system administrators as possible. —TheDJ (talk • contribs) 18:06, 12 October 2010 (UTC)

The bot was making a ridiculous number of edits. An auto-updating JS could be smarter - not make as many edits, setting to 'away' after an hour, with a timestamp that can be smart and change itself to "offline" after 3 hours or something. –xeno^talk 18:17, 12 October 2010 (UTC)

Perhaps, but it used the same approach. (perhaps half an hour instead of 3 hours, but same idea) Also, how would you do an autoupdating JS ? Keep your browser open all day long ? Install a JS for all the users (some who might not care for this? Neither seems desirable to me. —TheDJ (talk • contribs) 19:11, 13 October 2010 (UTC)

Only those who opt-in, I guess. The javascript should be able to detect a browser-close action and ask if they want to update their status to offline, else, it would set to away after some time, etc. –xeno^talk 19:13, 13 October 2010 (UTC)

Can we toggle Show-A/Show-B (~show/hide)

Rare characters do not show well often in our browsers, not even when using {{unicode}}. For some templates, the characters can be shown by image. What we are looking for is a toggling switch, that does the altering between show-fonts <=> show-images.
This sandbox {{Unicode chart Counting Rod Numerals/sandbox}}. uses two show/hide buttons, to get some idea, but which is not as elegant. What we don't want is user-side requirements. Some earlier talk here. Possible? -DePiep (talk) 05:02, 13 October 2010 (UTC) +link -DePiep (talk) 05:10, 13 October 2010 (UTC)

Isn't there an extension that supports mutually exclusive tabs, with only one tab displayed at one time ? It seems that I have seen this used within portals, just like it is used in the User Preferences page. verdy_p (talk) 15:38, 13 October 2010 (UTC)

I have tried the following code, but this does not work:

<div id="mytabs" class="jsprefs-container">
  <ul id="preftoc" class="jsprefs-toc">
    <li class="selected">[[#prefsection-0|Text]]</li>
    <li class="">[[#prefsection-1|Image]]</li>
  </ul>
  <div id="preferences" class="jsprefs">
    <div id="prefsection-0" class="prefsection">
Display as Text here...
    </div>
    <div id="prefsection-1" class="prefsection" style="display:none">
Display as Images here...
    </div>
  </div>
</div>

• Text
• Image

Display as Text here...

Display as Images here...

The problem being that the "jsprefs" class works works with the User preferences, but only with specific ids (to associate the prefsections with items of the leading unordered list). Unfortunately, there's no class defined for the container "mytabs" and for the list of tabs in "preftoc", that would allow subtituting the ids. The javascript also expects to find a matching fieldset element within the "preferences" div, an element can't be inserted in wiki code. If this existing javascript was looking for classes, it would work as well to create selectable tabs in the wiki page (provided that each tab can be given a unique id. verdy_p (talk) 16:02, 13 October 2010 (UTC)

But you could do that with a gadget based on the existing javascript for "jsprefs"... The default display without the gadget will be much like:

• Text
• Image

Display as Text here...

Display as Images here...

Ideally, there should not need to indicate "display:none" within sections, and as well we could avoid specifying the class "selected" in the list, assuming that the first tab should be implicitly selected by default, and the associated section will be left shown, when the other sections will be set to "display:none". There should always not need to specify any style for bordering each section, as this should be inferred from class "tabs-section". We could than use any variable id as the prefix for the container, and could choose any id for linking list items to the appropriate section.

verdy_p (talk) 16:11, 13 October 2010 (UTC)

Userpage showing up at CAT:CSD

Kubek15's userpage is appearing at CAT:CSD despite no apparent tagging. I deleted a bunch of his user subpages at his request, but his main userpage is clearly not tagged; he's looked at it and is stumped as well. Is there something transcluding there that's putting it in a deletion category? Acroterion _(talk) 13:57, 13 October 2010 (UTC)

Fixed now, I think. A null edit was all that was needed. –xeno^talk 14:02, 13 October 2010 (UTC)

Yes, you beat me to it. It transcluded at least one of the subpages that were deleted. Something had probably not been updated yet by the job queue. A null edit skips the queue. Note that purging the page after changes to a transcluded page will only update which categories are displayed on the page and not the listings of the page in those categories. PrimeHunter (talk) 14:09, 13 October 2010 (UTC)

Thanks for the help. Acroterion _(talk) 14:49, 13 October 2010 (UTC)

Bot to fix accidental category or interlanguage links

It is a known problem on some discussion pages where someone will try to wikilink to a category or to a page in another language, and accidentally put the page into the category or create an incorrect interlanguage link instead. AnomieBOT is currently in trial for fixing this sort of error. If you know of a discussion page where this sort of error is relatively common, please add it to Category:Pages automatically checked for accidental language links to give the bot more work to do. If you have any comments, go to Wikipedia:Bots/Requests for approval/AnomieBOT 43. Thanks. Anomie⚔ 00:59, 14 October 2010 (UTC)

Edit conflict problem

I am concerned about recent edits on Axiom of choice, and in particular this obviously broken edit. My theory about what happened is as follows.

I edited "Nonconstructive aspects" (section 5 as the section edit link calls it, though it's section "3" in TOC) to remove the section header of its subsection "Banach–Tarski paradox" (section 6). Meanwhile, another user opened the "Banach–Tarski paradox" subsection and tried to edit it. He only saved it after me. At this point, the software should have raised an edit conflict. It apparently didn't, and instead it replaced the content of the next section "Independence" (section 7 before, but section 6 now) with the edited content of the former section 6 (now part of section 5).

If this interpretation is correct, it means that the automatic edit conflict resolution system erroneously accepted the edit, failing to realize that the section number under which the edit is being saved is bogus because the numbering has changed meanwhile. Moreover, it probably also failed to realize that the two edits changed the same section, which should be another reason for an edit conflict (indeed, the second edit incidentally restored the section header deleted in the first edit). Is this a plausible analysis of the situation, and if so, should we do something about it?—Emil J. 15:47, 13 October 2010 (UTC)

It appears that the edit-conflict checker should have known how to catch this. Since the person you conflicted with is not on Wikipedia very much, it seems likely that he got the e/c warning and just did the wrong thing by inadvertence. See mw:API:Edit for some details on what the conflict checker has to work with. It seems that the section number is passed in, but not the section title. Two people trying to edit the same numbered section should have produced a conflict warning. Possibly he overlooked the warning, and saved his version anyway. EdJohnston (talk) 16:36, 13 October 2010 (UTC)

It seems you are right. I tested it in a sandbox, and it worked as expected. Sorry for the confusion.—Emil J. 17:08, 13 October 2010 (UTC)

Except the edit conflict checker is off for logged in users, the regression seems to date back for at least a few months possibly since last year. — Dispenser 17:12, 13 October 2010 (UTC)

Dispenser, I don't understand your comment about the conflict checker. I just verified by editing my sandbox from two different windows that edit conflicts are being announced in the usual way. Also, my experiments show that a person who hits 'Save' enough times can ignore the advice from the conflict checker, which is what could have happened in Emil J.'s report that started this thread. EdJohnston (talk) 15:52, 14 October 2010 (UTC)

Is this a bug or a feature?

On my watchlist, I have Lights (edit | talk | history | protect | delete | links | watch | logs | views). An admin performed two actions, and it appears thusly:

• (Deletion log); 19:54 . . AnemoneProjectors (talk | contribs | block) restored "Lights" (36 revisions restored: just did history merge with Kiss You Off)
• (Deletion log); 19:54 . . AnemoneProjectors (talk | contribs | block) deleted "Lights" (G6: Deleted to make way for move)

You'll notice that the "deleted" link is a redlink. I don't think that's on purpose, am I right? You can add the page yourself to your watchlists to verify this. And yes I've purged. :) Magog the Ogre (talk) 00:17, 14 October 2010 (UTC)

I've always assumed that it was a feature, to make this sort of thing stand out and be easier to see on a busy watchlist. It might be a bug, sometimes it's hard to tell the difference round here. DuncanHill (talk) 00:23, 14 October 2010 (UTC)

When a page is deleted, Mediawiki asumes the pages no longer exists, so it uses a redlink for in the edit summary, and since edit summaries cannot be changed, it stays as a redlink. Just a technical limitation. — Edokter • Talk • 18:59, 14 October 2010 (UTC)

Single login at Meta - double login everywhere else

I have a unified account on the various WMF projects. If I go to meta:Special:UserLogin and log in, I can jump to any other WMF project like Commons, Wiktionary or Wikipedia and I will automatically be loged in.

If I log in at any other project, like commons:Special:UserLogin, wiktionary:Special:UserLogin or wikipedia:Special:UserLogin and then jump to another WMF project, I have to log in a second time. After the second login, I am automatically logged in to all other projects.

Why is the login system set up this way? -      Hydroxonium (talk) 12:39, 14 October 2010 (UTC)

Auto log out problem

Does the sentence in Wikipedia:FAQ/Technical#Hey.21_Why_was_I_automatically_logged_out.3F that says "Session information may occasionally be lost or go missing temporarily, causing users to be logged out" mean the problem is on Wikipedia's end, not mine? If yes, is the problem being worked on? TresÁrboles (talk) 22:08, 13 October 2010 (UTC)

That quote was from Tim Starling in 2005. The issue still exists today, so I'd assume it's not being worked on. ;-) However, the issue could be with your browser (make sure you're accepting cookies and not deleting them with each end of your browser's session). Other than that, I'd probably blame Wikipedia. Killiondude (talk) 07:12, 15 October 2010 (UTC)

Moving multiple pages

When moving a page, is there any way of moving all associated subpages as well in one go? This can be particuarly tedious work, especially for templates that use many subpages. PC78 (talk) 00:35, 15 October 2010 (UTC)

Administrators are able to do this (Wikipedia:Administrators/Tools#Other, bullet 2). Intelligentsium 00:54, 15 October 2010 (UTC)

See Wikipedia:Requested moves/Closing instructions#Non-admin closure.--Fuhghettaboutit (talk) 00:58, 15 October 2010 (UTC)

I figured that was probably the case. Is there any reason why the rank and file don't get to do this? PC78 (talk) 01:02, 15 October 2010 (UTC)

It was made available to all autoconfirmed users when introduced, but later restricted to admins, possibly because of vandalism moves. Keith D (talk) 01:19, 15 October 2010 (UTC)

The relevant discussion is here.--Fuhghettaboutit (talk) 02:43, 15 October 2010 (UTC)

With new technology, ceasing abuse of that feature by non-admins would probably be easier. However, I agree with Mr.Z-man's post in that (archived) thread in that the cases for appropriate use for this feature are so rare that I don't really see a point in broadening the ability to use it. Killiondude (talk) 06:24, 15 October 2010 (UTC)

It's not that rare. Talk pages can often have archives, templates will often have /doc, /sandbox and /testcases subpages. In fact it's potentially damaging for non-admins not to have this feature, because it's all too easy for the subpages to get left behind when doing a move. PC78 (talk) 17:41, 15 October 2010 (UTC)

One possibility would be to include the move-subpages right with a bundle of rights as suggested here. –xeno^talk 13:23, 15 October 2010 (UTC)

I'll check that discussion out. PC78 (talk) 17:41, 15 October 2010 (UTC)

String manipulation

Has there been a change somewhere to the string manipulation templates, or the displaytitle functionality, recently? The title italicizer embedded in {{Infobox ship begin}} (based on {{Italic title prefixed}}) seems to have stopped working for titles which include a space within the part requiring italicization, although it handled these all right before.--Kotniski (talk) 13:17, 15 October 2010 (UTC)

OK, I think I've tracked it down to Template:Str index/getchar.--Kotniski (talk) 13:58, 15 October 2010 (UTC)

Search box tooltip obscures top results in drop-down suggestions

Whenever I do a search in the Search box in the upper right corner of Wikipedia, I click into the box and usually leave my cursor there while I type a search term. This causes a tooltip to appear ("Search Wikipedia [alt-f]") that unfortunately blocks the top search suggestion from my view (in the drop-down that appears with the search suggestions). Since generally the top search suggestion is the one I most want to see, I find myself constantly annoyed by this.

Could we get the tooltip removed? I think its negative effects outweigh its benefits. Thanks.

(Using IE 8/Windows 7) —Preceding unsigned comment added by 173.8.89.178 (talk) 14:55, 15 October 2010 (UTC)

Not to be an ass or anything, but couldn't you just move your cursor? Removing the tooltip entirely removes the functionality (important for non-sighted users), so that's not really an option. You could also use the keyboard shortcut (alf-f) to immediate bring focus to the search box without using your cursor, which would likely eliminate the tooltip from showing up. EVula // talk // ☯ // 20:54, 15 October 2010 (UTC)

Gray redirect

Resolved

I might be asking this in the wrong place. I clicked on a link from here and at the top of the article it said "Redirected from" with no blue link.Vchimpanzee · talk · contributions · 19:53, 15 October 2010 (UTC)

Found it. It's like the link on the page where I clicked on it in the first place. It's too small to be seen easily.Vchimpanzee · talk · contributions · 21:20, 15 October 2010 (UTC)

Zoom in (Ctrl+NumPad+, Ctrl+NumPad0 to restore default zoom), you'll see that the grave accent is there, but some fonts may cause it to collide with the closing parenthese, or to touch it so that it's difficultto see the separation between the two characters, notably with small font sizes like in « `) ». But the blue link is there, active, it gets underlined when the mouse gets over it, the arrow cursor changes into a hand cursor, and it can be selected separately from the parenthese, and clicked even if it's just two-pixels wide. verdy_p (talk) 04:36, 16 October 2010 (UTC)

Bot/Script to sort references in alphabetical order

GA reviewers frequently ask that the references section be sorted in alphabetical order. An automated tool to do this for those who use the {{citation}} template and other methods will help greatly. Zuggernaut (talk) 03:43, 14 October 2010 (UTC)

Are you talking about the references created using <ref> tags as displayed to the reader, or are you talking about the wikitext source when using list-defined references, or are you talking about manual referencing systems? If the first, those GA reviewers should be made aware that the order is determined by the software and is roughly based on the order the <ref> tags appear in the page. Anomie⚔ 11:09, 14 October 2010 (UTC)

No, I'm not talking about the <ref> tags. I've been using the {{sfn}} template together with the {{Citation|last=ABC|first=XYZ|year=2010...}} combination. An example is in Third Anglo-Maratha War. I'm currently trying to get Deshastha Brahmin promoted to GA and I'm finding that arranging it in alphabetical order is manual labor and painful. I was asked in the GA review for Third Anglo-Maratha War to sort in alphabetical order so I'm trying to do the same for Deshastha Brahmin before I nominate it. Zuggernaut (talk) 21:35, 15 October 2010 (UTC)

Not directly responsive, I'd comment that when I have the freedom to do this I try to regularize such References sections with authors listed as Last, First, and with the authors as the initial item in the wikitext of the list entries. The sort-key (|last1=, if the cites are templated) then appears in a predictable early location for each list item, minimizing the difficulty of finding the appropriate insertion point for newly-added items. Wtmitchell (talk) (earlier Boracay Bill) 03:17, 16 October 2010 (UTC)

I have created a tool to do this: [1]. Few caveats:

1. Each ref has to start with a * (I had to fix Deshastha Brahmin because of this)
2. There can't be any text after the last ref, so you have to manually remove things like {{refend}} and then add them again to the result. (Text before first ref is okay.)
3. Sorting of non-ASCII symbols is probably going to be incorrect (for any definition of what the correct sorting is).

I haven't spent much time on this, so if anything gets wrong, it's going to output non-helpful error messages or just plain wrong results. Svick (talk) 22:33, 16 October 2010 (UTC) Svick (talk) 22:33, 16 October 2010 (UTC)

Great, this is going to be a lot of help and I can already foresee using it on several other articles I am nominating for a GA review. Very helpful. Thanks. Zuggernaut (talk) 06:16, 17 October 2010 (UTC)

hiding particular editnotices with css

If the div class is editnotice_Editnotices/Page/Wikipedia:Administrators%27_noticeboard, how would I hide this? –xeno^talk 17:30, 14 October 2010 (UTC)

Put

.editnotice_Editnotices/Page/Wikipedia:Administrators%27_noticeboard {display: none !important;}

in Special:MyPage/Skin.css. Algebraist 17:47, 14 October 2010 (UTC)

Thanks, but for some reason it's not working =\ –xeno^talk 17:55, 14 October 2010 (UTC)

Yes, you need to escape the /, : and %, and I can't work out how at present. Algebraist 18:02, 14 October 2010 (UTC)

Maybe it's not a valid class name? –xeno^talk 18:04, 14 October 2010 (UTC)

Here we are:

.editnotice_Editnotices\2F Page\2F Wikipedia\3A Administrators\25 27_noticeboard {display: none !important;}

Algebraist 18:07, 14 October 2010 (UTC)

Marvelous! Many thanks, –xeno^talk 18:11, 14 October 2010 (UTC)

Don't add wrapping divs when that is not needed however. I just gave those editnotices ids, that should work much better

• #admin_noticeboard_editnotice
• #village_pump_technical_editnotice
• #village_pump_proposals_editnotice

Those id's are added to the div of the editnotice itself. —TheDJ (talk • contribs) 18:30, 14 October 2010 (UTC)

No objections here. See also: Template:Editnotices/Page/Wikipedia:Administrators' noticeboard/Incidents. Thanks, –xeno^talk 18:35, 14 October 2010 (UTC)

• Funny that all these edit notices are being created only so that we can find ways to hide them. :-) Killiondude (talk) 03:22, 17 October 2010 (UTC)

Odd article found in watch list

I just cleaned out my watch list and found an vandaly odd article. Something about finding my daughter and sex. I suspect that it was maliciously placed there. Any idea if my account or the watch list could have been corrupted? Zulu Papa 5 * (talk) 04:38, 16 October 2010 (UTC)

For clarification it was in the listing, never in the watchlist with an contributions. Zulu Papa 5 * (talk) 04:40, 16 October 2010 (UTC)

It's page move vandalism. When someone moves a page to a new title, that new title is added to the watchlists of everyone who has watched the old title. When the vandalism is undone, the malicious title still remains in watchlists even though the actual page has been deleted. You can just remove that item from your watchlist. PleaseStand ^(talk) 04:50, 16 October 2010 (UTC)

Possibly some kind of malware running on your PC, that reuses your existing session to post spams to the wikis you are visiting. Be careful! This may cause you to be blocked if someone discovers that your account posts these things repeatedly.

Are you connected via a shared proxy (at school, in university, or using some internet accelerator ? If your ISP is installing a proxy without your knowledge, and does not respect the proxy standard for safely identifying their forwarded HTTP sessions, your proxy may be blocked. You should then speak with Wikimedia admins to resolve the issue with your ISP, or could have to change your ISP

Beware or mobile phone ISPs, and roaming accesses... most of them use proxies, first to reduce the network bandwidth by filtering images, or compressing images an scripts, or because of IPv4 adressing space depletion : another user using the same proxy as you may have been identified as if it was you.

Beware of cross-site scripting sessions : make sure your browser is up to date, and make sure that the adress bar shows the domain name "en.wiki.x.io" (to avoid navigation through an invisible frame). Avoid navigating and connecting to your account from framing sites.

Beware of some fake browser extensions that propose you helpers for editing on Wikipedia or other collaborative spaces including blogs and online forums (some of these extensions are malicious and take control of it to allow spammers to post things using your current identification). Some of these extensions are also hidden in fake security suites or fake PC cleaners: never use those sorts of "free" tools unless it is referenced and evaluated on a large download site such as download.com. When in doubt before using a new free tool, try Googling to see what others are thinking about it : most serious (non-fake) free tools are available on download.com, cnet.com, and other wellknown generic software and PC solutions sites, many of them associated to a magazine, or supported and proposed by ISPs in their portals. verdy_p (talk) 04:57, 16 October 2010 (UTC)

Nah, it's what PleaseStand said. There's no mystery here, and no need to panic anyone by talking about "malware". — Gavia immer (talk) 05:14, 16 October 2010 (UTC)

Thanks I suspected something like page move vandalism, but could not fully comprehended it, and wanted to be sure, my account wasn't compromised. Zulu Papa 5 * (talk) 05:27, 17 October 2010 (UTC)

Out and in

Why does Wikipedia sometimes log me out even when i am very active? Simply south (talk) 15:19, 16 October 2010 (UTC)

• Login expires after 30 days. — Edokter • Talk • 15:32, 16 October 2010 (UTC)
  • Tat's true if you say "keep me logged in" but i often log out and log in and i am wondering why it sometimes logs me out automatically or does the same reason apply? Simply south (talk) 16:18, 16 October 2010 (UTC)
    • You might see #Auto log out problem, above. Killiondude (talk) 03:20, 17 October 2010 (UTC)

Template:*MP

Can I can attention to this unanswered question by Admrboltz? Is this template still one which should be in use? SFB 16:09, 16 October 2010 (UTC)

Interestingly the template is transcluded 16396 times but in only 25 articles. And no, I don't think it is necessary. –droll [chat] 00:16, 17 October 2010 (UTC)

problem with mediawiki:newarticletext

I asked about this twice in #mediawiki and GreenReaper tried to help but it wasn't a real solution and he was gone by the time I responded to him.

* jeremyb points to [[Special:ExpandTemplates]]?contexttitle=*foo&input={{mediawiki:newarticletext}}

< jeremyb> see the raw wiki syntax in the preview rendering

< jeremyb> what's the best way to escape the leading '*' in the pagename?

< jeremyb> fwiw, it's not an academic exercise, this really is affecting a current redlink

GreenReaper suggested using &#43;

* jeremyb tried &#42; and it worked but it would need to be done automatically (by some parser function?)

Any ideas? there's a urlencode parser function, how about htmlencode? --Jeremyb (talk) 20:24, 16 October 2010 (UTC)

We do not have functions to do such escaping of wikicode atm. The problem here seems to be an extraneous linebreak added somewhere, It would be interesting to know, if that linebreak is added by Expandtemplates, or simply by {{PAGENAME}}. —TheDJ (talk • contribs) 10:48, 17 October 2010 (UTC)

It's T14974, which is caused by the "fix" for T2529. Basically, T2529 is about a template using wikitext table syntax wouldn't render the table correctly when not placed at the beginning of a line. They "fixed" that by forcing a newline at the start of any template-like construct whose output with "{|", "*", "#", ":", or ";". Unfortunately that breaks cases where you want a literal "*", "#", ":", or ";" at the start of a template's output, which is T14974.

Sadly, while there is an easy workaround for 529 (just put the template at the start of a line) and no workaround for 12974, it seems we're stuck with the current situation of 529 "fixed" and 12974 broken. Anomie⚔ 16:37, 17 October 2010 (UTC)

Display problem with a page

The Portal:Olympics page can be stretched way beyond the display width in some cases when I open it. I'm using IE. I.e.: {{Broken}} Time to return to wikibreak :) Contact me at my talk page if I'm not responding. ANGCHENRUI _WP:MSE♨ 12:20, 17 October 2010 (UTC)

image border in Chrome

I've found recently that when viewing Wikipedia pages in Google Chrome, some images have a narrow double border of red around them (inside the larger box that surrounds image+caption in Thumb mode). This doesn't happen for images from Commons, and it doesn't happen in Firefox (I can't bring myself to test in Internet Exploder). Could someone look into this and fix it? If you want an example, the top image (Chavez at Turiamo) at 2002 Venezuelan coup d'état attempt is affected. Rd232 ^talk 12:57, 17 October 2010 (UTC)

I do not see this problem in Google Chrome 6. Perhaps you have some gadget/javascript tool installed that does this ? I believe the alt text fixer script used to do something like that. —TheDJ (talk • contribs) 13:27, 17 October 2010 (UTC)

Aha! Indeed - the problem disappears when logged out, so it must be one of my scripts. And it turns out to be User:Anomie/linkclassifier.js (and/or the associated css). Never mind then. Rd232 ^talk 13:57, 17 October 2010 (UTC)

It's a feature of the script and its default CSS, the images with the double red border are non-free. Anomie⚔ 17:00, 17 October 2010 (UTC)

Help needed with navbox template code

I just created Template:EstcatDecadeType which works in conjunction with Template:DBDecadesInCentury. Can someone spot what I'm unable to see which makes the rendering of Category:Musical groups established in the 1840s faulty? __meco (talk) 13:33, 17 October 2010 (UTC)

Fixed. diff. —TheDJ (talk • contribs) 13:42, 17 October 2010 (UTC)

And even though I couinted and recounted those parentheses. Much appreciated anyway! __meco (talk) 14:05, 17 October 2010 (UTC)

Enlarge a dithered image without smoothing?

Is there a way to enlarge an image in an article, pixel-for-pixel, without having the image smoothed and smeared? I want to preserve the blocky appearance of this dithered 3-bit RGB image when it is enlarged.

Original size:

Wiki-enlarged, with auto-smoothing:	Enlarged 300% in GiMP, no smoothing:

For the wiki-enlarged image, all the white dots have become smaller and darker, while the black dots seem to have stayed about the same.

Is there some parameter to have the wiki-enlarged image look like the GiMP result? DMahalko (talk) 07:48, 18 October 2010 (UTC)

No, wikipedia is a website, that is automated with sensible defaults that are useful for 95% of the usecases of images. It is not desktop publishing :D If you need a bigger size, uploading a bigger size is the best method. —TheDJ (talk • contribs) 08:19, 18 October 2010 (UTC)

It is usually your browser doing the smoothing; it is not under Wikipedia's control. In IE, you can try disbaling "smart dithering" to see if that helps. But as TheDJ said, uploading a bigger version is the better method. — Edokter • Talk • 11:02, 18 October 2010 (UTC)

Ah, I see.. it is Firefox 3.6.10 that is making the enlarged image blurry. When I view this page using IE 7, the two images are dot-for-dot correct.

I was under the impression that the wiki software recompresses images down (or up) to whatever size is needed, like when showing a 2048x1536 original at 320x200 in an article. So, does the wiki software only recompress images that are smaller than the original? Or does it not recompress those either? DMahalko (talk) 14:02, 18 October 2010 (UTC)

It creates thumbnails, i.e. images that are smaller than the original. Upscaling is one by the browser and might therefore give varying results. --Morn (talk) 14:15, 18 October 2010 (UTC)

Technical problem with Citation/core template in bnwiki

Hi, I know this place is for technical problems of English Wikipedia, but I'm asking for help from Bengali Wikipedia. Since it's a small wiki, and don't have enough people to help out with technical issues.

Lets get to the point. We copied the {{Citation/core}} from enwiki and localized it. We also made some modification according to Bengali grammar and citation style. But it's showing a bare link of url at the end, (which you can see here on my test page), and I don't know why. I will be thankful if some one will help to solve this issue. If someone want to contact directly, find me on IRC. My nick is Tanvir. Thanks in advance. — Tanvir • 12:48, 18 October 2010 (UTC)

You are missing the printonly class from MediaWiki:Common.css. —TheDJ (talk • contribs) 13:06, 18 October 2010 (UTC)

I was about to ask a similar question for the users of Simple English Wikipedia. In baby steps, how do you resolve this DJ? The Rambling Man (talk) 13:08, 18 October 2010 (UTC)

Well I know that that template adds a (normally) hidden link for when you print the article. Then I used the WebKit Web Inspector to look at the source of the page in question (Safari/Chrome, rightclick, "Inspect element"). The link is wrapped in a <span> element with the class "printonly". That class is part of the wiki specific CSS in MediaWiki:Common.css (which I knew from experience). —TheDJ (talk • contribs) 13:25, 18 October 2010 (UTC)

Wow. Any chance (any chance at all??) that you could have a look at the Simple English version and "fix" it? The Rambling Man (talk) 13:30, 18 October 2010 (UTC)

For the exact same problem, add the snippet below to simple:MediaWiki:Common.css. —TheDJ (talk • contribs) 13:34, 18 October 2010 (UTC)

/* For linked citation numbers and document IDs, where
   the number need not be shown on a screen or a handheld,
   but should be included in the printed version
*/
@media screen, handheld {
    span.citation *.printonly {
        display: none;
    }
}

Fingers crossed, thanks. The Rambling Man (talk) 13:35, 18 October 2010 (UTC)

Hmm, that didn't seem to work. Perhaps it's something else... The Rambling Man (talk) 13:38, 18 October 2010 (UTC)

CSS en JS is cached by your browser, so you have to bypass your browsercache before you can see the result. I'm also not sure of course if the "simple" templates use the exact same classname. If you have a link to a page that shows the problem, then I could look further into it of course. —TheDJ (talk • contribs) 13:54, 18 October 2010 (UTC)

This shows the problem in the examples... Thanks for your time so far! The Rambling Man (talk) 14:02, 18 October 2010 (UTC)

If I bypass my browser cache, the problem no longer shows. —TheDJ (talk • contribs) 14:09, 18 October 2010 (UTC)

DJ, Thanks very much for your help. — Tanvir • 14:17, 18 October 2010 (UTC)

Ditto, thanks again. The Rambling Man (talk) 14:36, 18 October 2010 (UTC)

Google Chrome display problem?

I have a question about setting column width and height. It appears that tables that have a set column width do not always display properly (that is, they are wider then the should be) in Google Chrome. I noticed this while editing My Chemical Romance discography (this version). I compared the display from Chrome to Firefox and Internet Explorer and found that the other two browsers had the proper display. The problem is evident in the first two tables (studio albums and live albums), which should be the same size, but aren't. My screen resolution isn't a problem because its 1280 × 800. Does anyone else have this problem or any ideas about what might be wrong? – Zntrip 22:54, 11 October 2010 (UTC)

If you want display/rendering to be consistent between browsers, you need to be (more) explicit. In this edit, I specified the table width to 95% of the content area. Those tables should now render at 95% in any reasonable browser. If that isn't the case, then there's an issue. :-) --MZMcBride (talk) 03:59, 12 October 2010 (UTC)

I understand what you did, but it didn't help. In the first two tables the second columns should be the same width, but they still aren't. :/ – Zntrip 04:13, 12 October 2010 (UTC)

Then specify column widths? --MZMcBride (talk) 04:21, 12 October 2010 (UTC)

I'm sorry if you don't understand the problem, perhaps I haven't explained it correctly. Column width was specified.

Note that column widths are specified in pixels. You seem to assume that all texts will fit, and this is not the case, and jighly depends on the fonts installed. some of your columns are in fact too narrow for their contents. As the widths specified are just hints for browsers, if anything does not fit, it will reajustate these widths to make the content fit in the minimum column width (that the width attribute or the width CSS properties do NOT specify).

When I spoke about overconstrained tables, this effectively includes all these cases. More generally, column widths should never specify pixels (unless the column only contains images or objects with static size also specified in pixels). All other columns shouls specify widths in percentage, or in 'em' (this last option is only available with CSS property style="width:...;", not with the HTML table attribute width="...".)

verdy_p (talk) 04:13, 16 October 2010 (UTC)

My problem is that Google Chrome is not displaying the correct width (specifically with respect to the the second columns of the first two tables), while other browsers are. Your most recent edit did not fix the problem, it just completely changed the width of columns in the first table. – Zntrip 04:43, 12 October 2010 (UTC)

A screenshot might help. This is how Chrome looks for me: http://pruebita.com/chem-romance-chrome.png The second columns look perfectly aligned to me. --MZMcBride (talk) 06:58, 12 October 2010 (UTC)

It doesn't look like that for me, it might have something to do with the fact that I'm using Windows 7. This is what the permalink version looks like for me: http://www.flickr.com/photos/30124318@N06/5074732992/ – Zntrip 07:48, 12 October 2010 (UTC)

Avoid pixel widths for columns, unless the columns are displaying only images (use widths in "em" or "en" instead, in CSS only because the width="" attribute only accepts pixels) : for the small columns (by country) they may be extended according to internal cell padding and their content. columns in pixels cause accessibility problems depending on font sizes affectively used by browsers or reader preferences.

Also the second column has no width at all, and given that the table itself has no width, this second column will have variable width according to its content, so it cannot align with the second column in the next table.

Finally avoid constraining the width of all columns in a table, AND the width of the table simultaneously. If your table is overconstrained, the widths specified will only be considered for their relative proportional value, and all columns will be resized first to their minimum width, and the rest (if any) will be distributed proportionally to all columns (according to their initial specified width), and if there's no rest of the sum of minimum widths exceeds the specified table width, the table width will be increased (even if this causes the table to be larger than page width, causing horizontal scrolling in the window).

When a table is overconstrained, the result is unpredictable across browsers, or versions of the same browser, as they can only used heuristics.

verdy_p (talk) 16:49, 13 October 2010 (UTC)

First of all I'd would like to say that I'm happy to "see" you again Verdy p. The last time we "spoke" was back in 2006 (here). Thanks for the advice, however I'm not too sure how to apply it. I'm not too sure what you mean by using "em" and "en" instead of pixels. Could you perhaps provide an example? Also, the second column does have a specified width (you have to look at this version of the page). Also, I don't believe that the width of the entire table is constrained, just the individual columns. – Zntrip 19:43, 13 October 2010 (UTC)

What is the problem, precisely ? I don't see any sizing problem with Google Chrome. If you mean that separate tables should align their columns, there is absolutely no way to do it, unless you specify a fixed width (using BOTH "min-width:" and "max-width:" CSS2 properties ; the CSS1 "width:" property, or the table attributes width="" only specify an hint, that browsers can increase if any content does not fit for its computed minimum width, but also can reduce if other columns need more space and the table itself has already reached its margin limits, the last solution being to increase the minimum width of the table itself has no maximum width specified (also causing any container of the table to have its minimum width also increased, if it is not itself constrained by a maximum width).

Computing element widths in HTML is a very complex algorithm in fact. So it would help if you wanted to explain what you expect. I don't see any problem of rendering with Google Chrome, the specified widths are working as indicated, just as hints. verdy_p (talk) 04:25, 16 October 2010 (UTC)

What I am trying to achieve is uniform column width across all the tables in the article. You said this could be achieved using the "min-width:" and "max-width:" CSS2 properties. How exactly do I do this; I'm not sure what the notation is with wikitables? Also, for an idea of what my display looks like see this link. My OS is Windows 7 and my screen resolution is 1280 × 800. – Zntrip 07:58, 16 October 2010 (UTC)

There, did it. I set the width to 100% for both tables. If you want it to be a certain distance from the right or left, just set a margin. For a certain percentage from the right, just set the width to a lesser percentage, and the browser will split the difference. Arlen22 (talk) 01:19, 20 October 2010 (UTC)

Wow, thanks a lot! :D – Zntrip 02:30, 20 October 2010 (UTC)

"What links here" problem

Hi, I am trying to look through the "What links here" pages for Republic of Ireland. Out of the first 50, 1 was a talk page, 15 linked to Republic of Ireland as part of the article, and none of the rest had a link to Republic of Ireland, only that they happened to have a template such as world countries or {{English official language clickable map}}. The Republic of Ireland article is linked to by many 100s if not thousands of articles so obviously such a high rate of useless "What links here" returns makes for extreme difficulty, and obviously this problem will apply to many 1000s of articles, not just Republic of Ireland. It seems that every country and significant city or region, which is probably in the thousands, will return as linked when in fact the true number may be less than a percentage. That goes for languages and who knows what other topics. So if there is a solution please tell or at least someone interested in making a work around might see. ~ R.T.G 18:29, 16 October 2010 (UTC)

Someone didn't understand this problem too well so I tried using the example of Football. Some templates are on many hundreds of articles, especially templates of countries of the world, languages, sports... etc. If the article Football is on a template, and the template is used on 1,000 articles, when you look for "What links here" for Football you will have an extra 1,000 articles on that list which may be about hockey or skydiving just that they are sport. So they are nothing to do with Football and don't even have a link except some sport template which, once you look at the template once, you know what the link looks like and how it is used. Looking at it 1000 times will not reveal anything new from the first time. To highlight this, the first 15 on the What links here/Football page, excluding talk pages, include: 2 dates, August 22 and April 11, and 7 articles which do not have a link to Football in the article text, Australian rules football, Baseball, Basketball, Bowls, Canadian football and Curling. Obviously the link is found in a template. So, that is around 50% false positives. Excessive. The page lists 3954 "what links here" under "(Article)" space. It's not hard to imagine that 1000-2000 of those are incorrect. That's a lot of clicks and article searches for nothing. ~ R.T.G 19:22, 16 October 2010 (UTC)

If you look through the wikitext you can see that e.g. the line with Stjepan Bobek has a link to football in the Aug. 22 article. So there are no false positives there; all those pages really have links to the football article (sometimes from templates of course).

But why you don't use e.g. portals or categories to find related content? That's much easier than with "what links here". --Morn (talk) 19:40, 16 October 2010 (UTC)

I posted some links at Wikipedia:Help desk/Archives/2009 September 2#What Links here - filter problem? PrimeHunter (talk) 20:44, 16 October 2010 (UTC)

Look Morn, I went through every one of them. If you found something I missed well done. Otherwise, check them. Press ctrl+f and look for "Football". Nada. Nothing. Look properly. ~ R.T.G 11:10, 17 October 2010 (UTC)

The whole point of my post was that you haven't looked properly. Go to August 22, search for "Bobek", and there's the football link! So I'm not sure what you mean by false positives. There are none. --Morn (talk) 11:15, 17 October 2010 (UTC)

The dates were not the false positives. The other ones are. There is no link in the article text. Try any of the other six. The link is in a sport template and this particlar template, for instance, links to 200 articles. I am pretty sure that more than half those 200, such as the six above, do not have Football in the article text. When it comes to articles about countries, there are many more templates and they link to many more articles of which most do not even have relation to each other. Do you really have no inkling or are you just berating me about adding the dates? ~ R.T.G 18:24, 17 October 2010 (UTC)

The other pages have links in the templates at the bottom of the page. So what's your point? That links in templates shouldn't count? Well, PrimeHunter and others have suggested that already and apparently so far no one has been interested enough in that feature to implement it in MediaWiki. What you can do is vote for bugzilla:3241 and bugzilla:17728 and hope a developer will take it up. --Morn (talk) 18:38, 17 October 2010 (UTC)

RTG "just won't use MW bugzilla".[2] OK, but that limits your ability to influence some things. PrimeHunter (talk) 01:05, 18 October 2010 (UTC)

(cynical laugh) All it really limits is your ability to naively hope that you might influence some things. A brief period of activity at bugzilla was enough to convince me that it has absolutely no effect on anything - the most you might get is a huffy response from a developer explaining why you have no right to want whatever feature it is you're suggesting.--Kotniski (talk) 05:37, 18 October 2010 (UTC)

That is because bugzilla is a system to track bugs and feature request, to discuss their potential impact, benefits etc. It is not a demand/ransom list nor is it a timetable. —TheDJ (talk • contribs) 13:18, 18 October 2010 (UTC)

Of course, but as has been said here many times by many different people, it's an extremely ineffective forum for discussion (assuming that the discussion is not intended to be purely academic, but actually to lead to improvements being made to the project(s) through the software).--Kotniski (talk) 13:33, 18 October 2010 (UTC)

Depends. I would say that it is as effective as having a red link in a Wikipedia article in order to get a new article written on a subject. It works eventually, but it will take a long time. The most efficient method by definition is always to hire and manage your own developers. —TheDJ (talk • contribs) 13:41, 18 October 2010 (UTC)

To get back to the technical problem at hand, maybe Mediawiki could create a list of all links in the article before template substitution and use that list later when adding new backlinks, setting a flag in PAGELINKS (e.g., "PL_IS_NOT_FROM_TEMPLATE"). The problem I see is that updating all links each time a page is saved would be an unacceptable performance hit, but that doing it incrementally might leave many links incorrectly flagged. Good performance and high accuracy might be difficult to achieve.

But even without template links, the WLH list might still be unmanageably large. E.g., a lot of player bios probably have a non-template link to football/Association football/American football, etc. So the question is whether ignoring template links will gain you much. Even without templates, Wikipedia articles are heavily linked. --Morn (talk) 13:11, 18 October 2010 (UTC)

Maybe it would be simple to add an "Ignore templates" button or "ignore transcluded templates" leaving the template(s) itself to show up in the result. Note: It's not a problem that the list is long but it is a problem that there's a load of wrong answers and if I need to check how the Football link is being used, Association football and American football links are of no use. It is not the sport, the ball or the foot I am looking at... it's the links between Wikipedia articles. I have a lot of my task, checking the use of pipe-linking, done now but if I lost a quick 3 seconds 1000 times, there's the making of an hour gone so I wager that, on this one little repitition and accounting for all editors, some weeks and months of valuable wiki time has been been lost. There are only 8,500 hours in a year... what are the odds? How many FA could be wrote in a solid year of work. ~ R.T.G 13:49, 18 October 2010 (UTC)

I think the solution you are looking for is a semi-automated bot. MediaWiki itself should mainly concern itself with features a lot people use every day, not making maintenance projects easier that are finished after a few weeks or so. --Morn (talk) 05:50, 19 October 2010 (UTC)

A script that would search through the links and filter out the template ones. Ah yes that would fix the whole thing alright. Much better than a script that would search through without filtering them. I think I read something like this before somewhere :) ~ R.T.G 19:31, 19 October 2010 (UTC)

signature toolbar button - cursor placement

This is a really pokey complaint, but I've noticed that (recently, in vector) clicking on the signature button in the toolbar adds a signature (--~~~~) but places the cursor at the beginning of the sig rather than at the end. was it always this way, and is there a reason for it? I find it annoying, but I can deal if it was done for a reason. --Ludwigs2 05:32, 17 October 2010 (UTC)

bugzilla:25460 and revision 74613. —TheDJ (talk • contribs) 10:42, 17 October 2010 (UTC)

Ah, ok. and I assume that the reason this isn't yet happening is that that revision is not yet implemented? not a complaint, I'm just curious: I don't know how implementation works on wikipedia. I assume there's some sort of periodic updates where revised code is brought online, but I don't know if that's an automatic thing, or a haphazard volunteer "ah, gee, better update the code" sort of thing, or a collective "let's talk about it for a while until we're all sure" sort of thing. or maybe a combination of all three? wikipedia is interesting that way... --Ludwigs2 19:29, 17 October 2010 (UTC)

This is part of the usability initiative (employed developers part of the ) code. It usually gets deployed about every 3 weeks. The volunteer code hasn't been deployed since the last mediawiki upgrade to 1.16, which was somewhere just before august. 1.17 will probably take a few more months before it will be deployed, but critical issues (security, things that cause data loss etc are handpicked at prioritized). This is not a critical issue, but should have higher priority because it is code from the paid developers. —TheDJ (talk • contribs) 13:49, 18 October 2010 (UTC)

Ah, I see. thank you for the explanation. --Ludwigs2 17:16, 19 October 2010 (UTC)

Don't Forget the Lyrics!

I was fixing the theme song in the infobox but it won't display and I don't know what's wrong.Vchimpanzee · talk · contributions · 20:23, 18 October 2010 (UTC)

According to the documentation of {{Infobox television}}, the parameter for opening theme is opentheme. Changing it to that worked. Svick (talk) 21:04, 18 October 2010 (UTC)

Thnaks. That's what I get for not clicking on edit when I checked another show.Vchimpanzee · talk · contributions · 16:33, 19 October 2010 (UTC)

No Bugzilla you cannot have my email address to display in public

Self explanitory. Bugzilla site is not safe. It shares your personal information by public display. Creating new email accounts does not solve this problem... it only creates the extra problem of checking more and more email accounts. New email accounts can be spammed just as much as old ones. Spam attacks can be ridiculous rendering any email account useless for months. I have personally experinced it. Page after page of crap, all from individual dodgy websites. Bugzilla is, in this, a bug in itself. What is the point? Does the most basic computer course not warn you about sharing your email info in public? Can I not be contacted on Wikipedia by email without displaying the address to the public? Probably a complicated issue to fix but is it not significant? Does the idea, "Give them an email anyway," fix the problem? ~ R.T.G 08:41, 18 October 2010 (UTC)

I always use a Google Mail address for mailing lists and other things where you will get a lot of e-mails per day, both wanted and unwanted. And guess what, their spam filter works very well! So spam is a poor excuse for not using Bugzilla IMHO.

Still, feature enhancements like this--if they get implemented at all--take some time. E.g., a long time ago I had submitted requests for an HTML print view without links and PDF export of articles. We did get those features eventually, but it took many, many months, if not years and it was probably not even because of my requests that they were implemented.

In all seriousness, buying a book on PHP, installing Mediawiki on your own PC and figuring out how to improve WLH yourself might actually be a faster option. Then again, getting your Mediawiki patch accepted on Wikimedia projects would probably also take some time. --Morn (talk) 09:37, 18 October 2010 (UTC)

I also point out that since a few upgrades ago, email addresses are not visible to people who are not logged into bugzilla. Of course that doesn't prevent spambots from registering, but it's better than it used to be. Another suggestion is to write a bugzilla identification plugin that can work with our CentralAuth module. Otherwise you will have to wait till we finally have openID identities, and then bugzilla with wikipedia openid identities could be used I guess. —TheDJ (talk • contribs) 10:16, 18 October 2010 (UTC)

I have to say that it is rather hypocritical for us on Wikipedia to routinely blank email addresses from requests for help, and then force people to post them for requests for help on Bugzilla. DuncanHill (talk) 10:43, 18 October 2010 (UTC)

I don't think that preventing spambots or google being tough justifies displaying *my* email in public. I won't be displaying it like that one way or the other. Can anyone tell me why? Is there some need to display an email address? Why exactly do Bugzilla want my email address to be publically displayed for all time? ~ R.T.G 13:59, 18 October 2010 (UTC)

Bugzilla is a third-party open source platform. We use it to manage our bugs, but we didn't design it. The designers were aiming for software projects in general, and in most cases such projects don't come with huge user databases. The assumption is that, in the general case, the main way that developers communicate with each other and with the public is through email. In other words, it is meant to help facilitate those email communications, and not to replace them. In addition, Bugzilla comments are generally forwarded to and logged on publicly accessible mailing lists. (In other words, even if the email were obscured on the Bugzilla portal itself, the current approach also exposes it in the log.) Perhaps it would be better if the Bugzilla paradigm wasn't built around email, but that's how it has always been. Dragons flight (talk) 19:24, 19 October 2010 (UTC)

You could register with a throwaway account. The only emails bugzilla sends out are updates to bugs and such. –xeno^talk 14:02, 18 October 2010 (UTC)

That wouldn't appeal to me as much as, "Privacy policy: We do not share your personal information with anyone." It's a rule of thumb, isn't it? With someones email address you can look them up on a lot of other sites, possibly revealing much more personal information about them... ~ R.T.G 16:59, 18 October 2010 (UTC)

Register a throwaway account and have it forward your email to your main account (minus the spam--which gmail is good about getting rid of, as mentioned above). It's not difficult and takes like 3 minutes to do. Killiondude (talk) 17:02, 19 October 2010 (UTC)

If I register a throwaway account will Bugzilla stop displaying peoples private information in public? ~ R.T.G 19:33, 19 October 2010 (UTC)

You seem to be quite paranoid. Heaven forbid they show an account to logged in members that you register for the specific purpose of using bugzilla. Killiondude (talk) 19:36, 19 October 2010 (UTC)

Killiondude, I'm logged in here, please display your email address to me and all other logged-in users. DuncanHill (talk) 19:45, 19 October 2010 (UTC)

bugzillaacountforkillion@gmail.com. Killiondude (talk) 19:56, 19 October 2010 (UTC)

How many accounts do you have? One for every website? Or do you find that any ethically run website will not display your email account publicly? DuncanHill (talk) 20:47, 19 October 2010 (UTC)

Just make it MySSNumber@gmail.com. Ok, not really. Actually, that is a bad idea, of course, but name it something like bugzilla5569345@example.com. That way the meaningful names aren't used for useless stuff. Arlen22 (talk) 21:57, 19 October 2010 (UTC)

Okay but let's say I stopped at the, "Actually, that's a bad idea." part and thought, "I gave up those!" then didn't bother displaying my email in public. How many bugs that I found would be reported to Bugzilla? Not saying that I have a million bugs to report but must say that I didn't report this one, for instance... ~ R.T.G 16:22, 20 October 2010 (UTC)

You could always just report them to VPT. –xeno^talk 16:25, 20 October 2010 (UTC)

bits.esams.wikimedia.org is down

bits.wikimedia.org resolves to bits.esams.wikimedia.org in my area, which does not reply to TCP connection requests since yesterday. Pointing my hosts file to bits.pmtpa.wikimedia.org solved the problem.

bits.pmtpa.wikimedia.org: 403 error, good

bits.esams.wikimedia.org: no connection allowed, bad --Bootchy9 (talk) 09:15, 18 October 2010 (UTC)

A huge CPU load on esams seems to have been the culprit behind yesterday's outage; the second time this has happened within a week or so. So it might be that the admins have taken it down temporarily to investigate the issue. --Morn (talk) 09:43, 18 October 2010 (UTC)

I'd say there's a memory leak somewhere. The CPU spike coincides with all the RAM being used up. Happened twice before, see the monthly log. The US-based bits caches show the same memory consumption pattern,[3] though so far apparently without outage. Lupo 12:22, 18 October 2010 (UTC)

Indeed, it looks like a memory leak. But it only becomes a problem for serving files when the systems finally land in swap hell (with the almost non-existent swap they appear to have). Your second plot suggests something has changed in the configuration 3 or 4 weeks ago.

These outages are a bit embarrassing for the 6th (or 7th) most popular site on the web. Then again, by Twitter standards it's completely normal I guess. :-) --Morn (talk) 13:25, 18 October 2010 (UTC)

(e/c caused this to get lost earlier) It is under investigation now. They (WMF) found the same graph yesterday. —TheDJ (talk • contribs) 13:07, 18 October 2010 (UTC)

This problem in varnish 2.1.3 might fit the bill. RAM gets used up, servers start swapping, and since the bits machines seem to have next to no swap space, they thrash or stall. Lupo 09:53, 20 October 2010 (UTC)

I notice that—at least judging by the plots—the US bits servers did not seem to swap at all during their recent high-memory use crises and recovered, whereas esams did (and died as a result). So why not disable swap memory on esams entirely? I know some older Linux kernels supposedly ran into trouble without swap, but that has been fixed.[4] --Morn (talk) 11:18, 20 October 2010 (UTC)

Excess white space

The problem is on List of tallest buildings in Charlotte. I don't want to just take out the photos but there must be a better method of putting them in the article.Vchimpanzee · talk · contributions · 16:38, 19 October 2010 (UTC)

What? Killiondude (talk) 16:50, 19 October 2010 (UTC)

I only get a lot of excess white space on that article if I compress my browser window to be ludicrously narrow. Looks fine to me otherwise. EVula // talk // ☯ // 17:05, 19 October 2010 (UTC)

I haven't done anything strange, but if I scroll down there's almost an entire screen of white.Vchimpanzee · talk · contributions · 18:53, 19 October 2010 (UTC)

Does this look better? Phil Bridger (talk) 12:31, 20 October 2010 (UTC)

No, I still get an entire screen that is white except for the photos on the right.Vchimpanzee · talk · contributions · 16:57, 20 October 2010 (UTC)

I'm not sure what I did, but it looks better now.Vchimpanzee · talk · contributions · 20:44, 20 October 2010 (UTC)

Help needed enabling task force assessment stats

At Wikipedia:WikiProject Sociology/Social movements task force. Here I was told that I need to create categories listed here; I can do it - but I don't know what their parent categories should be. Could somebody tell me that, or create the first few so I can look at an example? --_{Piotr Konieczny aka Prokonsul Piotrus| talk} 16:52, 19 October 2010 (UTC)

Their parent will be the last one on that list: Category:Social movements task force articles by quality. If you press the "create" link for each category, the preloaded code should help you out. — Martin (MSGJ · talk) 22:06, 19 October 2010 (UTC)

The "create" template is pretty useless, I still had to create a parent category (I hope Category:Social movements task force articles by quality is correct) and categorize it manually (to Category:Sociology articles by quality). Is this what I need to do for all categories? I have to say, I am pretty wiki-savvy, but this process is arcane; somebody should really make it more editor-friendly. --_{Piotr Konieczny aka Prokonsul Piotrus| talk} 00:01, 20 October 2010 (UTC)

It's really not that difficult! If you press the "create" link and then follow the instructions at the top of the screen, everything should work fine. In your case you will need to set |project=Sociology and |topic=social movements task force and then save the page. (The result is like Category:GA-Class social movements task force articles which I have created as an example for you.) — Martin (MSGJ · talk) 08:08, 20 October 2010 (UTC)

Ok, I see now, but the instructions ARE confusing; I followed them as well as I could and did not create a proper page. What works (easily) is if I copy and modify the following string: {{cat more|Wikipedia:WikiProject Sociology}} {{cat class | topic=social movements task force | class=GA }} {{CategoryTOC}} I suggest that the "how-to" instructions are revisited and modified. --_{Piotr Konieczny aka Prokonsul Piotrus| talk} 15:56, 20 October 2010 (UTC)

Simple guide to MediaWiki delta compression?

I assume that MediaWiki/English Wikipedia makes use of Delta compression, though the documentation I've seen for it seems to be at a knowledge level that's beyond me (e.g. wgCompressRevisions). Is there a simple guide to how delta compression is used on Wikipedia and some examples of the space savings it makes? Thanks Rjwilmsi 18:34, 19 October 2010 (UTC)

Probably better ask this on the mediawiki or wikitech mailinglist. —TheDJ (talk • contribs) 12:12, 20 October 2010 (UTC)

I thought never used delta compression, just zipped the individual revisions. Though I could be wrong. — Dispenser 16:03, 20 October 2010 (UTC)

Yes, my understanding is that MediaWiki saves the entire raw wikitext of each revision and generates the diffs dynamically. — Gavia immer (talk) 17:06, 20 October 2010 (UTC)

See mw:Manual:CompressOld.php. Old revisions are concatenated into chunks of related revisions and then each chunk is gzip compressed. By grouping related revisions, it allows gzip to efficiently compress the repeated information. Hence this has a similar effect to a delta representation, but that's not actually the approach we take at present, as I understand it. However, I'm also suspicious that the Mediawiki page may not actually cover everything we do for compression since there have been multiple efforts at developing efficient representations. Dragons flight (talk) 17:35, 20 October 2010 (UTC)

Category intersect

What would be a sample SQL query for the tool server to do a category intersect of two specific categories on a specific namespace?Smallman12q (talk) 01:16, 20 October 2010 (UTC)

First, I have to ask, is there a reason why you don't want to use CatScan?

But anyway, let's say you want to list all articles that are both in Category:1988 births and Category:All articles that need to be wikified:

mysql> select page.page_title from page join categorylinks as cl1 on page.page_id = cl1.cl_from join categorylinks as cl2 on page.page_id = cl2.cl_from where cl1.cl_to = '1988_births' and cl2.cl_to = 'All_articles_that_need_to_be_wikified' and page.page_namespace = 0;

+-------------------------+
| page_title              |
+-------------------------+
| Katie_O'Donnell         |
| Marcin_Gawron           |
| Sebeca_Zahra_Hussain    |
| Kingi_Akau'ola          |
| Mathieu_Baudry          |
| Louis_Royal             |
| Stephen_Lambdin         |
| Alysha_Brillinger       |
| Chris_Molitor           |
| Jennison_Myrie-Williams |
| Tahis                   |
| Alexander_Turnquist     |
| Princess_Hijab          |
| Dorothea_Gundtoft       |
| Chantel_Jones           |
+-------------------------+

Note that underscore in the category names in the query are important. Svick (talk) 21:02, 20 October 2010 (UTC)

I'm rewriting WolterBot's replacement so it can run nightly on the toolserver. It essentially takes a WikiProject category and intersects it with another few hundred categories. Thanks for the example! Documentation and examples seem scarce.Smallman12q (talk) 22:19, 20 October 2010 (UTC)

Proposed top icons script

Moved from Wikipedia:WikiProject_User_scripts/Requests Per the suggestion of User:Roux, I have copy-pasted the following discussion from the scripts requests:

Feasible? I was thinking about the {{Top icon}}s {{Featured article}}/{{Featured list}}/{{Featured portal}} and {{Good article}} and I thought it might be handy to have a script that does the same thing for all of the other assessment classes (possibly excluding namespace classes like Portal, Book, or Category as being redundant.) The way it would work--as someone who does not know enough scripting to save his life--is that it would "read" the talk page of an article and find the WikiProject banner(s) that have the Class= parameter and then assign that rating to the article with a small icon at the top right to alert editors of what--if any--class has been assigned to the article. Further ideas that could be useful:

• If the rating is Stub, then users can click on the top icon and be directed to the appropriate WikiProject for more information on how to expand it (in the case of multiple banners, just choose the first one.) Alternately, go to the stub category and users can be directed from there on which WikiProject has adopted that set of articles.
• In the case of Unknown class or an empty class, then users can click on the top icon and be directed to http://en.wiki.x.io/wiki/Talk:PAGENAME&action=edit&section=0 to edit the banner(s) and easily apply a rating. I know that this would encourage me to rate more articles, especially stubs.
• If someone is feeling really adventurous, you can try to incorporate the Importance as well, somehow: through a background color to the icon, a tooltip, etc.

Below, I have displayed the class icons from {{Class/icon}}; see a full list (without all of the nice buttons) at {{Cat class}} and the color definitions at {{Class/colour}}, in case that is necessary.

• Featured Article: {{Classicon|FA}}
• Featured List: {{Classicon|FL}}
• A-Class Article: {{Classicon|A}}
• Good Article: {{Classicon|GA}}
• B-Class Article: {{Classicon|B}}
• C-Class Article: {{Classicon|C}}
• Start-Class Article: {{Classicon|Start}}
• Stub-Class Article: {{Classicon|Stub}}
• List-Class Article: {{Classicon|List}}
• Non-Article Page: {{Classicon|NA}}

• Unknown-Class Article: {{Classicon|No}}
• Book Page: {{Classicon|Book}}
• Category Page: {{Classicon|Category}}
• Disambiguation Page: {{Classicon|Disambig}}
• Media File Page: {{Classicon|Image}}
• Portal Page: {{Classicon|Portal}}
• Project Page: {{Classicon|Project}}
• Redirect Page: {{Classicon|Redirect}}
• Template Page: {{Classicon|Template}}
• Unknown: {{Classicon|}}

Note that Bplus is only used by two projects; I suppose that they can just get "B" ratings...

Other classes include: Current, Future, Merge, Needed, Deferred (???), Sl (???), and User.

Does this seem like a feasible or useful feature to anyone else? Does someone want to adopt this project and make a script that provides this functionality? Thanks. —Justin (koavf)❤T☮C☺M☯ 03:37, 20 October 2010 (UTC)

Addendum I primarily use Firefox to edit. —Justin (koavf)❤T☮C☺M☯ 03:40, 20 October 2010 (UTC)

This is not a bad idea at all. Though given how many pages it would affect, might be better to propose at the VP. Also will need a bot to do the needful. → ROUX ₪ 03:41, 20 October 2010 (UTC)

Clarify Just in case we're miscommunicating here, I am not proposing that a bot go add {{A-class article}} to every A-class article, rather these icons will only appear for users who are logged in an have this added to monobook.js/vector.js (which seems like an altogether small and reasonable server load, unless I'm missing something here.) —Justin (koavf)❤T☮C☺M☯ 03:58, 20 October 2010 (UTC)

---

Thoughts? —Justin (koavf)❤T☮C☺M☯ 06:04, 20 October 2010 (UTC)

Doesn't the existing Gadget "Display an assessment of an article's quality as part of the page header for each article" basically cover this, but without pretty icons? I find it very useful. Rd232 ^talk 10:50, 20 October 2010 (UTC)

My thought exactly. —TheDJ (talk • contribs) 12:10, 20 October 2010 (UTC)

! Oh wow; that is very similar. I feel foolish. I'll contact the creator of that script/gadget to see if he can help me modify it in the ways that I have suggested above. Yeesh. Thanks. —Justin (koavf)❤T☮C☺M☯ 17:24, 20 October 2010 (UTC)

Regex to remove all refs

What regex would I use to remove all the refs on a page? Arlen22 (talk) 17:38, 20 October 2010 (UTC)

Assuming no ref tags are nested, you could match a ref tag and its contents with /<ref>.*<\/ref>/. However, there is no regular expression that, in general, can correctly match balanced tags (or balanced parentheses, which is the usual example). — Carl (CBM · talk) 17:41, 20 October 2010 (UTC)

(e/c) It's not that easy. Usually, pattern matching is greedy, hence your regexp will also match "<ref>ref1</ref> unrelated text <ref>ref2</ref>". Second, the pattern will not match named references, and ref tags with atributes in general.—Emil J. 17:50, 20 October 2010 (UTC)

Assuming names do not contain > or /, this might work: /<ref( [^>/]*)?(\/>|>([^<]|<([^/]|\/([^r]|r([^e]|e([^f]|f[^>]))))*<\/ref>)/. In Perl, this could also work: /<ref( [^>/]*)?(\/>|>.*?<\/ref>)/. Both untested.—Emil J. 18:07, 20 October 2010 (UTC)

That's unlikely to be a big problem. Here is how a footnote in a footnote is displayed:

Test.<ref>Footnote 1<ref>Footnote 2</ref></ref>

---

Hans Adler 17:48, 20 October 2010 (UTC)

Nesting isn't an issue for <ref> tags, as XML-style extension tags cannot be nested in Mediawiki wikitext. In Perl or another language with regexes supporting lazy quantification you could use a substitute like s/<ref(?:\s+[^>]*[^\/>])?(?:\/>|>.*?(<\/ref>|$))//oig (which is rather well tested). References can be nested using the {{#tag:ref}} construct, which cannot be easily removed with a regular expression. Anomie⚔ 22:18, 20 October 2010 (UTC)

This is the AWB one (only works for C#): (<\s*ref\s+(?:name|group)\s*=\s*[^<>]*?/\s*>|<\s*ref\b[^<>]*?>(?>.(?<!<\s*ref\b[^>/]*?>|<\s*/\s*ref\s*>)|<\s*ref\b[^>/]*?>(?<DEPTH>)|<\s*/\s*ref\s*>(?<-DEPTH>))*(?(DEPTH)(?!))<\s*/\s*ref\s*>) Rjwilmsi 21:29, 21 October 2010 (UTC)

Cross Linking Wiki's

Is there any way to show the all pictures from one category of commons to any wikipedia (Automatically synchronized while commons category gets added with more pictures)

--Kalarickan | My Interactions 18:58, 21 October 2010 (UTC)

Script for fine-grain control of watchlist?

Is there a coder available to write a script(or a script already written) that would allow you to eliminate items from your watchlist with fine grain control? i.e. edits by user X marked minor, edits to Talk:Z (but not edits to "Z") etc. ? (the latter is already available in Wikipedia:Hide Pages in Watchlist, but figured it would be a good fit). –xeno^talk 00:04, 19 October 2010 (UTC)

Anyone?...Anyone?...Bueller? –xeno^talk 13:49, 22 October 2010 (UTC)

If you already load jQuery (through vector, new toolbar, live-preview) then it's trivial:

if (wgCanonicalNamespace=="Special" && wgCanonicalSpecialPageName=="Watchlist") addOnloadHook(function(){
  jQuery("ul.special > li:has(abbr.minor):has(.mw-userlink[title=User:Xeno])").hide();
});

That hides all minor edits that are by you. By script, so you'll probably see it for a bit, which might be annoying. Amalthea 17:43, 22 October 2010 (UTC)

Thanks! I'll have to figure out how to load jquery =] –xeno^talk 18:02, 22 October 2010 (UTC)

Reference

Why is the same reference list appearing for two different pages at Wikipedia:WikiProject United States Public Policy/Help/Reference reuse?Smallman12q (talk) 13:08, 22 October 2010 (UTC)

See Template:Reflist#Multiple uses. In an attempt to make things a little bit faster, MediaWiki caches certain template invocations (i.e. those with no parameters) the first time they are encountered in a page so it doesn't have to re-parse them every time they are encountered. That breaks for templates that can have side effects or different output each time they are transcluded. The workaround is to supply a parameter on each use. Anomie⚔ 19:48, 22 October 2010 (UTC)

Template:Superseded-Image

The link in “Afterwards please add the following code to here” in Template:Superseded-Image leads to a fully protected page. It seems that only one such subpage exists. How can the template best be fixed? --Leyo 12:04, 20 October 2010 (UTC)

The URL had "index.php" with a /wiki/ prefix instead of /w/ (and the protected page was the Index.php article) - I've fixed this and the discussion link. 86.152.209.187 (talk) 23:02, 22 October 2010 (UTC)

Thanks. As Wikipedia:Deletion requests/Superseded/2010/05/20 is the only subpage of this type, I am concerned if new subpages would be noticed. --Leyo 12:28, 23 October 2010 (UTC)

Alphabetising interwikis

As of now, an excerpt of the list of interwikis goes as follows:

• فارسی
• Fiji Hindi
• Føroyskt
• Français
• Frysk
• Fulfulde
• Furlan
• Gaeilge

As it is so, the alphabetical order is mistaken. Føroyskt should be placed after Furlan and immediately before Gaeilge, because Ø is the last letter of the Faroese alphabet, coming far after R and U. Generally in alphabets of Scandinavian languages special characters like ÄÆÖØÅÞ are considered separate letters (rather than variants of other letters) and are placed after Z, and this order is observed when alphabetising words or names. Could something be done so the alphabetical order gets fixed? --Theurgist (talk) 14:03, 23 October 2010 (UTC)

Interwiki links are shown in the order they are in the article, but at English Wikipedia, the order defined at meta:Interwiki sorting order#By order of alphabet, based on local language should be used and some automated tools (e.g. AWB) enforce it. I think discussion to change that order should go there, although the talk page seems quite dead. Svick (talk) 16:31, 23 October 2010 (UTC)

Special:Prefixindex

Resolved

 – I hardly believe any of the current living people are alive when year 3000 has come... HeyMid (contributions) 15:22, 23 October 2010 (UTC)

Is it somehow possible to make it so that a specific page does not appear in Special:Prefixindex? HeyMid (contributions) 15:01, 23 October 2010 (UTC)

To what end? (It's not) –xeno^talk 15:03, 23 October 2010 (UTC)

For example, there is no reason for my edit notice to show up here. Is it technically possible to do so? Or do I have to live with that? (I can, if it isn't possible.) HeyMid (contributions) 15:09, 23 October 2010 (UTC)

I don't think it is, and xeno seems to agree. But if you wanted to, for example, just show to archives, you could use Special:PrefixIndex/User talk:Heymid/2, course, that'll only work till year 3000... - Kingpin¹³ (talk) 15:12, 23 October 2010 (UTC)

That is true, but it's highely unlikely I will live for 1000 years... but that's OK. HeyMid (contributions) 15:15, 23 October 2010 (UTC)

Archives are usually in a subpage starting with "Archive", for example Special:PrefixIndex/Wikipedia:Help desk/Archives. See Help:Archiving a talk page#Subpage archive method. There are archiving templates assuming this by default. And then you can live forever without Special:PrefixIndex missing your archive pages. PrimeHunter (talk) 15:38, 23 October 2010 (UTC)

"For example, there is no reason for my edit notice to show up here." Sure there's a reason - it matches the prefix =) –xeno^talk 15:18, 23 October 2010 (UTC)

Technically, that's correct =). But I think the main focus is how the user him/herself wants to have it. HeyMid (contributions) 15:22, 23 October 2010 (UTC)

question

Is it possible to make an image link to a diff? Access Denied ^{[FATAL ERROR]} 17:51, 23 October 2010 (UTC)

Yes. ―cobaltcigs 18:11, 23 October 2010 (UTC)

Do I use the standard image link code? Access Denied ^{[FATAL ERROR]} 18:56, 23 October 2010 (UTC)

See source code of example above, or more generally use:

[http://any.freakin.url/whatso/ever.php?foo=bar&useskin=monobook [[File:Something.fmt|link=]]]

The key lies in suppressing the default link to the file description page. It also makes the latter inaccessible from the point of image display, so please exercise restraint using this on pages that matter, thanks. ―cobaltcigs 19:24, 23 October 2010 (UTC)

You can just assign something to the link parameter. See Wikipedia:Extended image syntax#Link. PrimeHunter (talk) 19:35, 23 October 2010 (UTC)

Problem with links to file description pages on mobile site

When using the mobile version, clicking on an image loads an abbreviated file description page. (example) On that page, there a link directing viewers to click it for "information about authorship, licensing, and additional descriptions". However, that page, rather than linking to the correct page (in the case of the above example, File:International Space Station after undocking of STS-132.jpg) it goes to a page with the same title, but in the article space. Anyone else seeing this? I saw it first on the Opera-based browser on my Nintendo Wii (Internet Channel) and duplicated it on Firefox 3.6.11.--Fyre2387 ^{(talk • contribs)} 20:33, 23 October 2010 (UTC)

Note that the link immediately below that (“View this page on regular Wikipedia”) works correctly. Svick (talk) 21:48, 23 October 2010 (UTC)

I'll make sure this gets fixed. —TheDJ (talk • contribs) 00:44, 24 October 2010 (UTC)

Import in phpMyAdmin

I'm trying to import the 3.8GB enwiki-20101011-categorylinks.sql.gz sql dump into mysql with phpmyadmin in XAMPP and its giving me a 413, "Request Entity Too Large". Does anyone know how to solve this?Smallman12q (talk) 01:11, 24 October 2010 (UTC)

I don't know, but the MediaWiki support desk might be a better place to ask your question. Graham87 02:57, 24 October 2010 (UTC)

There are two ways that I can think of. first, you can edit the phpmyadmin settings to allow larger upload sizes, assuming you have uberuser access (the software sets a limit, mostly (I think) to keep the server from getting swamped by tons of large file uploads). just set it back when you're done. however, it's probably easier overall to move the file onto the machine that's running the server, drop down to MySQL command-line mode and use the SOURCE command to load the data in directly. just go into the database you want to fill, and then type 'source filepath'.--Ludwigs2 04:23, 24 October 2010 (UTC)

Dropped edits

I've been having issues in the last day or so where edits appear to go through but in reality did not. One time was a rollback that gave me the confirmation screen, went back to the article and it never happened. Another was a Huggle AfD, the edit to the article never happened even though Huggle thought it did. I asked on IRC and two other people reported similar incidents over the last two days. Is there something that is known-broken? Gigs (talk) 08:21, 24 October 2010 (UTC)

Interwiki help request for category problem

Hi, this seems like a reasonable place to ask, apologies if not. I'm an editor on gv.wiki and have noticed a strange categorisation issue. As we are a very small wiki we don't have enough expertise to work out what's the problem. Basically, some pages (all templates, so far) are not listed in categories they've been assigned to, though those categories are listed on the template page itself. I have spotted a few by accident, which tends to suggest there are more unspotted out there. I have not been able to spot where the problem lies or a consistent pattern to it. For example, gv:Clowan:Ronney elley does not appear in the categories listed for it, and nor does gv:Clowan:Kishtey fys lioar. I have no idea how pages can be in a category in one place but not in another. I haven't found discussion of this issue anywhere, so I'm posting in the hope that someone can tell me what's going on; since I've no idea what might cause that problem I don't know how to set about fixing it. -- Shimmin Beg (talk) 20:59, 24 October 2010 (UTC)

Seems to be working now, I made a null edit to each of the pages, so maybe that forced it through. I expect it was just a problem with the job queue. - Kingpin¹³ (talk) 21:08, 24 October 2010 (UTC)

Thanks Kingpin¹³, I see they're visible now. I'm puzzled by how long they've been lying invisible though, and concerned about how many others may be around (though I can tidy up the ones I know about if that's all it takes). You don't happen to know any way to find them? I'm imagining not if it's a job queue thing rather than a matter of content. -- Shimmin Beg (talk) 06:00, 25 October 2010 (UTC)

FA Link

(I paste from Help desk, since they told me to write here instead) Hi, does anyone know why english wikipedia doesn't have any bot which puts a star near the interwikis in order to show they are GA/FA in other wikipedias? Thank you. --Arnaugir (talk) 12:13, 22 October 2010 (UTC)

It has one, LinkFA-Bot, but it appears to be inactive at the moment. Graham87 03:55, 23 October 2010 (UTC)

Bot will run somewhen this week.... I had to re-code and test a few changes. --Guandalug (talk) 21:12, 25 October 2010 (UTC)

Wiki's rasterization

I don't understand why Wikipedia rasterizes anything, including math formulas and SVGs. Of course PNG are better for interoperability, but why can't modern browser users like me choose to directly see SVG and MathML (this feature does exist as of now but doesn't work at all) or some other math renderer like jsMath, Mathjax, mathtex which are scalable.--Netheril96 (talk) 15:43, 23 October 2010 (UTC)

Not sure about the mathematical elements, but SVGs are rendered as PNGs presumably because there is little in the way of a downside and few browsers actually have consistent support for in-place SVGs. Firefox 4 beta 7 (a month overdue) will be the first to carry the requisite img-tag based support, for example, and let's not forget that 20% of visitors are still on IE6. It's against that a developer putting in time to give the option of a switch any time soon is unlikely. But possible in future, as support broadens. - Jarry1250 ^{[Who? Discuss.]} 16:29, 23 October 2010 (UTC)

Per the October report for Wikimedia Traffic Analysis Report - Browsers, reader browser use is 45% for IE overall and 5% for IE6. ---— Gadget850 (Ed) ^talk 16:54, 23 October 2010 (UTC)

Still adds up to over 75% of browsers that can't view SVG images via an img tag. -- WOSlinker (talk) 16:59, 23 October 2010 (UTC)

Mmm, I must be working six months behind mentally. Ah well. But the point is, it's important for Wikimedia to work around the world, in developing and developed counties. - Jarry1250 ^{[Who? Discuss.]} 18:53, 23 October 2010 (UTC)

Seems like it would be easy to serve a different version based on user-agent. As for the math tags it’s sure nice being able to select formulae as text for pasting to another medium without the aid of OCR, error-prone whether by human or cyborg. I just wish the setting “HTML if possible or else PNG” were (a) more aggressive about what is “possible” html and (b) the default preference. It seems to give up too easily! ―cobaltcigs 18:04, 23 October 2010 (UTC)

But the question remains... why? What's wrong with a PNG? - Jarry1250 ^{[Who? Discuss.]} 18:53, 23 October 2010 (UTC)

One cannot drag the cursor to select text from a raster graphic. This presents an accessibility failure, or a captcha, depending on how you look at it. ―cobaltcigs 19:24, 23 October 2010 (UTC)

But if you really cared you could just click through? If you're going to make a case for adding more prefs in the present climate, you need a good case. - Jarry1250 ^{[Who? Discuss.]} 19:36, 23 October 2010 (UTC)

Can’t do that with the math-tags as their output has neither a description page nor an intermediate svg format. Also some diagrams just as are awful, like this one (url inaccessible, legend revised in non-matching font, uggh). ―cobaltcigs 23:26, 23 October 2010 (UTC)

Is showing naive SVG that difficult to implement? I don't say it should cease support for older or clumsy browsers like IE; I mean we, especially registered users should be able to choose. Now I don't feel like I'm using a modern browser with MathML and SVG feature.

As for math, I actually think MathJax is a good idea because it will decide upon users' browser type whether to transmit MathML, pure images or CSS, which caters to both modern browser users and the large group of IE users. (I suppose promoting an open source software is not considered ad, as long as it is off the main namespace, is it?)--Netheril96 (talk) 03:49, 24 October 2010 (UTC)

An interface to allow on-wiki SVG editing, and to generate SVG from parameterized wiki-text (templates), might be a more broadly useful approach. In fact I’d favor scrapping the math tags if/when this comes available to us. PNG conversion can still exist for browsers which can’t display SVG, but even IE users may wish to download SVG content for use outside the browser. ―cobaltcigs 07:19, 24 October 2010 (UTC)

Several issues

1. A 12px PNG rendering of an SVG is much more efficient than downloading the SVG itself.
2. MediaWiki currently cannot deliver user-agent dependent content. It cannot deliver you an SVG instead of a PNG if your browser supports it. Adding such capabilities would be an undertaking and definetly not "easy" as claimed above by cobaltcigs. Also, there is no clear list of which browser started supporting which features from which version on. Such a list would need to be constructed before this can be implemented.
3. You could add an option to the prefs to "Always give me raw SVGs". That means every user who wants this feature, will have to enable it by hand (unless 2 is tackled).
4. We don't deliver "multiple" versions of pages to anonymous users, so the whole discussion is only relevant to what registered users experience.
5. Support for math in HTML is improving, but as long as we still run into regular issues where fonts have different "glyps" for the same character, there continues to be a problem. And again, it suffers from all the same issues as listed above.
6. jsMath is very CPU intensive, not suited for mobiles, not accessible for blind users (unlike our current approach, which has the latex math in the alt text) and would need integration into an extension, because it is much too large to load on every single wikipage. It is great, but has it's own drawbacks none the less. I'm also not clear if it is fully compatible with our current math syntax. This would need to be explored.
7. Who is doing the work ? No one ever volunteers to solve these particular issues (that have been open for years already), so perhaps someone should hire a team of developers on their own budget and get it done. Code donations are always welcome, it's how the project is build.

OK, so I think the most realistic way is to realize the option "MathML if possible".--Netheril96 (talk) 14:44, 25 October 2010 (UTC)

Besides, is it Firefox's not supporting MathML in html instead of xhtml/xml why the option "MathML if possible" doesn't work?--Netheril96 (talk) 14:49, 25 October 2010 (UTC)

So we are talking about features of limited value for a minor group of users. That means that such features get the priority that they deserve (low). Developers are sensitive to all arguments raised about how desired these things are, but some things are just difficult to do. Some might just require the web to grow up and support new things and supporting new things on the web can take 5-15 years. Anyway, the relevant bugzilla entries: client side SVG rendering, jsMath, list of all math bugs. —TheDJ (talk • contribs) 12:52, 25 October 2010 (UTC)

CAPTCHA?

For some reason, when I go to save some pages, I am being asked to type in a word that appears at the top of the screen. Two issues: (a) this is new and I recall no announcement about it (may have missed it, of course), and (b) it is causing some issues for Huggle. Thanks! — SpikeToronto 07:43, 24 October 2010 (UTC)

I believe your account (the new one) isn't autocomfirmed, so it makes you do the captcha when you add links. DC T•C 07:48, 24 October 2010 (UTC)

Ah! Brilliant deduction! It’s less than four days old … Thanks DC! — SpikeToronto 08:02, 24 October 2010 (UTC)

You could request WP:CONFIRMed. –xeno^talk 13:10, 25 October 2010 (UTC)

Secure server timing out

Is the secure server timing out for others? It's been down for about 12 hours for me. Rjwilmsi 08:40, 25 October 2010 (UTC)

12 hours? Wow, it's also down for me. For now, use the normal server. The secure server has always been unstable and unreliable. I prefer the normal server, but the secure version is more safe. In the future, you can use a site like downorisitjustme.com to confirm that the server(s) you're trying to connect to are down. HeyMid (contributions) 09:44, 25 October 2010 (UTC)

The foundation blogs and some other sites are down as well. It's not really critical services however, so they'll probably get to it when the sysadmins go to work today. Secure is not really unstable though. I know more cases where everything else is down and secure survives, so it's more a matter of perspective. —TheDJ (talk • contribs) 12:23, 25 October 2010 (UTC)

Yes, but the secure server has, in my experience, always been slower. And most of the times, it's down when the normal servers are not. Also, when the normal servers went down in March this year, everyone switched to the secure server, so the Wikimedia tech was forced to shut down that server. HeyMid (contributions) 13:59, 25 October 2010 (UTC)

The secure server, including the tech blog, is now up again. HeyMid (contributions) 16:03, 25 October 2010 (UTC)

http://status.wikimedia.org/ is a nice site to check out about the servers' health status, for future reference. Killiondude (talk) 16:36, 25 October 2010 (UTC)

WP:FAS

I don't know how to do this, nor do I have the time even if I did. Could anyone update and maintain the graphs at the bottom of FAS? SandyGeorgia (Talk) 14:18, 25 October 2010 (UTC)

Question

I have 2 questions:

1. Where can I download peachy? http://wiki.peachy.compwhizii.net/wiki/Downloads is giving me a 404.
2. What do negative cat_pages, cat_subcats, and cat_files in category.sql mean?

Smallman12q (talk) 16:31, 25 October 2010 (UTC)

1. You can try their site at Google Code.
2. It's probably just a glitch, hard to say what caused it. The last note on mw:Manual:Category table seems to indicate this can happen.

Svick (talk) 22:15, 25 October 2010 (UTC)

Section edit summaries in history changed arrows?

I'm looking at the history of a page, and it looks like the arrows to indicate a section edit have changed (or am I going crazy?) Specifically, I thought it used to look like

→Section name: edit summary

But it now seems to use a "<" rather than a →. However, this only happens on non-mainspace pages. Was it always like this? (Am I really crazy?) /ƒETCHCOMMS/ 17:01, 25 October 2010 (UTC)

I still see the arrow for all namespaces. — Martin (MSGJ · talk) 17:08, 25 October 2010 (UTC)

Extremely odd, the issue seems to have been fixed, now. Perhaps it is because I am using Chrome now, which has had (for me) a multitude of character/font support issues. /ƒETCHCOMMS/ 18:01, 25 October 2010 (UTC)

Status.wikimedia.org

Is it somehow possible to make it so that http://status.wikimedia.org does not redirect to https://secure.wikimedia.org/wikipedia/status/wiki/, when being on the secure server? HeyMid (contributions) 17:21, 25 October 2010 (UTC)

Fixed. —TheDJ (talk • contribs) 19:00, 25 October 2010 (UTC)

Thanks. HeyMid (contributions) 20:10, 25 October 2010 (UTC)

Handbook of Texas online template

Please be advised that Template:Handbook of Texas may need to be revised shortly, due to a new TSHA website going online TSHA. I noticed today they were "testing" and links were re-directed to a new Texas State Historical Association page, but only pointed the user to an additional link to click. It did not automatically roll over to the correct page. As of this posting, it's temporarily back the way it's always been.

Maile66 (talk) 19:16, 25 October 2010 (UTC)

As of this posting, the new TSHA website is up and running. They suggest users re-bookmark. If you look at the address bar in how it changed, it's like this Re-direct page to this New link. I'm not sure if there is a quick template fix on this, as how it changed is the right-hand end of the address line. But it will affect all Wikipedia articles that have used this template. And possibly affect any new articles using the template.

I am posting a copy of this on Wikipedia talk:WikiProject Texas.

Maile66 (talk) 22:53, 25 October 2010 (UTC)

Pending changes box on top of pp-move icon

See this, taken on Russell Brand. Is there a way to tweak the css so the protection icons (in this case, pp-move) are shifted when PC is active on a page? /ƒETCHCOMMS/ 21:06, 25 October 2010 (UTC)

 Fixed here. — Edokter • Talk • 21:17, 25 October 2010 (UTC)

Not moved far enough for Vector, it seems. /ƒETCHCOMMS/ 22:20, 25 October 2010 (UTC)

Did you purge your cache? Changes to vector.css may take a while unless you purge. — Edokter • Talk • 22:36, 25 October 2010 (UTC)

Yep, that fixed it. I had to empty out my Chrome cache via browser settings, and that did it. Thanks! /ƒETCHCOMMS/ 01:01, 26 October 2010 (UTC)

Is it possible to get an old version of an image and use both versions?

I feel the old Belk logo should be in the article. Someone just changed File:Belk_logo.png rather than letting both be displayed in the article.Vchimpanzee · talk · contributions · 21:08, 22 October 2010 (UTC)

No in fact the old version should really be deleted as WP:CSD#F5: Unused unfree images. You would need to upload a fresh file and have a very strong rationale for including two non-free logos in an article. Rambo's Revenge (talk) 21:23, 22 October 2010 (UTC)

I have no knowledge of how to do that. I contacted the person whose name is on the old image but got no response. I'm not sure how to word this, but the article just seems incomplete if you don't have the old logo.Vchimpanzee · talk · contributions · 21:35, 22 October 2010 (UTC)

The image without the blue design is very likely {{PD-textlogo}}, it's just text. The one with the blue design may also not meet the threshold of originality (it's text plus three copies of a simple geometric shape), but a second opinion would probably be a good idea. Anomie⚔ 00:00, 23 October 2010 (UTC)

The old image perhaps would be PD-textlogo, but I would disagree with the new image being the same. While made up of simple shapes, the confluence of these shapes, in my opinion, becomes a complex form not covered by PD-textlogo. — Huntster (t @ c) 00:23, 23 October 2010 (UTC)

I know we're getting into policy territory now, but are these last two comments helping make my case?Vchimpanzee · talk · contributions · 17:31, 23 October 2010 (UTC)

Yes, if the old logo is found to be PD-textlogo then it can be uploaded as a seperate image and put in the doucument (and being free it won't need to meet fair-use requirements). Suggest you ask for the second opinion as linked previously. Rambo's Revenge (talk) 17:35, 23 October 2010 (UTC)

Thank you. Done.Vchimpanzee · talk · contributions · 18:30, 26 October 2010 (UTC)

User:Magog the Ogre seems to believe it's okay to use both.Vchimpanzee · talk · contributions · 15:03, 27 October 2010 (UTC)

Internal links turn white when hovered over

Internal wiki links (like this one) occasionally disappear - or possibly turn the same color as the background - when hovered over. I've noticed this twice. It could be a Mediawiki problem, a problem with the new skin, a Firefox problem, a 7 problem, or (god willing) a problem with my computer that seems to only affect this website. Anyone else notice? Doc Strange^MailboxLogbook 19:42, 26 October 2010 (UTC)

I haven't--SPhilbrickT 21:55, 27 October 2010 (UTC)

Merging the history of two articles

Is it somehow possible to merge the history of two separate articles together? (Like if a copy-and-paste move has been done.) Or can they only remain divided? HeyMid (contributions) 20:58, 26 October 2010 (UTC)

Yes, but only admins can do it, Wikipedia:How to fix cut-and-paste moves. Mr.Z-man 21:31, 26 October 2010 (UTC)

Thanks for the link. HeyMid (contributions) 22:07, 26 October 2010 (UTC)

Wikipedia speed

Normally, this is the place where we complain about Wikipedia outages or sluggishness; well, it's extremely fast now. Of course, I can't tell whether that might be because of reduced load or any improvement, but it's so dramatic that I felt I should leave a note – in the hope of not spooking it. -- Michael Bednarek (talk) 09:19, 27 October 2010 (UTC)

One thing that has changed in the last few days is that the esams bits cache problem has been solved apparently.[5] Depending on from where you are accessing WP, that might contribute to better load times. --Morn (talk) 12:19, 27 October 2010 (UTC)

Suggestions

I am make three suggestions.

1- I want two thing from Wikipedia employees

I want the say of suggestions about Wikipedi. To do this, create a page. Wikipedia employees will read yhe suggestion and forward them to relevant units.

2- State name is continuous written in war, battle and event subjects in Wikipedia. But; races is not write. For example: Hunnic empire with ...... is fight in (date). Please write: Hunnic empire that mostly composed of Turks with ..... is fight in (date) or similiar. Race must emphasize by Wikipedia writer. for this, please notify the authors on this subject. Please missing paragraph should be added the race by Wikipedia volunteers. You should request the adding the races from Wikipedia volunteers

3- There is New section button in http://en.wiki.x.io/wiki/Wikipedia:Village_pump_(technical). Plase change as New create subject or similiar. If this problems is available in other language Wİkipedia, please apply the this my suggestion to other languages Wikipedia

4-http://en.wiki.x.io/wiki/Wikipedia:Village_pump_(technical)

in left in Village pump page:

العربية Česky Español فارسی ქართული Қазақша Magyar Македонски Bahasa Melayu Português Русский Suomi Українська 粵語 中文

Village pump may be in other languages Wikipedia. FOr example: Turkish Wikipedia. http://tr.wiki.x.io/wiki/Vikipedi:Köy_çeşmesi Please add the links of other languages to left section. FOr example: Turkish

5- Some subject have 粵語 (languages). I think install language packs. PLease add information to top of website about this problem. This information must fixed. WHen I change pages, this information must not forget

Please five suggestion must be/available in other languages Wikipedia

—Preceding unsigned comment added by 85.96.145.54 (talk) 19:58, 25 October 2010 (UTC)

Ad 1. Almost everything regarding Wikipedia is done by volunteers like you or me (including deciding what the rules are). Wikimedia Foundation that runs Wikipedia has only a handful of employees and probably won't be able to deal with your suggestions (but that largely depends on what suggestion do you have in mind). If you still want to contact them, see Wikipedia:Contact us.

Ad 2. That's what links are for. Click on the link Hunnic Empire to find that … it wasn't mostly composed of Turks, mainly because no such nation existed at the time.

Ad 4. Those links link to technical Village pumps and it seems Turkish doesn't have one. Wikipedia:Village pump does contain a link to tr:Vikipedi:Köy_çeşmesi.

Svick (talk) 22:43, 25 October 2010 (UTC)

Ad 5 - do you mean the list of languages that appears at the bottom the navigation area? This is located on the left-hand side of the standard skin. They are interlanguage links, and have to be added manually to each article. However a bot (computer program) will attempt to copy the interlanguage link from one language's article to all the other languages' articles. For example, if you link tr:BAR to en:FOO, then all the language articles that en:FOO links to will be linked back tr:BAR and tr:BAR will be linked to all those language articles. CS Miller (talk) 12:09, 28 October 2010 (UTC)

Does Google stop indexing deleted pages after a while?

I've had someone complaining on my talk page that a deleted page is ruining their reputation on the web, see Ankit Singh Gehlot. He doesn't like the fact that the deletion reason is visible. I'd like to tell him not to worry and that Google will stop indexing it very quickly but I don't know if that is actually true. Is there anything that we can do to help? Theresa Knott | Sort that Knee! 16:21, 27 October 2010 (UTC)

Google does have a web page on the topic: http://www.google.com/support/webmasters/bin/answer.py?hl=en&answer=164734 —RJH (talk) 21:40, 27 October 2010 (UTC)

OK so the noindex in the page means that next time Google crawls us the result should disappear right? So I can tell him to wait and it will sort itself out in a couple of weeks? Theresa Knott | Sort that Knee! 00:18, 28 October 2010 (UTC)

Google treats Wikipedia as a special case, to begin with it does not rely on usual crawling. The generic webmaster instructions thus do not apply. It is reasonable to expect that the page will eventually disappear from the index, but only Google knows how and when. I am in fact surprised that it did not disappear already, as otherwise Google picks up changes to WP pages in a matter of ~5 minutes. It may be that page deletions are treated as a potential temporary server failure, note that the Google search now gives a cached version of the article.—Emil J. 10:42, 28 October 2010 (UTC)

Template problem

Since the template {{Fb rbr header}} was modified to allow for use with all options up to 46 it appears to break after the 40th #ifexpr:. Is there a limit on the number of these that can be used in a template or is there some other reason for the problem occurring? Keith D (talk) 00:26, 28 October 2010 (UTC)

There are limits on template expansion. If you have an example of how the template breaks, post it in a sandbox or look at it in preview, and then look at the page source. In a comment (often somewhere near the bottom), there's a "NewPP limit report" that specifies how near to each of the limits in question the page has become; it's relatively rare for a page to hit the limits nowadays, but it can happen. See also Wikipedia:Template limits. --ais523 12:14, 28 October 2010 (UTC)

I have done it on the section that fails in an article with the following result

NewPP limit report
Preprocessor node count: 6288/1000000
Post-expand include size: 30457/2048000 bytes
Template argument size: 1642/2048000 bytes
Expensive parser function count: 0/500

Keith D (talk) 12:21, 28 October 2010 (UTC)

Hmm, that's nowhere near any of the limits, but there's nothing obviously wrong with the template (and in particular, no obvious reason why it would break at 40). Could you show me a specific example of what goes wrong? --ais523 12:28, 28 October 2010 (UTC)

You can see it on 2010–11 Hull City A.F.C. season but it happens on all the football season articles since the IP changed to allow it to be used for each number. Keith D (talk) 12:45, 28 October 2010 (UTC)

It appears as through there's a limit of 40 on nesting {{#if:}} statements. I've remved the nesting and it is now working. -- WOSlinker (talk) 13:03, 28 October 2010 (UTC)

Thanks for the fix. Keith D (talk) 20:07, 28 October 2010 (UTC)

Wrong inter-wiki links

Resolved

 – Thanks, Δ and Anomie. Apologies to any bots I've maligned. TFOWR 15:13, 28 October 2010 (UTC)

This bot edit, and similar ones in the articles recent history. On the face of it, it's bots doing their job well - adding an inter-wiki link from Nick Robinson to fr:Nick Robinson. However... the bots are all wrong: fr:Nick Robinson is un acteur américain. en:Nick Robinson is a British journalist and political editor for the BBC. Several different bots have made this mistake. Is there a quick and easy way to prevent this? TFOWR 14:51, 28 October 2010 (UTC)

Ive fixed it, you just need to go to frwiki and remove the link to enwiki. ΔT _{^{The only constant}} 14:58, 28 October 2010 (UTC)

But why was it linked to enwiki in the first place? Surely that needs to be fixed, instead of having to manually do so. Aiken (talk) 15:00, 28 October 2010 (UTC)

That would be human error [6] not much we can do about that. ΔT _{^{The only constant}} 15:02, 28 October 2010 (UTC)

(edit conflict) A careless human, apparently. For future reference, in general when this sort of interwiki problem occurs you have to go and fix it on every wiki involved. Otherwise the bots will usually re-make the error. Anomie⚔ 15:04, 28 October 2010 (UTC)

Problem with diffs and empty lines

I just submitted the the following bug:

When viewing a diff that contains empty context lines (gray), those lines are not shown. This is due to all cells in that row having no content, which reverts the row height to zero. It used to show a thin gray line, but a change in diff.css (table.diff td {padding: 0px;}) made that disappear as well. Example in URL should show an empty line directly above "==Professional acting career==".

Having the following CSS in your own stylesheet restores the empty line: td.diff-marker {height: 1.5em;}

However, that is a hack; A better solution is not to have rows with all empty cells. At least one cell needs to have content, and all rows start with a cell with a diff marker; these are either empty, or contain a + or - sign. I suggest putting a non-breaking space (&nbsp;) in the empty diff marker cells. This will cause the row to have content and force the proper height automatically.

I wonder if there is any objection to applying this temporary fix in vextor.css (or common.css) until this bug is fixed. — Edokter • Talk • 20:56, 28 October 2010 (UTC)

Both ideas seem good. Only problem with an nbsp in the diff-markers, is that they are copy pasted along with the content. Might have some unforeseen effects for some people I guess. We could also use a &#8203; zero width space, that would be harder to see, but also more difficult to remove from copy pasted content I guess, because you wouldn't spot them in the first place. —TheDJ (talk • contribs) 00:44, 29 October 2010 (UTC)

I see zero-width spaces always as squares; it is not a widely supported character (it may also be zero-height), so that may cause more trouble then a simple nbsp. The nbsp being copied is not much of a problem; the plus- and minus signs also get copied. But it could be replaced by any neutral character. — Edokter • Talk • 11:22, 29 October 2010 (UTC)

I just realized who made the change in SVN :) Thanks DJ. As a side suggestion: could the hyphen be replaced with &minus;? Also, how can I check if a revision is live here? — Edokter • Talk • 00:21, 30 October 2010 (UTC)

Fixed in r75658. Since it may take a while for it to go live, I still want to implement the temporary fix. — Edokter • Talk • 20:37, 29 October 2010 (UTC)

Done in monobook and vector. Added a to-do box on both talk pages. — Edokter • Talk • 21:47, 29 October 2010 (UTC)

I'm just curious about this

If the English Wikipedia has now been edited 422,480,253 times (I used {{subst:NUMBEROFEDITS}} to produce this number), then why are the oldid-numbers of the most recent versions just above 393,500,000? --Theurgist (talk) 23:22, 28 October 2010 (UTC)

I'm not too sure, but I think the {{NUMBEROFEDITS}} feature counts some actions (maybe log actions?) that the oldid number does not take into account. The disparity between the two figures has been gradually increasing over the last five years. Graham87 02:35, 29 October 2010 (UTC)

Lies, damned lies, and statistics. ;-)

I know that certain actions (like page moves and page protections) will register a new entry in the revision table. Perhaps they don't increment oldid? I can't investigate now, but I suppose that could be the source of a large percentage of the offset. --MZMcBride (talk) 03:35, 29 October 2010 (UTC)

Hmmm, it seems that this diff indicates that page protections increment oldid. This diff indicates the same thing for page moves, AFAICT. Graham87 14:48, 29 October 2010 (UTC)

Do null edits increment oldid? —DoRD (talk) 15:28, 29 October 2010 (UTC)

"A null edit will not record an edit, nor make any entry in the page history or in Recent Changes, etc." from WP:NULL. Killiondude (talk) 18:46, 29 October 2010 (UTC)

Spoiler-related proposal

A proposal to make the link to the general disclaimer more prominent in the standard Vector skin by putting it into the sidebar.

From the original proposal:

Wikipedia does not tag spoilers in articles, and instead the content disclaimer for the site warns that Wikipedia may contain spoilers for works of fiction. In the default skin the link to the disclaimers is not very prominent. Should its prominence be increased, and if so in what way and by how much?

There is presently general agreement to alter the sidebar so that the general disclaimer should have more visibility. As this is a site-wide change it's best to have wide discussion.

Some technical comments on implementation would be particularly appreciated, particularly by those familiar with Mediawiki:Sidebar.

Please join the discussion at Wikipedia talk:Spoiler#RFC:_Change_prominence_of_site_disclaimer_link_in_default_skin. --TS 19:39, 29 October 2010 (UTC)

No new message banner?

Is it just me, or has the 'new message' banner disappeared in the last couple of days? I have some CSS set on mine to reposition it, but nothing that should interfere with it entirely. Still, I don't seem to be getting any banner when I get a new message. --Ludwigs2 00:12, 29 October 2010 (UTC)

Only way to know wehter it is your CSS is to disable it and try again. — Edokter • Talk • 00:38, 29 October 2010 (UTC)

I've disabled it. would someone care to send me a test message? (please, not everyone - lol). --Ludwigs2 18:36, 29 October 2010 (UTC)

I've sent you one; noting it here mostly to avoid flooding your page with more. — Gavia immer (talk) 18:44, 29 October 2010 (UTC)

Got it, and that works, so it must have been something with my CSS. I'll look into that, thanks. --Ludwigs2 19:08, 29 October 2010 (UTC)

---

Ah, reopening this briefly, because I discovered the cause of the problem. I work on a Mac using Safari, and what happened is that (because of a flurry of messages) my user talk page crept into Top Sites. After that, every time Top Sites would rebuild its page cache it would check my user talk page, which would cause the wikipedia software to cancel the message notification. Since Top Sites and other RSS like things are becoming more prevalent, would it be worth looking into software modification so that the message banner is only removed when user actually visit the page in an appropriate browser? I assume that could be done just by checking the request headers.

I've resolved the problem on my end simply by banning my user talk from Top Sites, so this is more of a general question. --Ludwigs2 15:04, 30 October 2010 (UTC)

Category Verification

If I add a category, say [[Category:University of Warwick alumni]], to an article then I have no in-preview way to know if that is an existing category until I save. I have to search in a separate web browser, or save and see if it's red. Can the categories be included in the previews? — Fly by Night (talk) 21:40, 29 October 2010 (UTC)

Seems to work for me, look at the very bottom of the preview page (past the edit form, the disclaimers, and so on). Anomie⚔ 01:56, 30 October 2010 (UTC)

Yes, I see that too. What would be nice, though, in case there are any developers watching, would be if the software automatically notified you at the "Save" stage if you'd added a red link or category, and gave you the chance to correct it before the edit goes live (an even nicer touch would be if it notified you that you'd added a link that goes to a disambiguation page, and offered you a drop-down list of dsisambiguated titles to choose from, which it would then substitute in the wikitext automatically - ah, but we can dream.)--Kotniski (talk) 13:03, 30 October 2010 (UTC)

Ah, I can see! It's right at the very bottom isn't it? Wouldn't it be better to include it in the main preview section of the page? — Fly by Night (talk) 16:46, 30 October 2010 (UTC)

List of most widely used templates

Resolved

Hi tech guys. :-) To begin one of the most important jobs on the to-do list of the Wikipedia:WikiProject Accessibility, I need to have a list of the most widely used templates. The top 200 will do, as a starter. But on the long run, I bet we will have to review the top 1.000. Does anyone knows how to get such a list? Or is a bot owner willing to make this list? Kind regards, Dodoïste (talk) 12:51, 30 October 2010 (UTC)

Like this? mgiganteus1 (talk) 12:59, 30 October 2010 (UTC)

Heh, 8 minutes and I get an answer already. Tech guys sure are quick, thanks. :-)

This is exactly what I was looking for. ^_^ And it's updated regularly by a bot. How handy. :-) Dodoïste (talk) 13:12, 30 October 2010 (UTC)

Bar graphs

Is there any way to create a bar graph using wiki syntax? Specifically I am trying to create flow charts for the Missouri River article. Currently I am using timelines but this leaves me to choose between either cfs or cumecs, but I need both. Any suggestions? Shannon^talk _contribs 17:51, 30 October 2010 (UTC)

No there is not other way than using <timeline> tag or creating an image, which is what I'd do. Svick (talk) 18:19, 30 October 2010 (UTC)

Slow

I want edit a my text. http://en.wiki.x.io/w/index.php?title=Talk:Romanos_IV_Diogenes&action=edit&section=3. I change some information. I click save page. New page is open but not save. similiar page to this page http://en.wiki.x.io/w/index.php?title=Talk:Romanos_IV_Diogenes&action=edit&section=3 is open. Again I save page. But this page is very slow open. changes is save. Please save more fast changes. More fast —Preceding unsigned comment added by 78.185.137.98 (talk) 21:45, 30 October 2010 (UTC)

Page tabs

In the File: namespace (using the current new Vector theme), we used to have tabs like (in the respective colours):

1. For a file that exists: File, Discussion, Read, Edit, View history
2. For a file that doesn't exist: File, Discussion, Create

   These tabs were then changed to:

3. For a file that exists: File, Talk, Read, Edit, History
4. For a file that doesn't exist: File, Talk, Edit

For an editor who frequently works within the File: namespace, the new colouring could be quite misleading and wastes valuable time. In sample 4, I propose:

• The recolouring of the current black "File" tab back to red. Alright, now it seems to be red...
• Renaming the blue "Edit" tab to "Create" as it was before. If this needs too much work across multiple namespaces, then at least recolour it to red.

-- Rehman⁽⁺⁾ 02:32, 30 October 2010 (UTC)

The "File" tab appears in red for me for non-existent files on the Vector theme. — This, that, and the other (talk) 10:18, 30 October 2010 (UTC)

Alright thats weird, it was black just a few hours ago... ;) Changed the above proposal. Rehman⁽⁺⁾ 10:33, 30 October 2010 (UTC)

It seems that the change you suggest may be non-trivial from our end. I suggest filing a bug at Bugzilla. — This, that, and the other (talk) 09:05, 31 October 2010 (UTC)

 Done, created bugzilla:25722. Rehman⁽⁺⁾ 09:19, 31 October 2010 (UTC)

Testing skin

Is there a way to create a skin test that looks like mw:Help:Extension:ParserFunctions? What I am asking is whether it is possible to create a #switchskin: ParserFunction like so:

{{#switchskin:
  |monobook=text displayed if Monobook is used
  |vector=text displayed if Vector is used
  |#default=text displayed if any other skin is used
}}

This would be useful to display different versions of CSS in different skins, as skins vary greatly in CSS implementation. It could also be used to display a different message to different skin users. —MC10 (T•C•GB•L) 05:17, 30 October 2010 (UTC)

Proposed features of Wikipedia's software can be submitted over at Bugzilla. — Blue-Haired Lawyer ^t 14:25, 30 October 2010 (UTC)

Created bugzilla:25729. —MC10 (T•C•GB•L) 19:18, 31 October 2010 (UTC)

Nested templates and nested tables

I am currently working on a template ({{User:Matthew25187/Infobox tram network}}) that produces a standard table. In addition to several general sections, this template also supports up to 4 sections which will all look the same, for which I decided to create a helper template ({{User:Matthew25187/Infobox tram network/Era}}). The helper template should generate the script for a collapsible child table which is embedded within a cell in the main table. By itself, the helper template appears to work fine.

The problem is that the helper template is not being evaluated when called from the main template. It appears that the parameters passed to the helper template from the main template are being interpreted as part of the table and the remainder of the call to the helper template is being embedded in the output of the main template as plain text (see Example section at User:Matthew25187/Infobox tram network/doc#Example). I'm guessing this has something to do with a misinterpretation of pipe symbols as table cell delimiters instead of template parameter delimiters but can't see precisely where it is going wrong. This occurs on both IE8 and Safari 5. Any ideas? --Matthew25187 (talk) 04:52, 31 October 2010 (UTC)

Change the pipe templates "{{!}}" to pipe characters "|" for the calls to the helper template in User:Matthew25187/Infobox tram network. The pipe template is only needed where something should not be parsed as a template parameter separator (usually where a table is nested within a template argument). But in this case, you do want the helper template to be called with each of the parameters listed, so the plain pipe is needed to separate them. — Richardguk (talk) 05:46, 31 October 2010 (UTC)

Thanks for spotting that, all fixed now. — Matthew25187 (talk) 00:35, 6 November 2010 (UTC)

An infobox is a data table. And for accessibility, it is very important to not produce nested data tables. See Wikipedia:Manual of Style (accessibility) and it's dedicated subpage Data tables tutorial#Avoiding nested data tables.

Plus, the rows of the table are unsemantic: there is no row header to be associated to the corresponding data cells. See the recent correction of Template:Infobox/row. Infobox made with {{infobox}} are now completely accessible; I wish all Infoboxes could be standardized using {{infobox}}.

I am at your disposal to help you make this infobox accessible. Kind regards, Dodoïste (talk) 11:52, 31 October 2010 (UTC)

I acknowledge that accessibility is a valid concern for some users of Wikipedia. In this case I suspect that most uses of this template will not require the collapsible sections (nested tables) thus avoiding the most problematic issues for users with accessibility requirements.

I didn't use the {{Infobox}} template in part because it was not used by the infobox on which I based my own work ({{Infobox rail network}}). For some articles in which my infobox template may be used there might be a large number of "quick facts" suitable for inclusion in the infobox, but too many (taking up too much room in the article) to always be displayed. This is why I wanted to use collapsible sections so only those users requiring this information need to see it and the infobox does not overwhelm the article for everyone else. Based on numerous examples in other existing infobox templates, collapsible nested tables seemed to me to be common practice to achieve this without requiring complicated Javascript.

I note that even some infoboxes using the {{Infobox}} template still use nested collapsible tables to, for example, optionally display images (e.g. {{Infobox station}}).

I am certainly not opposed to addressing accessibility concerns but for now I'm just glad to have got it to work! I assume what you are suggesting is essentially a rewrite of the template which is something I don't have the time to do right now. — Matthew25187 (talk) 00:35, 6 November 2010 (UTC)

Edit on arrival

I propose a simple new option in Preferences where upon selection, clicking on any page would drop the user straight to the editpage mode, instead of loading the page normally. Rehman⁽⁺⁾ 07:44, 31 October 2010 (UTC)

This feature might be better suited for a userscript gadget. Not many people would use it, and too many options in the preferences will work confusing for users. —TheDJ (talk • contribs) 14:22, 31 October 2010 (UTC)

Enable Navigation popups. Hover over a link, then select actions → edit. You can even right click and select open in new tab. ---— Gadget850 (Ed) ^talk 14:43, 31 October 2010 (UTC)

HotCat updated

I have updated our gadget HotCat to a new version as used on Commons. Some new features.

• Possible to make multiple category changes
• More translations
• Support for sortkeys
• Quick child or parent categorization

Please read the documentation for information on the new options. —TheDJ (talk • contribs) 14:34, 31 October 2010 (UTC)

If you experience any problems with the new version of the gadget, please leave a message at Wikipedia talk:HotCat‎. Lupo 22:34, 31 October 2010 (UTC)

automatically setting a parameter

How do I extract the first word from {{PAGENAME}} and place it into the variable (or parameter, I'm not sure what the right name is) {{{year}}}? __meco (talk) 17:00, 30 October 2010 (UTC)

Mediawiki does not have the concept of 'variables' that one can assign values to; they are all read-only. Parameters are variables that are set when a template is called; they pass information, but you cannot store information in them. — Edokter • Talk • 17:28, 30 October 2010 (UTC)

Are you saying that the title of a page cannot be parsed to a template other than wholesale? __meco (talk) 17:38, 30 October 2010 (UTC)

Correct. Though there is a hack in the form of {{First word}}. Usage of this template is however an expensive operation, and developers have stated that the observed behavior is a hack that might break at any point in time and should not be relied upon. And that goes for all templates in Category:String manipulation templates—TheDJ (talk • contribs) 17:47, 30 October 2010 (UTC)

I saw a link to a page about expensive template calls, but I don't remember what it was. Do you know what it is? __meco (talk) 18:24, 30 October 2010 (UTC)

Maybe Wikipedia:Don't worry about performance or Wikipedia:Template limits? PrimeHunter (talk) 00:39, 31 October 2010 (UTC)

I'm not ure that was it, but those links are certainly relevant to the topic. __meco (talk) 17:48, 1 November 2010 (UTC)

New toy to play with – {{Random-color}}

This is the closest thing to a true random color generator that can currently be created on the MediaWiki platform. The source code is below:

<includeonly>#</includeonly>{{<includeonly>subst:</includeonly>#switch: {{<includeonly>subst:</includeonly>#expr:{{<includeonly>subst:</includeonly>NUMBEROFEDITS:R}} mod 16}} |0=0|1=1|2=2|3=3|4=4|5=5|6=6|7=7|8=8|9=9|10=A|11=B|12=C|13=D|14=E|default=F}}
{{<includeonly>subst:</includeonly>#switch: {{<includeonly>subst:</includeonly>#expr:{{<includeonly>subst:</includeonly>NUMBEROFFILES:R}} mod 16}} |0=0|1=1|2=2|3=3|4=4|5=5|6=6|7=7|8=8|9=9|10=A|11=B|12=C|13=D|14=E|default=F}}
{{<includeonly>subst:</includeonly>#switch: {{<includeonly>subst:</includeonly>#expr:{{<includeonly>subst:</includeonly>REVISIONTIMESTAMP}} mod 16}} |0=0|1=1|2=2|3=3|4=4|5=5|6=6|7=7|8=8|9=9|10=A|11=B|12=C|13=D|14=E|default=F}}
{{<includeonly>subst:</includeonly>#switch: {{<includeonly>subst:</includeonly>#expr:{{<includeonly>subst:</includeonly>NUMBEROFARTICLES:R}} mod 16}} |0=0|1=1|2=2|3=3|4=4|5=5|6=6|7=7|8=8|9=9|10=A|11=B|12=C|13=D|14=E|default=F}}
{{<includeonly>subst:</includeonly>#switch: {{<includeonly>subst:</includeonly>#expr:{{<includeonly>subst:</includeonly>NUMBEROFUSERS:R}} mod 16}} |0=0|1=1|2=2|3=3|4=4|5=5|6=6|7=7|8=8|9=9|10=A|11=B|12=C|13=D|14=E|default=F}}
{{<includeonly>subst:</includeonly>#switch: {{<includeonly>subst:</includeonly>#expr:{{<includeonly>subst:</includeonly>NUMBEROFPAGES:R}} mod 16}} |0=0|1=1|2=2|3=3|4=4|5=5|6=6|7=7|8=8|9=9|10=A|11=B|12=C|13=D|14=E|default=F}}

Correct usage would be like this:

<span style="border:1px solid {{User:Access Denied/Random-color}};background:#dde;">pretty colors</span>

The following box should have a different border color each time you refresh the page.

pretty colors

Feedback is welcomed. Cheers, Access Denied 05:20, 31 October 2010 (UTC)

Doesn't seem to work. I tried refreshing several times, even clearing the cache, and it always came up in gunmetal gray. --Trovatore (talk) 09:19, 31 October 2010 (UTC)

Not working here either, but not suprisingly, as those variables are only updated when the page or template is re-cached. See if you can play with {{rand}} instead. — Edokter • Talk • 12:13, 31 October 2010 (UTC)

What connection does generating random colours have to the building of an encyclopedia? Phil Bridger (talk) 00:25, 1 November 2010 (UTC)

Indeed, this toy would seem at best a waste of time, at worst a MOS violation and undue page complexity. OrangeDog (τ • ε) 16:53, 1 November 2010 (UTC)

Userpage beautification!!! Killiondude (talk) 02:00, 2 November 2010 (UTC)

Four tilde

1- http://en.wiki.x.io/w/index.php?title=Talk:Romanos_IV_Diogenes&action=edit&section=new

There are remember to sign your posts by typing four tildes(~.~~.~) text in this page and similiar talk pages. Plase not write the sign your post expression. Because; the sign can be understood wrong. People think Am I append the signature?. Please not write the sign your post expression

2- http://en.wiki.x.io/w/index.php?title=Wikipedia:Village_pump_(technical)&action=edit&section=new

http://en.wiki.x.io/w/index.php?title=Wikipedia:Village_pump_(technical)&action=edit&section=29

There are – — ‘’ “” ° ″ ′ ≈ ≠ ≤ ≥ ± − × ÷ √ ← → · § Sign your posts on talk pages: ~(4 tilde)~.~~ Cite your sources: <ref></ref>

in bottom of page

Plase not write Sign your posts. You must not use the this expression in any page. Plase completely remove the this expression from Wikipedia. Please write the Add 4 tilde (~.~~.~) (Within point) to the end of your post. and also would be more descriptive. Because; people can not know the what do 4 tilde.

Plase put the this warning to top of page within sign your posts

Please you must create all my suggestions in other languages Wikipedias —Preceding unsigned comment added by 78.185.201.64 (talk • contribs) 08:06, 31 October 2010 (UTC)

We can't change what other language Wikipedias do. We're also unlikely (or able) to completely remove a phrase from Wikipedia. Given that you didn't sign that post, I'm going to assume that you haven't understood something. Typing ~~~~ will sign your post. It's the easiest but not the only way. In summary then - please sign your posts. OrangeDog (τ • ε) 16:49, 1 November 2010 (UTC)

E-mail suggestion

info-en-q@wikimedia.org

info-en-q@wikimedia.org e-mail adress is available in some pages. For example: http://en.wiki.x.io/wiki/Wikipedia:Contact_us/Article_problem/Factual_error_(from_subject)

Please add the Please write English text to pages which be have to info-en-q@wikimedia.org e-mail adress. Please open parenthesis and write Plase write English text into of paranthesis

Please add this suggestion to all languages —Preceding unsigned comment added by 88.235.238.98 (talk) 19:15, 1 November 2010 (UTC)

Given that this is the English Wikipedia, the page in question is written in English, and "en" is in the email address, I don't see how someone could not conclude that English should be used. --Cybercobra (talk) 21:43, 1 November 2010 (UTC)

Login with Firefox

I can't login with the new Firefox(version 3.6.12). I get the message "Login error There seems to be a problem with your login session; this action has been canceled as a precaution against session hijacking. Please hit "back" and reload the page you came from, then try again." Seem okay in other browsers. Any ideas? Regards, SunCreator ^(talk) 23:34, 1 November 2010 (UTC)

Been running the same version for a while with no problems so it doesn't seem to be that. Rambo's Revenge (talk) 23:53, 1 November 2010 (UTC)

I've had that issue once or twice. It's so infrequent I figured it's a weird server glitch. DC T•C 00:51, 2 November 2010 (UTC)

Thank you. It's working now. It seems the process is somewhat of a cookie overload and I may of blocked some of them. Regards, SunCreator ^(talk) 01:20, 2 November 2010 (UTC)

New special page?

I just discovered that Wikitravel has a special page that we don't: Special:Mytalkpage, which is analogous to Special:Mypage both there and here. Two questions — (1) Besides leaving a note at Bugzilla, is there anything we could do to ask for its implementation? (2) If so, what? Nyttend (talk) 23:57, 1 November 2010 (UTC)

Special:Mytalk works fine both here and on Wikitravel. Svick (talk) 00:04, 2 November 2010 (UTC)

I've never before seen that link; thanks. Nyttend (talk) 00:08, 2 November 2010 (UTC)

Me too. Rehman⁽⁺⁾ 00:38, 2 November 2010 (UTC)

"Hidden" blocks

Shouldn't notice of all blocks be on the user's contributions page? As an example, http://en.wiki.x.io/w/api.php?action=query&list=allusers&aufrom=Troll&auprop=blockinfo says that User:Troll is blocked and gives a block reason, but the user's contributions page does not say this. Is this because an oversighter attempted to delete the block reason? PleaseStand ^(talk) 15:46, 30 October 2010 (UTC)

Their block-log says they've never received any... strange. ╟─TreasuryTag►duumvirate─╢ 15:49, 30 October 2010 (UTC)

Oh, no: the link you provided shows which users with a name starting Troll have ever been blocked: they're all Troll78 and Troll like the wind and things, but not actually Troll (talk · contribs) itself. ╟─TreasuryTag►You may go away now.─╢ 15:50, 30 October 2010 (UTC)

Troll was blocked in July 2004. See[7]. It's a very old block and there was a bug around for ages which meant some blocks didn't show up in block logs. I'd put it down to that. It can also happen when a user is suppressed or globally locked, but that doesn't seem to be the case here. -- zzuuzz ^(talk) 16:09, 30 October 2010 (UTC)

It's likely that Special:Contributions only follows Special:Log/block. As you can see from Special:Log/block, it only goes back to December 2004. However, Special:BlockList goes as far as February 2004. I think that the users are blocked based on Special:BlockList (or Meta's block and lock logs). HeyMid (contributions) 16:50, 2 November 2010 (UTC)

(EC: I didn't notice the reply above, oops!) The block log for Troll (talk · contribs) does not appear in the user's contribs page because he/she was blocked before our current log system was introduced. That user was blocked in July 2004; see Wikipedia:Block log/Archive2 and search for the term "Troll" with the quotes. Graham87 16:14, 30 October 2010 (UTC)

Here's the very first page of logs that use our current logging system, from December 2004. Graham87 16:19, 30 October 2010 (UTC)

Of course you are also right Graham. If in any doubt, any existing blocks can always be found in Special:BlockList. There are still some blocks from 2005-2007-ish which still aren't recorded in the block log. -- zzuuzz ^(talk) 16:34, 30 October 2010 (UTC)

What explains Troll Account Troll Account (talk · contribs · count · logs · page moves · block log) though? That one is considerably newer. PleaseStand ^(talk) 20:28, 30 October 2010 (UTC)

That's a global lock. You'll find the log on meta[8]. Admins can see the block reason if they attempt to change the block: "crosswiki abuse, globally locked & hidden<!--[[m:User:StewardBot|bot]]-->" -- zzuuzz ^(talk) 20:41, 30 October 2010 (UTC)

A few questions

1. Regarding time zones (CEST and CET), is it possible to make it so that the time still matches, although it switches from CET to CEST? (Halloween today, so we've turned our clocks back 1 hour.)

2. Is there any tool available to see how many times like a particular template has been transcluded? Also possible to see how many times a template has been substituted?

/HeyMid (contributions) 09:32, 31 October 2010 (UTC)

Regarding 2, transclusions can be counted using Special:Whatlinkshere and unchecking links and redirects. Don't think substitution-counting is possible. sonia♫ 10:06, 31 October 2010 (UTC)

Oh yeah, what am I talking about? Obviously I meant substituted templates, as the transcluded ones are listed there. However, what if some transclusions are removed? Can they still be found? HeyMid (contributions) 10:16, 31 October 2010 (UTC)

Also, is it possible to make an editnotice for an editnotice for your user page? If so, how do I do so? HeyMid (contributions) 10:28, 31 October 2010 (UTC)

...yes, but I highly don't recommend it. What's the point? sonia♫ 10:36, 31 October 2010 (UTC)

Nothing, really. HeyMid (contributions) 10:39, 31 October 2010 (UTC)

I don't understand what you mean by "time still matches". Do you mean server time? It is always on UTC, which does not take summer time. (Additionally, I'm fairly sure you mean "switches from CEST to CET"). Regarding your second question, some substituted templates (for example, {{subst:AfD}}) have tracking categories or use things like {{Void}} to track usage, but in general I am not aware of any way to track substituted templates. Perhaps an off-wiki tool exists, such as on the toolserver? Intelligentsium 00:21, 2 November 2010 (UTC)

Perhaps an off-wiki tool exists, such as on the toolserver? That's what I thought. HeyMid (contributions) 19:46, 2 November 2010 (UTC)

There is no reliable way to count substituted templates. (Does it still count if somebody changes the substituted text a little? What if the template changes?) But many templates that should be substituted add hidden code (like <!-- Template:Some template -->) to the pages, which could be counted from the database dump. Also, some templates which should be transcluded add Category:Pages with incorrectly substituted templates when they are substituted. Svick (talk) 21:04, 2 November 2010 (UTC)

'Database server lag' message - should it be in minutes?

I've just got this message looking at my watchlist: 'Due to high database server lag, changes newer than 2,070 seconds may not appear in this list.. I think the lag itself is just a glitch. I'm more concerned with the message though: now I can figure out that 2,070 / 60 = 34.5 minutes, but it seems a bit unnecessary to have to do the maths. Presumably the message only appears if the expected lag exceeds a given amount, so unless this is set very low (<60s?) wouldn't it be better expressed in minutes in the first place? Just a thought... AndyTheGrump (talk) 02:27, 2 November 2010 (UTC)

I would assume that anyone who knows what database lag is also knows how to divide by 60. Besides, if it's in minutes then it'll never go over 9000. OrangeDog (τ • ε) 14:34, 2 November 2010 (UTC)

Firesheep security issues

This new publicised Firesheep brings up some security issues. The possibility of others taking over your Wikipedia session when your logged into a public Wifi(with no security set) is now very real. I suggest that the E-mail options of Special:Preferences requires a password to confirm the change otherwise your password can be changed by changing your email address and requesting a new password, thus taken over your account completely. Regards, SunCreator ^(talk) 12:59, 2 November 2010 (UTC)

If this sheep is for real, then ultra strong support for this proposal. I am currently using my home wifi; wondering if anyone could do the dew right now... Rehman⁽⁺⁾ 13:30, 2 November 2010 (UTC)

You can also use the HTTPS server. Then no one can highjack your session regardless of how you connect. Just make sure you get the secure-only cookie then viewing images will not send your session cookie. HumphreyW (talk) 13:47, 2 November 2010 (UTC)

Using only a secure HTTPS connection is a good idea, see https://secure.wikimedia.org/wikipedia/en/wiki/Main_Page. I did not realise that https was currently supported by Wikipedia. Regards, SunCreator ^(talk) 14:21, 2 November 2010 (UTC)

Yet the secure server is mentioned every single time you visit the login page. —TheDJ (talk • contribs) 00:04, 3 November 2010 (UTC)

Yes, somehow I had come to the mistaken belief that login with the secure server was only making your login secure and now I realise it keeps your session secure. So folks, always use the secure server. Regards, SunCreator ^(talk) 22:57, 3 November 2010 (UTC)

Rehman, if you are using home wifi you can turn WPA encryption on besides it's less likely other people will be accessing your wifi point. The case I was highlighting above is for people accessing Wikipedia with wifi in a public place like in public cafes, for example Starbucks. Regards, SunCreator ^(talk) 14:26, 2 November 2010 (UTC)

I've got WPA on. Thanks for the info, I thought you meant all wifis... ;) Kind regards. Rehman⁽⁺⁾ 14:29, 2 November 2010 (UTC)

Okay, my lack of clarity, it's not all wifis, only on unsecured ones. Editing to clarify that. Regards, SunCreator ^(talk) 14:36, 2 November 2010 (UTC)

For those who haven't seen it, Firesheep was also discussed on Wikitech-l last week, as summarized in the Signpost. Regards, HaeB (talk) 15:05, 2 November 2010 (UTC)

You should always use the secure server when editing from a network that is not under your control like a public wifi network, especially if you're an admin or other privileged account. Possibly even just for reading, since the articles you read may suggest private information. From a secure home network you're relatively safe, but routers can still eavesdrop on your traffic en route, so you might use it all the time just to be safe. A little-known fact is that other language versions of Wikipedia are also available via the secure server (e.g. https://secure.wikimedia.org/wikipedia/fr/wiki/). Dcoetzee 01:08, 3 November 2010 (UTC)

except for... using interwikis kicks you back out... Choyoołʼįįhí:Seb az86556 ^{> haneʼ} 01:20, 3 November 2010 (UTC) scratch that... nobody ever bothered to install the secure.js @ nv... done now, works. If somebody reads this from a smaller wiki and it doesn't work, you need to copy Mediawiki:Common.js/secure.js and the "import"-line. Choyoołʼįįhí:Seb az86556 ^{> haneʼ} 01:38, 3 November 2010 (UTC)

Why does the web site info box in Firefox say "Your connection is only partially encrypted and does not prevent eavesdropping" for secure.wikimedia.org? That doesn't sound very secure to me. --Morn (talk) 12:11, 3 November 2010 (UTC)

Because our images are not available over the secure connection. —TheDJ (talk • contribs) 13:29, 3 November 2010 (UTC)

Then the suggestion to use secure for reading WP is a little useless. If I can see the images your browser is loading, in many cases I can easily guess which page you are visiting. Many of our articles have images that are linked only once or twice from the English WP. Also, if Firefox reports the connection to be insecure, this might deter people from using secure, just as it should. Any hope that this issue will be fixed or is this due to images being stored on Commons? --Morn (talk) 14:13, 3 November 2010 (UTC)

Your cookies are not sent to the image upload server, making you well protected from session hijacking. It's just that the browser cannot know that (older browsers actually didn't even consider non-secure images a problem). If you pour a lot of money in the sys admin team of the foundation, so that they have a lot of time to spare, then this will be fixed. —TheDJ (talk • contribs) 14:27, 3 November 2010 (UTC)

Then maybe there should be a clearer description of secure that points out that while it does protect against session hijacking, it does not currently offer truly private browsing. I see there is an explanation at Wikipedia:Secure_server, but it seems a bit too technical and disorganized. Perhaps that page needs Template:Nutshell at the top? --Morn (talk) 15:02, 3 November 2010 (UTC)

Fundraiser notice

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.

---

Sometimes, this button is come in top of Wikipedia:

Please read: A personal appeal from Wikipedia founder Jimmy Wales

1- Please apply the button in image to other languages Wikipedia and all websites of Wikimedia Foundation (http://www.wikimedia.org/). But; need that be in their own language.

2- Please apply the button in image to all languages of all websites of Wikimedia Foundation. But; need that be in their own language. 88.235.117.4 (talk) 18:41, 3 November 2010 (UTC)

It's being dealt with at meta & foundation. This is the English wikipedia; it has nothing to do with other language editions. Thank you. Choyoołʼįįhí:Seb az86556 ^{> haneʼ} 18:48, 3 November 2010 (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.

Strange glitch has been bothering me lately.

The standard toolbar (The one that goes Bold, Italic to <ref /ref>, {{ Cite }}) occasionally loses all the buttons to the right of Horizontal line, and I need to refresh the edit page to get them back.--occono (talk) 03:26, 2 November 2010 (UTC)

Perhaps it's a browser-related issue? Do you have a recent version?—RJH (talk) 19:20, 3 November 2010 (UTC)

It has affected me across mulitple computers and browsers. I thought getting rid of some gadgets and moving to the new skin might have fixed it, but it's been happening again.

It's still happening. --occono (talk) 03:00, 5 November 2010 (UTC)

Adding a photo

Commons has a new photo of Emma Bull but I can't figure how to add it to her article. Can someone do that?--filceolaire (talk) 23:02, 3 November 2010 (UTC)

 Done by User:Keith D, see this diff. – ukexpat (talk) 16:59, 4 November 2010 (UTC)

LaTeX \int

Compare $\int \int _{X}\int _{Y}^{Z}$ with $\int _{X}\int _{Y}\int _{Z}$, [generated from "Compare <math>\int \int_X \int_Y^Z</math> with <math>\int_X \int_Y \int_Z</math>, "], and note the line breaks before and after the second part. What is happening? --Rumping (talk) 10:49, 4 November 2010 (UTC)

The second example is rendered using HTML, albeit in a convulted way (in a table??). You could force it to render as PNG like this: \,\!\int_X \int_Y \int_Z: $\,\!\int _{X}\int _{Y}\int _{Z}$. — Edokter • Talk • 11:40, 4 November 2010 (UTC)

This bug occurs whenever large operators are used with subscripts but no superscripts. Another example is $\sum _{i}x_{i}$.—Emil J. 11:46, 4 November 2010 (UTC)

Strange rendering on image gallery on Safari

How it appears

Something odd is happening with the rendering of text in an image gallery. When the text goes onto a third line the enclosing text is displayed with a wide shadowed right border and only the top of the third line of text is displayed. I don't know if this is a browser dependant issue, I'm using the latest Safari on Mac.--Salix (talk): 15:26, 4 November 2010 (UTC)

• 
  In a Gallery: Part of the contentious frontage behind the ruined Charles Church.

That shadowed area is supposed to be a scrollbar. It looks like Safari (or OS X's graphics toolkit) can't handle rendering such a short scrollbar, it draws the trough but no arrows or scroller. Anomie⚔ 16:24, 4 November 2010 (UTC)

We are talking {{Gallery}} template here, not the actual stock gallery code. The scrollbar is due to deviating line-height. That screws up Safari. —TheDJ (talk • contribs) 20:16, 4 November 2010 (UTC)

Webkit bug: https://bugs.webkit.org/show_bug.cgi?id=33557 —TheDJ (talk • contribs) 21:00, 4 November 2010 (UTC)

I don't think that bug is related. I can reproduce the problem scrollbar in an offline test page that just puts some text in <div style="width:20ex;height:3em;overflow:hidden;overflow-y:auto">, no line-height adjustment anywhere. Anomie⚔ 23:03, 4 November 2010 (UTC)

Warning template not showing icon

The warning triangle isn't showing in Template:Uw-delete3. What's wrong? __meco (talk) 18:52, 4 November 2010 (UTC)

There appears to be a problem with thumbnails on the image servers. We are seeing similar questions on the Help Desk. I am sure it will be fixed soon. – ukexpat (talk) 19:02, 4 November 2010 (UTC)

Picture not displaying properly

For http://en.wiki.x.io/wiki/David_James_(South_African_actor) the picture is not loading; it's been displaying fine for a week. The permissions for the picture has been approved and the ticket recorded (or however you say it) so it's not a permission issue. When you click on the picture box, the picture page loads but not the thumbnail on that page. Help! I'm very new to Wiki so I'm lost. There were no new edits done between times when it was displaying properly and now. I tried both Explorer and Firefox and neither works. —Preceding unsigned comment added by Bobbyandbeans (talk • contribs) 01:48, 5 November 2010 (UTC)

I've seen many articles like that lately. Probably some server issue that will be fixed soon ;) Rehman⁽⁺⁾ 02:06, 5 November 2010 (UTC)

God I hope so ... he's going to be looking at it tomorrow and I said it was done and brilliant! It was brilliant before today! *sad face* Thanks for the reassurance, I'll try to be patient.Bobbyandbeans (talk) 02:27, 5 November 2010 (UTC)

Proposed tools launched

I've created a new project page Wikipedia:Proposed tools designed for contributors to propose new ideas for tools to help edit and administrate Wikipedia and recruit interested developers. If you have a tool idea you've been thinking of proposing but aren't sure how to get in touch with developers, or if you're a developer with a complex tool idea but don't have the time to implement it all yourself, or if you just want feedback on your idea, please use the process described there to add it. If you're a developer who would be interested in evaluating and possibly working on proposed tool ideas, please add the page and any subpages of interest to your watchlist. I'm also welcoming feedback on the discussion page regarding the Proposed tools process itself and how it can be improved. Thank you! Dcoetzee 02:41, 5 November 2010 (UTC)

External link icons proposal

This is something I've always thought and a brief discussion has pointed me here. Basically all PDFs give a symbol in links (http://www.example.pdf) instead of the diagonal arrow link thing (http://www.example.com/). I think xls should use {{XLSlink}} type icons instead (and equivalently for Word Documents, .doc) as they a user may be much more reluctant to open these links, may be technically restricted or otherwise. Also, let's face it. The |format= parameter in citation templates isn't utilised by everyone. Rambo's Revenge (talk) 15:19, 30 September 2010 (UTC)

I am fairly sure this has been discussed before. The PDF icon is defined at MediaWiki:Common.css. You can add the XLS icon with:

#content a[href$=".xls"].external, 
#content a[href*=".xls?"].external, 
#content a[href*=".xls#"].external,
#content a[href$=".XLS"].external, 
#content a[href*=".XLS?"].external, 
#content a[href*=".XLS#"].external,
#mw_content  a[href$=".xls"].external, 
#mw_content  a[href*=".xls?"].external, 
#mw_content  a[href*=".xls#"].external,
#mw_content  a[href$=".XLS"].external, 
#mw_content  a[href*=".XLS?"].external, 
#mw_content  a[href*=".XLS#"].external {
    background: url("http://up.wiki.x.io/wikipedia/commons/b/ba/Page_white_excel.png") center right no-repeat;
    padding-right: 16px;
}

You can test by adding the CSS to Special:MyPage/skin.css. This link should have the icon: http://www.example.org/example.xls ---— Gadget850 (Ed) ^talk 16:13, 30 September 2010 (UTC)

I have amended the initial example from http://www.example.pdf/ to http://www.example.pdf because the presence of the trailing slash means that the URL doesn't end with ".pdf", which defeats the object of the example. --Redrose64 (talk) 16:16, 30 September 2010 (UTC)

Think this is a great idea...many of this types of links are sometimes considered dangerous and i would love to know in advance if this is the type of link i am about to click on.16:49, 30 September 2010 (UTC)

This seems helpful for DOCs (which are occasionally used as links or sources), but XLS is pretty rare and it would often be clear from context or description. Might as well have both though. Rd232 ^talk 16:56, 30 September 2010 (UTC)

That is the perfect solution and works brilliantly Gadget850. I think it's worth making this the default for everyone. You mention this being discussed before. If an elegant solution such as yours was so feasible, I'd be suprised to see anyone oppose it. Is there any disadvantage of having such a function which is already so universally used for PDFs? Rambo's Revenge (talk) 17:40, 30 September 2010 (UTC)

IIRC, it was opposed because "we should discourage linking to propriety formats" and there was no free icon. And you may have to write a detailed explanation why it's free. — Dispenser 19:27, 30 September 2010 (UTC)

• I def support making icons for .doc, .xls, and possibly other non-web formats default for users. Protonk (talk) 17:44, 30 September 2010 (UTC)
  • I've imitated what you did to give a code that works with .doc[9] Is it worth integrating docx & xlsx into those. Anything else common enough? Rambo's Revenge (talk) 19:36, 30 September 2010 (UTC)
• While I like that this makes it clear when a document is in one of these formats, it doesn't address the basic problem with using such proprietary-format sources. The vendor can render them unreadable practically overnight. We need to have them rendered into a more stable, more trustworthy, and more widely accessible format. PDF would suffice. If we need to do this then the links to such a stable format should be provided, if not in preference to the proprietary format, then at least in parallel to it. A WebCitation or link may be helpful in archiving the result, and List of PDF software suggests several converters. LeadSongDog come howl! 21:39, 30 September 2010 (UTC)
  • I don't think such a solution is feasible or practical for most of these links. Protonk (talk) 22:51, 30 September 2010 (UTC)
    • I second this and also don't think it's related to this proposal. Can we try and keep this specific proposal rolling as I don't wish for it to sink into obscurity. What would be the next steps? Rambo's Revenge (talk) 23:44, 30 September 2010 (UTC)
  • What is the issue with proprietary document formats? PDF did not become open source until 2008, and was well used here before that. DOCX, XLSX and other Office Open XML formats are an open standard. ---— Gadget850 (Ed) ^talk 06:33, 1 October 2010 (UTC)
    • The second "O" in OOXML is widely treated as a very bad joke in the open software community. While some parts of the format are disclosed, it doesn't come close to a fully open format. The fact that the vendor has a long history of embrace and extend tactics doesn't help. But the real issue is that their active content enables so many malware attacks that the formats are routinely blocked by firewalls. Still, you are correct that visibly marking them with their file formats is helpful to users independently of whether they consider the content to be safe to use. LeadSongDog come howl! 20:03, 1 October 2010 (UTC)
    • LeadSong is absolutely right. Providing the icons is less an issue of promoting open formats and more an issue of providing a convenience to users who (esp. if they are on linux or a mobile device) might want to know that the link target will appear as a format which they can't or won't read. I would prefer that external links all land on simple HTML, but that isn't the way of the world. Protonk (talk) 21:46, 3 October 2010 (UTC)

I have the CSS for XLS, DOC and RTF at User:Gadget850/ExternalLinkIcons.css. I will probably add DOCX, XLSX and others as I find icons. You can copy the CSS or import it by adding this to your Special:MyPage/skin.js:

importStylesheet('User:Gadget850/ExternalLinkIcons.css'); // Linkback: [[User:Gadget850/ExternalLinkIcons.css]]

---— Gadget850 (Ed) ^talk 16:40, 2 October 2010 (UTC)

• Excellent work Gadget850. Should we make this the default to all users? There seem to be no disadvantages to users so shall we boldly update MediaWiki:Common.css Rambo's Revenge (talk) 18:29, 3 October 2010 (UTC)
  • I think we ought to solicit a little more input before we make a wide ranging change like that. Protonk (talk) 21:44, 3 October 2010 (UTC)
    • Okay, where? (don't say RfC) Rambo's Revenge (talk) 23:53, 3 October 2010 (UTC)
      • Naw, it could be here. Just make a little sub-section to this one saying what you want to be changed, post a notice on the CSS talk page you want to change and put something up on WP:CENT. If we have our heads up our asses, people will tell us. Otherwise we can make the change and be cool. Protonk (talk) 23:55, 3 October 2010 (UTC)

Icons

After some fiddling, I find that the icons should be 16px wide and a max of 16px high, and must be bitmap. SVG will not work since the image is being called directly and not through ImageMagic that would convert it to PNG. Icons and are certainly open to more tweaking. ---— Gadget850 (Ed) ^talk 12:56, 4 October 2010 (UTC)

The icons are going to have to be smaller than the current size. They clip when the font size is reduced, as with {{reflist}}. ---— Gadget850 (Ed) ^talk 15:15, 4 October 2010 (UTC)

The PDF icons are 16x16, is it just a case of doing this? Rambo's Revenge (talk) 15:20, 4 October 2010 (UTC)

Proposal

I propose the contents of User:Gadget850/ExternalLinkIcons.css is added to MediaWiki:Common.css to supplement the existing PDF icon and indentify other formats (namely doc, docx, xls, xlsx, rtf, and txt) that are not html.

• Support (as proposer) Rambo's Revenge (talk) 11:48, 4 October 2010 (UTC)

NB. This proposal has been listed at Template:Centralized discussion — Rambo's Revenge (talk) 12:01, 4 October 2010 (UTC)

• I don't think it's a big deal, so unless someone brings up a good reason why this should not be done, go for it. /ƒETCHCOMMS/ 12:46, 4 October 2010 (UTC)
• Support, small improvement but useful nonetheless. It adds consistency with the similar PDF icon. Dodoïste (talk) 12:48, 4 October 2010 (UTC)

• Weak oppose as unnecessary CSS bloat. Some browsers will download the additional icons whether they appear in a page or not. All browsers will have to deal with the longer CSS files. We should not be using these formats in the first place, at least when we can avoid it, and certainly not appear to condone them by having special icons for them. Many users do not have the necessary software to open them. Those who do often do not normally use this software. For a user who does not normally use Microsoft Office or Open Office, and consequently has disabled their quickstart function (which bogs down a computer at startup), loading an Office document takes ages.

However, I would support one common warning icon for all proprietary formats that are prone to viruses and loading problems. Hans Adler 13:15, 4 October 2010 (UTC)

• (edit conflict) Support for easy identification of file formats and for consistency with the PDF icon. It might look small at first, but I'm sure the readers will benefit from this. ~NerdyScienceDude 13:16, 4 October 2010 (UTC)
• Oppose Links to these file are far and few between, and I feel it does not justify bloating common.css for these icons. PDF is by far the most linked document format (after HTML). This is a solution looking for a problem. — Edokter • Talk • 13:31, 4 October 2010 (UTC)

• (@Hans Adler) This isn't about whether we should use these formats (ideally we wouldn't use PDF either, and PDFs were also used long before they became open source) it is about warning users that they may not to click on a link because they might not be able to or take a long time to open. Often |format etc. is omitted.

That's the reason for my alternative proposal of one icon for all of these crappy formats that clueless secretaries use when asked to publish something on the web. Hans Adler 14:15, 4 October 2010 (UTC)

It might not be a good idea to employ ad-hominem attacks against those who chose to put content online using a file type that you dislike. Vanilla HTML is all very well, but sometimes there are quite good reasons for using proprietary formats. Especially when the intended audience was somebody other than wikipedians, as is often the case with content on the rest of the internet that we might want to cite. bobrayner (talk) 15:14, 4 October 2010 (UTC)

• (@Edokter) I disagree with the "solution looking for a problem". A wiki search shows the existance of many (>4000) with a few false positives of those including "format" and ".xls" and, seemingly, even more for .doc.

Rambo's Revenge (talk) 13:55, 4 October 2010 (UTC)

4000 is not much. If you do the analogous search for PDF you get 70,000. Hans Adler 14:15, 4 October 2010 (UTC)

That's a terrible search. Most of them I looked at contained "format PDF" and had "xls" in some other context. OrangeDog (τ • ε) 18:46, 19 October 2010 (UTC)

• Support I think it would be helpful. I do not consider the possibility of changing standards to be a big problem, since client software to read common proprietary formats (such as Word, Excel) is generally quite good at displaying documents written by a slightly earlier version of the software - the chances of any one cited item failing to be readable by a wikipedia user because of subsequent changes to the standard are, I think, much lower than the chances of the link failing due to vanilla linkrot (my brand-new laptop copes effortlessly with Excel spreadsheets that my team created in the 20th century). And even if it does fail, it's not Wikipedia's job to guarantee that every person must be able to read every cited document even to the extent of removing citations or substituting in less-than-ideal ones. The widespread use of universally-readable formats is a laudable goal, but in the real world there will sometimes be a better .doc or .xls out there, and most people browsing wikipedia will be able to read them; if you want to cater to those who cannot, maybe citing a second source would be better than removing the first choice. bobrayner (talk) 15:10, 4 October 2010 (UTC)
• Support. It's worth noting that those who oppose the use of these formats (for which there are valid reasons) should consider that ensuring that their use is identified might help reduce their usage. Making them more visible may encourage people to replace them with alternatives. Rd232 ^talk 15:21, 4 October 2010 (UTC)
• Support. About bloody time, too. If the best reference available is in Excel/Word format, the least we could do is to forewarn the readers of that. This has nothing to do with the "promotion" of proprietary formats.—Ëzhiki (Igels Hérissonovich Ïzhakoff-Amursky) • (yo?); October 4, 2010; 15:42 (UTC)
• Weak Support Not strictly necessary, but I don't see much of a downside. --Cybercobra (talk) 15:59, 4 October 2010 (UTC)
• Support Be bold and do it. Lugnuts (talk) 17:26, 4 October 2010 (UTC)
• Support, as should be clear from my comments above. I also think we need to steer clear of "XYZ format is crappy, why support it" discussions. We don't support document formats in external links. We just link to what the source material is. If the source material exists only in .DOC (common for corporate or government sources), our provision of an indication is not a support of that format. It is simply a convenience to the reader. I can't speak to the technical issue of CSS bloat, but I don't feel bloat is a universal argument against changing the status quo. Protonk (talk) 18:49, 4 October 2010 (UTC)
• Support I think it is a good idea. Armbrust ^Talk _Contribs 19:30, 4 October 2010 (UTC)
• Support Not everyone has the capability to read DOC or XLS files. However, those icons should be linked to the article on the respective format. What if someone doesn't know what they mean? — Train2104 (talk • contribs • count) 02:19, 5 October 2010 (UTC)
• Weak Support per Cybercobra. If we can minimize the downsides... let's do it. Jclemens (talk) 04:09, 5 October 2010 (UTC)s
• Support will only help editors understand the abnormal links and it gives them the choice to click on certain types of links before hand.Moxy (talk) 05:59, 5 October 2010 (UTC)
• support seems really quite helpful, and the expressed downsides seem minor. Hobit (talk) 00:38, 6 October 2010 (UTC)
• Question Re: Hans Adler's mention of CSS bloat. How much additional data are we talking about that would need to be loaded, and how much of a potential loading speed decrease for users? Isn't there a new resource loader being written that is supposed to only load the things that each individual user needs to see? Also, are there any legal issues with making icons depicting these formats? Thanks!--Danaman5 (talk) 12:26, 6 October 2010 (UTC)
  • Comment For the css bloat, I'd say it wouldn't be significantly more than the audio and video link icon definitions in the monobook stylesheet, which I would guestimate to be around 1-2 kilobytes. As for the images and the formats they represent I would probably suggest using some sort of generic icon set since there are other similar file formats (.odt, .ods, .odp, etc.) out there too. --Dlrohrer2003 19:19, 6 October 2010 (UTC)
• Support, a sensible proposal and sound idea, makes sense. -- Cirt (talk) 05:15, 8 October 2010 (UTC)
• OPPOSE - because using icons of brands in an article is advertising - and that's not allowed in wikipedia. Spamelgoog (talk) 22:24, 8 October 2010 (UTC)

  Comment moved to correct place chronologically, also this was the users first Wikipedia space edit and the account was less than 10 hours old. Rambo's Revenge (talk) 22:34, 9 October 2010 (UTC)

• Comment-What about OpenDocument, ppt, image, executable, zip extensions?Smallman12q (talk) 16:54, 9 October 2010 (UTC)
• I don't really buy the CSS bloat argument. In the coming months Wikimedia wikis will be using ResourceLoader, which will make the argument even more spurious (or perhaps dubious, you pick). I can agree that these icons are kind of silly and rare, so I won't outright support this proposal, but I certainly wouldn't oppose it based on the current arguments put forward. Think of this as a mild way of saying don't worry about performance: if you like the icons (aesthetically) and think they should be there, support. If not, oppose. Keep it simple, Sally. --MZMcBride (talk) 22:13, 9 October 2010 (UTC)
• Support Seems a good idea to me, would be helpful. Acather96 (talk) 07:19, 10 October 2010 (UTC)
• Support. Seems like a decent proposal to me, and I don't see much of an issue with performance, especially not in the grand scheme of things. If it increases users' awareness of what they are about to click, then it is a good thing. — Huntster (t @ c) 09:08, 22 October 2010 (UTC)

• Oppose: Showing icons might be nice, but basing that on extensions is bad. Only a subset of documents will have such extensions. A subset of this extensions will not follow your interpretation. Some of these will be fatally wrong.
  Checking and amending format entries is a far better solution. -- Tomdo08 (talk) 20:20, 25 October 2010 (UTC)

• Comment: People who want to check for file extensions, can always do this in the status line of the browser. On the other hand a wrong format designation is not distinguishable from a correct one. Following the proposal would be invalidating all format designations! -- Tomdo08 (talk) 20:33, 25 October 2010 (UTC)
• Support – Useful and consistent. —MC10 (T•C•GB•L) 04:58, 30 October 2010 (UTC)
• Support. While not a perfect solution, it is much better than the status quo, unlike it has been claimed above. This solution catches most of the cases, which is enough. Titoxd^{(?!? - cool stuff)} 02:23, 5 November 2010 (UTC)
• Support- a very sensible suggestion. I cannot open non-html documents when browsing from my phone so it would be very useful to see plainly which links are incompatible with my hardware. Thryduulf (talk) 22:14, 19 February 2011 (UTC)

Polls are evil, especially when ignoring the technical niceties of how the WWW works

I de-rail this bandwagon by noting, as no-one else has done so far (unexpectedly for the technical Village Pump, and also unexpectedly since we do have an encyclopaedia around here that sort of explains this stuff), that there's no such thing as a file extension in a URL. Internet media type is determined by the Content-Type: header in the HTTP server response. One could publish an image/jpeg object with a URL ending in ".xls" and it wouldn't make it a spreadsheet file. The little pictures are not determined by actual content type. They have no guarantee of being right for readers. You really are better off ensuring that every non-HTML external hyperlink has a |format=[[Portable Document Format|PDF]] or a |format=[[Microsoft Word|DOC]] parameter in the citation template or uses one of the external link file type templates. The right thing cannot be deduced from just the URL. For universally correct results, it has to be specified alongside the external hyperlink, from knowledge of what the thing being hyperlinked-to is. Uncle G (talk) 02:11, 5 October 2010 (UTC)

• How often is a URL used on WP that looks like an XLS or a DOC not one of those? I've never come across that, so I'd guess it's rare. Rd232 ^talk 07:57, 5 October 2010 (UTC)
  • On Wikipedia we'd have to measure. But in the world at large it's not rare. The tag ends of URLs not matching, when misinterpreted as filename extensions, the Content-Type: of the object are why Internet Explorer gained "MIME Handling Enforcement".

    "Internet Explorer MIME Handling Enforcement". Microsoft TechNet. Microsoft.

    Uncle G (talk) 17:14, 5 October 2010 (UTC)

    • It doesn't matter how common it is Out There, it matters how common it is In Here. How often do articles use such URLs? Generally, the type of thing we're talking about will be a source; a source that doesn't even get the extension right is quite likely not a reliable source. So in general, I would call it rare, and calling attention to the cases where such extensions are used would likely make it rarer still. Rd232 ^talk 10:20, 6 October 2010 (UTC)
      • That you still talk of "getting the extension right" indicates that you've missed the point. There is no extension to "get right". We really do have an encyclopaedia around here that sort of explains this stuff. Uncle G (talk) 01:52, 7 October 2010 (UTC)
        • Er, what? Of course there's an extension to get right - extensions haven't been abolished (albeit superseded by mimetypes where possible), and it remains true that a document ending in .XLS ought to be an Excel spreadsheet. Mimetypes schmimetypes, that's what people expect, not least because it's usually the case. Rd232 ^talk 11:58, 7 October 2010 (UTC)

You err there. Wikipedia should not use file extensions:

• Extensions never were reliable.
• Giving extensions a special meaning was always a bad idea.
• To ignore and remove extensions is good advice.
• Documents ending in ".XLS" do not ought to be an Excel spreadsheet.
• Most documents with this ending will be, admittedly - but the case is different for ".doc" for example.
• It's not what people expect. Some will, a lot know better, and most will expect nothing.
• Even if a lot of people will expect extensions to be something special, it still is a bad idea.
• I would like to challenge the notion that most citable documents have endings with your interpretation.
• Having a special ending is rather a bad sign for quality than a good one. More so if relying on such ending. Databases of citable documents will rather have just an URL - an uniform resource locator.
• Wikipedia does not have to be the leading edge in such matters, but it's not a good idea to make it an obstacle to changes to the better.
• Giving hints that might be severely wrong is not a good idea.

-- Tomdo08 (talk) 19:53, 25 October 2010 (UTC)

• I think if we were designing a security protocol, you would have exposed a serious hole. There is no requirement that the stated extension be declared in the given URL and no reason that a given URL be a final landing point (and not a redirect page). A more full featured solution would be to write a web crawler which follows external links, classifies the content and adds a tag. But we are merely adding a convenience icon. False negatives may produce confusion (resolved by the format tag), but false positives would be the only really disagreeable outcome for readers. And like RD 232 says, how many false positives have you discovered on the web? Protonk (talk) 15:51, 5 October 2010 (UTC)
  • More than you'd like to think. Author of User:PDFbot and WP:Checklinks — Dispenser 16:40, 5 October 2010 (UTC)
    • Thats a possibly indeterminate number. :) enough to make URLs a bad first approximation? Protonk (talk) 17:06, 5 October 2010 (UTC)
    • ...or remove the existing PDF icon? Rambo's Revenge (talk) 17:08, 5 October 2010 (UTC)
  • See above. This problem is why Internet Explorer has MIME Handling Enforcement. And this isn't discovering a security hole. This security hole actually existed, some several years ago. It is CVE-2001-0727 if memory serves correctly.

    It's a fairly good idea to learn from experience in these things, and the experience is that "extensions" (which are not in fact extensions at all, and were never intended to be) don't match content types, that this is widespread enough that IE has a specific mechanism to avert the problems that this causes with IE's cache, and that really it isn't a good idea to make these assumptions, lest one repeat an error that was discovered the same year that Wikipedia was launched. Uncle G (talk) 17:14, 5 October 2010 (UTC)

Question: would it be (a) possible and (b) desirable for a bot to check for mismatches of extensions and MIME type? Rd232 ^talk 10:21, 6 October 2010 (UTC)

Yes possible - but desirable? It certainly would be limited protection against abuse, since MIME types can change, and presumably we would only check once, and as Uncle G says the hole is fixed, and we would probably be finding mistake rather than attacks. Or if you mean to "icon" the urls better, then I'm not sure I see the advantage of "iconing" them, certainly with what look like Microsoft logos. As to the extent of .xls terminated urls, there seem to be about 8,730 articles (compared with some 307,000 for ".pdf") containing them which is small compared with the total number of articles, but large in absolute terms. However many of these are "references no one will ever check" e.g. a spreadsheet of populations for several hundred towns, ref'd in each of the several hundred articles. Rich Farmbrough, 04:19, 7 October 2010 (UTC).

I didn't envisage protection against abuse, an issue which seems somewhat distant from the origin of this thread. I envisaged mistake checking. At the least, it would give statistical data on the issue. Rd232 ^talk 11:58, 7 October 2010 (UTC)

Caveats

URLs for sites that us a query will not show an icon. For example, this New York Times article is a PDF:

http://query.nytimes.com/mem/archive-free/pdf?_r=2&res=9F00E6DE1E3CE633A25754C1A9659C946396D6CF

There are probably other ways in which an URL extension nay not be recognized or spoofed. ---— Gadget850 (Ed) ^talk 17:27, 9 October 2010 (UTC)

a bot could recognise common ones (and/or check mimetypes) and add the format= parameter to cite templates. Rd232 ^talk 18:54, 9 October 2010 (UTC)

PDFbot already does that, its just it's a pain running the bot. — Dispenser 20:03, 9 October 2010 (UTC)

Consensus?

Ext	Count
html	2890966
php	2038904
htm	1362586
asp	615023
pdf	594196
aspx	426159
cgi	251490
shtml	240630
cfm	199854
stm	174180
(digits)	166514
fcgi	154070
jsp	129049
dll	104817
pl	96219
jspa	51689
do	51288
jpg	42905
action	38570
ece	34938
jhtml	30602
sd	28798
gif	27233
txt	26790
story	21465
phtml	16065
xls	12199
php3	11746
doc	11673
dsml	11366
cms	10351
ashx	9697
jp	9613
dbml	9061
xml	8708
mp3	7460
exe	5184
asjx	5114
zip	4497
shtm	4396
php4	4294
app	3754
ssf	3622
article	3277
com	3037
csv	2981
sps	2665
py	2408
mspx	2327
xpd	2277
xhtml	2084
lasso	2063
qst	1896
zhtml	1889
prl	1860
mma	1804
pperl	1639
ppt	1497
136m	1442

I'm wondering if we have some sort of consensus here now (I'm also posting so this doesn't get archived). The proposal seems to have significant support, and despitie a couple of concerns about CSS bloat, it doesn't seem to be that much especially when considering MZMcBride's comments about WP:PERF and the impending ResourceLoader. Obviously (as proposer) I'm biased so I just thought I'd see if someone much less involved would agree with the above summary and we can then get this implemented. Rambo's Revenge (talk) 10:24, 14 October 2010 (UTC)

Well I'm opposed, as it's based on a flawed assumption (see Uncle G's section and the point about php/query served documents), we shouldn't really be using these links anyway, and there's no guard against it spiralling out of control with wars over who's pet format gets to have an icon. OrangeDog (τ • ε) 17:04, 17 October 2010 (UTC)

There is nowhere that says we shouldn't use these links. Next will we rule out offline sources such as books etc. that people can't access? Also spiral out of control like it has since PDFs were added back in 2006? Rambo's Revenge (talk) 00:00, 18 October 2010 (UTC)

I don't know, it's a LOT of bloat. Isn't it better to use explicit templates such as {{PDFlink}} that adds the icon ? That works wherever you use it, and it doesn't have any of the problems listed above. I've never really been a fan of these link icons. —TheDJ (talk • contribs) 14:30, 19 October 2010 (UTC)

The problem is people don't use templates or format parameters. I've definately clicked on a PDF link before unawares and began downloading a whole large file that I don't want. Rambo's Revenge (talk) 16:39, 19 October 2010 (UTC)

A bot to fill in the format parameter is a much better solution. OrangeDog (τ • ε) 15:43, 20 October 2010 (UTC)

But won't work because some links don't even use a citation template. Rambo's Revenge (talk) 15:50, 20 October 2010 (UTC)

Indeed. This proposal at least does something to make people aware, rather than sitting at the current status queue. May not be perfect, but nothing in this world is. — Huntster (t @ c) 09:07, 22 October 2010 (UTC)

This proposal also won't work, as explained above. (wikt:status quo). OrangeDog (τ • ε) 21:49, 23 October 2010 (UTC)

The limitations of this proposal are identical to those of the existing PDF icon which has been in use for 4 years. Also consensus seems to be for implementation. Rambo's Revenge (talk) 21:43, 24 October 2010 (UTC)

Indeed there seems to be. The poll above currently sits at 16 to 4 in favour...80%. — Huntster (t @ c) 20:09, 26 October 2010 (UTC)

First polling is evil. Second, the majority of votes were added before the major flaws were pointed out. OrangeDog (τ • ε) 22:56, 26 October 2010 (UTC)

Using cite templates with format parameter is good, but using a bot for automatic filling of the format parameter is bad, too: Extensions are ambiguous. -- Tomdo08 (talk) 20:27, 25 October 2010 (UTC)

Showing icons might be nice, but basing that on extensions is bad:

• Only a subset of documents will have such extensions.
• A subset of this extensions will not follow your interpretation.
• Some of these will be fatally wrong.

Checking and amending format entries is a far better solution. -- Tomdo08 (talk) 20:09, 25 October 2010 (UTC)

I'm a bit confused by your statement. Are you saying that having at least a large percentage of existing links to documents iconified is worse than having none? That the only viable solution is to have each link checked by hand? I apologise if I'm mis-interpreting, but these arguments seem a bit outlandish. — Huntster (t @ c) 20:06, 26 October 2010 (UTC)

So I'm here to provide fuel for the burn *.* campaign ;-). I wrote a tool to count all the extension from the 16.5 million external links and produce the table (right) after a few hours of playing around with OurSQL. The more recognizable extensions are in bold. There are nearly 50 times more links with the extension of .pdf than .xls or .doc.

We have definitions in MediaWiki:Common.js which override the default icon. Monobook and Vector use the following extension to style the link icon:

audio (blue speaker or musical note icon)

.ogg, .mid, .midi, .mp3, .wav, .wma

video (film strip or reel icon)

.ogm, .avi., .mpeg, .mpg

document (blue document or gray document icon)

.pdf

Notes: The match on .pdf? and .pdf# is done so linking to specific pages (example.pdf#page=3) or anchors will display correctly, it is generally not needed for other types of files. Acrobat (default for most people) web browser can take up to two minutes to load before rendering the PDF. Additionally, the major of PDFs are more than 0.5 MB and picture heavy ones can be over 100 MB. — Dispenser 04:21, 28 October 2010 (UTC)

What are all those exe links doing there? They should almost certainly be removed. OrangeDog (τ • ε) 22:30, 28 October 2010 (UTC)

Many of them are CGI programs on web servers (1,568 for zipinfo.com), but 316 seem to be regular .exe. Half of that link to http://elections.sos.state.tx.us/elchist.exe CGI program, some more appear to be Microsoft issued executable files, and other are from domains I generally don't recognized. — Dispenser 21:38, 2 November 2010 (UTC)

Firstly a very belated thank you to Dispenser for knocking together the very informative table on extension formats. As for depecting "." formats I still don't see evidence not do so. To me the facts are that most extensions do match (security threats and type II errors hardly seem applicable to the majority of readers/users). Furthermore, people fail to use the format parameter (or sometimes cite templates) and, like the PDF icon that has been around for years, we have a good way to inform readers that a link is potentially undesirable for them to open regardless of how the link is formatted. I've been advised in audited content work to add XLSlink which does a worse job of this idea. Surely making it automatic is beneficial in spite of above caveats. I'd also propose some form of icon for images is added in light of Dispenser's findings. Rambo's Revenge (talk) 00:40, 1 November 2010 (UTC)

Things like this are slow. These styling matches are relatively complicated matches and on pages with many links (think Barack Obama), and on older computers, you can probably notice the difference in HTML rendertime between them being there, and them not being there. I'm not against them per se, I just want everyone to really grasp the fact that the more complicated CSS matches we add, the slower pages will be become, especially to those not fortunate enough to have the latest hardware and software (think 3rd world countries). Complicated CSS adds up, just as complicated Javascript adds up. The more complicated a page becomes, the more users we are excluding from having a usable browsing experience. It's not the raindrop, it's the fact that at some point the drops will form a pool of water. The increase in client rendertime will affect every page and the benefit cases are 50x more rare than the PDF case (if one doc link per page, then 1 in 290 or so pages will have a match for a link with a .doc extension). Fixing "rare" cases is relatively expensive (It is why we often use inline css for templates, instead of adding it all to the main stylesheet [only has wikitable, infobox and warningboxes]). —TheDJ (talk • contribs) 01:10, 1 November 2010 (UTC)

I understand what you are saying but surely that argument goes two ways. Won't older computers also want to identify and not open document links etc. as they are large and can crash browsers etc. Rambo's Revenge (talk) 23:28, 4 November 2010 (UTC)

If people are concerned they can look at the extension. Unless the extension doesn't match of course. OrangeDog (τ • ε) 09:37, 5 November 2010 (UTC)

The extension is not always visible. In fact unless there is a status bar in Firefox you wouldn't be able to see it at all (unless you opened/saved/bookmarked it and then examined that). Not very likely for users. The point is an icon makes a visual impact of beware you might not want/be able to open this. The casual user isn't examining every url/extension they click on. Rambo's Revenge (talk) 22:51, 5 November 2010 (UTC)

Image scalers

What happened with the connection to the image scalers at about 13:50 UTC?[10] At the same time, the load on the upload squids went up, too.[11]. Currently, no thumbnails are served for me. Lupo 14:17, 2 November 2010 (UTC)

If i understand correctly, the NFS connection between the scalar and the fileservers went down, causing congestions due to requests that take long to time out. —TheDJ (talk • contribs) 23:59, 2 November 2010 (UTC)

And then apparently some failing processes on the upload servers logged their failures, and filled the disks with the logs.[12] sq51 still has a full disk.[13] Lupo 11:35, 3 November 2010 (UTC)

I'm not sure if you're referring to my problem, but I uploaded new versions of File:Mitsubishi Celeste.JPG and File:DodgeArrowGS-rear.jpg and they're not showing up in any format besides the original. No thumbnails, no resizes. Should I just sit back and wait or should I upload them again?  ⊂| Mr.choppers |⊃  (talk) 17:29, 4 November 2010 (UTC)

For two days unable to alter image size, increase or decrease, in infoboxes. Been advised to wait for how long????REVUpminster (talk) 19:42, 4 November 2010 (UTC)

It seems they are working on it at Commons. --Saddhiyama (talk) 21:43, 4 November 2010 (UTC)

• Now working at "275px and 300px but not 250pxREVUpminster (talk) 12:42, 5 November 2010 (UTC)

XCF files

We have a few .xcf files on English Wikipedia[14] XCF is probably quite useful on Wikimedia Commons, but I don't see why we allow them to be uploaded here because the mediawiki software doesn't have a renderer for XCF. I think the free files should be moved to Commons, and non-free files like File:Old Weather logo.xcf replaced with a svg. John Vandenberg ^(chat) 22:09, 2 November 2010 (UTC)

Now we are gonna make a in what file formats are allowed to be uploaded to what wikimedia project ? Why further complicate it for people ? I don't see the point in this. It's a file. Transfer it if you want but I see no reason whatsoever to disallow uploading a free and valid fileformat solely on the English Wikipedia. —TheDJ (talk • contribs) 23:54, 2 November 2010 (UTC)

Maybe because it's not rendering? Just a thought. This is an encyclopedia, and keeping files that aren't renderable doesn't seem to be a part of its scope. Commons is more of a more "if it's freely licensed and a file, we'll take it" nature. Killiondude (talk) 05:20, 3 November 2010 (UTC)

Exactly. We even have a CSD for it - Files that the MediaWiki software is unable to read or generate resized thumbnails of. OrangeDog (τ • ε) 11:06, 3 November 2010 (UTC)

I see no reason why we cannot host free formats that can be useful. The only reason to upload an xcf, is to upload something in the original format, so that others can recreate images based on your actions. such seems perfectly reasonable, and it not at all like what CSD F2 should be tackling. In that case, we have some high res PNGs that we could be deleting. —TheDJ (talk • contribs) 13:32, 3 November 2010 (UTC)

It is in scope, if it concerns for instance a restoration of an image that is PD in the US, but not in the country of origin (not allowed on Commons). The file is perfectly renderable, what you guys mean is that it cannot be used inline in articles, because the software can't make automatic thumbs of it. But nothing stops you from downloading, creating a new rendering and uploading a new PNG based on the original files. The point of these kinds of files is often exactly this. XfD .xcf files that you think no longer serve a purpose, move to Commons files that are allowed on Commons. —TheDJ (talk • contribs) 13:47, 3 November 2010 (UTC)

It's not renderable by MediaWiki, which is why it is out of scope. For the same reason uploaded .doc or .txt files are removed, even if they are free and relevant. OrangeDog (τ • ε) 16:42, 4 November 2010 (UTC)

You advocate forcing future photo-restoring editors to start from scratch when we could easily provide them an easily-editable format from which to make improvements? Bad idea, IMO. Doc and txt files are a different matter, because they're text which can be better stored as wikitext for our purposes. Anomie⚔ 17:40, 4 November 2010 (UTC)

No, I advocate moving them to Commons, as does the OP. OrangeDog (τ • ε) 19:04, 4 November 2010 (UTC)

You mean the Commons that chooses to abide by national copyright rules unlike English Wikipedia, which often happily ignores copyright on works outside the US if it is pre 1923 ? Not everything fits on Commons. —TheDJ (talk • contribs) 21:41, 4 November 2010 (UTC)

Nor here. OrangeDog (τ • ε) 09:39, 5 November 2010 (UTC)

This looks like an entirely a theoretical problem - the xcfs listed in that search are mostly on Commons, and the few that are on en could be moved there. If they are acceptable on Commons, they should be moved to Commons of course... However Commons cannot host certain files due to copyright issues, which en.wikipedia can host (those that are PD in US but not country of origin) and its possible that restoration work on such images will involve xcf files. If that happens its probably something to ignore, after all these files do still have an encyclopedic use - even if they can't be included in articles. Note also that if you upload a jpg thumb over the xcf you get a viewable jpg File:Henry Cuyler Bunner.xcf with the xcf in the history - that dodges the clause of F2.--Nilfanion (talk) 12:06, 5 November 2010 (UTC)

I'd recommend not uploading a jpg over the xcf like that. It's misleading to the reader, it's liable to be moved to the .jpg name by a well-meaning editor who isn't aware of that trick, and it's quite possible the devs may at some point remove the ability to do this just as it is currently not possible to upload a jpeg with a .png extension. Anomie⚔ 16:13, 5 November 2010 (UTC)

Question about qantity of categories in english Wikipedia

I am currently in the process of writing a paper about Wikipedia Categories. Ive done most of my research using Wikipedia API to get approximate numbers and to analyze categories structure. But i also want to know how many categories does Wikipedia has at the moment, and i just cant find this information. I could, of course, just install a previous Wikipeda DB dump and make a query to cout all categories, but that is an overkill. And using API for that would take too much time and strain Wikipedia servers.

So i am asking this question here: How many categories does Wikipedia have and is there a place where this info is posted or could be extracted from? i've beed told on the help desk that this can be done with magic words, but the required functuion seems to be disabled at the moment. Is there another way? --DSUmanskiy (talk) 11:21, 3 November 2010 (UTC)

mysql> select count(*) from page where page_namespace = 14;
+----------+
| count(*) |
+----------+
|   638834 |
+----------+
1 row in set (16.30 sec)

That answer your question? ΔT _{^{The only constant}} 11:33, 3 November 2010 (UTC)

It does! Thank you very much:)--DSUmanskiy (talk) 11:59, 3 November 2010 (UTC)

Here's another place that's got the info -- Wikipedia:Database reports/Page count by namespace -- WOSlinker (talk) 15:02, 3 November 2010 (UTC)

That's a rather interesting page since with the old revisions you can graph the growth by namespace. — Dispenser 08:00, 4 November 2010 (UTC)

Wow! Thats pretty nice! Thank you! --DSUmanskiy (talk) 20:54, 5 November 2010 (UTC)

Pages removing themselves from Watchlist

I added Wikipedia:Village Pump (proposals) to my watchlist and three days later it somehow managed to change to unwatched. I rewatched the page, and it moved back to unwatched 10 hours later. I readded it and it became unwatched two days later. I then readded it and it stayed on my watchlist until today (about 5 days). I just readded it to my watchlist. What is wrong with Watchlist? I haven't removed it myself. It has happened to other, non=article space pages. Thanks, --Alpha Quadrant ^talk 13:36, 3 November 2010 (UTC)

Do you use AWB? OrangeDog (τ • ε) 16:38, 4 November 2010 (UTC)

I have used it once, but not recently. --Alpha Quadrant ^talk 04:19, 6 November 2010 (UTC)

Templated Link & Category splitting & General Background for a Wikia-Project

Hi, I hope that I can ask my Questions here, even if they dont refer to wikipedia itself, but an wikia project. But the tehnics shouldbe the same?

However. I have gut two questions to ask. First of all: Does somebody know what to do, if I want to use a lage background image for the main background (the background that cannot be more in the background, behind the wikia main frame). And this Image shall vary its size, depending on how large the screen setting is set. Also, when scrolling up or down, the background shall stay where it is, meaning don't scroll alongside wth the wikia frame. I do know that it is possible with html. I'm not the Admin in this Wikia Project, but I shall say how it works. Is HTML being used for such Wikia settings?

The other one should work with wikipedia, too... I have got a problem with a template... I wrote this template:

{| class="toccolours" style="float: right; clear: right; margin: 0 0 1em 1em; width: 225px"
|-
!style="background:#beb998"|{{PAGENAME}}
|-
|
{| style="background: #ffffff; width: 225px; margin: 0 auto;"

[...]
Blabla (some lot more written here I spared out here now...)
[...]

|- style="vertical-align: top;horizontal-align="
|'''Typ'''|| {{#if:{{{Typ|}}}|{{{Typ}}} [[Kategorie:{{{Typ}}}]]|[[Image:Warning Yellow.svg{{!}}19px{{!}}Bitte ergänze den Typ des Zauberspruchs.]]<includeonly>
[[Kategorie:Unbekannter Zauberspruchtyp]]</includeonly>}} 

[...]
|}|}<noinclude>

Now, If you write any Word behind |Typ= if you use the template, a category and link is generated there later. My Problem is, that i want to generate automatically two, three or even four links and kategories if I write more words behind this |Typ= . It would be possible, that every word is generated for its own as a category and link.

E.g. If I write: |Typ=Door, Chair, I want to have Links Generated for Door and one for Chair, not one Link for [[Door, Chair]], same for the category. One for each, not one for both together... What shall I change? I tried loads of things, but at the end I dind't find out what to do there... As I said, its also possible that each word gets a category and link, even an if or and or or or whatever. I could also use something different than the , ... Hope someone can help =) Greetz Marc 134.93.45.143 (talk) 16:25, 3 November 2010 (UTC)

Regarding your second question, you could use several parameters, e.g. Typ1, Typ2, …, each of them adding a category of its own, if it is set. Svick (talk) 22:59, 3 November 2010 (UTC)

Thanks, the background cannot be solved, as I found out (cause we don't have server access) and you helped for the second one thx =) —Preceding unsigned comment added by 79.220.53.45 (talk) 19:11, 5 November 2010 (UTC)

Creating a link from a radio button

If I wanted to create a link for the:

preferences >> appearance >> skin

where a person clicks a link and their skin changes, how would I go about doing this?

so for example, MonoBook, the radio button coding is:

<input name="wpskin" type="radio" value="monobook" checked="checked" id="mw-input-skin-monobook" />

I have read over Manual:Parameters to index.php and am not sure how to plug in these values.

Thank you so much! Adamtheclown (talk) 16:13, 4 November 2010 (UTC)

If you're wanting to give them a link to view a particular page in a different skin (as done by the "preview" links in WP:SKIN#Screenshots), use useskin. If you're wanting to actually change their preference, there is no way. Anomie⚔ 16:29, 4 November 2010 (UTC)

Is there a way to use this to view a page as a logged out user would see it?

Lemmiwinks2 (talk) 02:40, 6 November 2010 (UTC)

Odd login problem

I've been contacted by Concertphotos (talk · contribs), formerly Festivalphotos (talk · contribs), whose username was changed. He gets "Login error. There is no user by the name "Concertphotos". User names are case sensitive. Check your spelling, or create a new account." when he tries to log in. He has (once) successfully used the account but has since consistently gotten this message. I've offered the usual suggestions about spelling, trying different browsers, clearing caches, etc., and they seem to be sufficiently clued that they've tried the obvious remedies themselves. They can't get into Commons or svwiki either. Does anybody have any ideas? The login sequence is definitely not my area of expertise. I've suggested just creating a new account, but thought I'd see if there's a known issue. Acroterion _(talk) 03:55, 5 November 2010 (UTC)

When I tried to login under the name "Concertphotos", the error i got was "Login error Incorrect password or confirmation code entered. Please try again." And that is expected. Are you completely sure he is entering the correct user name and that he is logging to the English Wikipedia? He can't login to commons or svwiki, because the account doesn't exist there, it is registered only here (see [15]). Svick (talk) 21:21, 5 November 2010 (UTC)

He's apparently managed to log in. Since he's interested in photography I'll get him to log in to Commons while logged in here for SUL purposes. Acroterion _(talk) 22:06, 5 November 2010 (UTC)

Table background colour depending on entry

Hi @ all. It's me again and I search for a possibility to write a template for a table in such a way, that it automatically chosses the background colour accoring to the entry for the field.

Meaning: The table will only get 3 kinds of thing written for the area of interest: That's * -10% * -5% and * 0%
Depending on what is written in the text, the background coulour of the boxes shall use an according colour. Meaning: all table boxes with -10% written in shall use the colour code #dddddd for its background. All boxes with -5% shall use the colour code #dafbff, all ones with 0% written in it shall use the colour code #dbffda... Is that possible e.g. using an #if command?

Hope someone can help, greetings Marc 79.220.53.45 (talk) 19:20, 5 November 2010 (UTC)

You can write a template to do that, see [16] for the code and [17] how to use it. Svick (talk) 22:11, 5 November 2010 (UTC)

WOW! Perfectly done! Tahnk you very much! It's all working now =) greeting marc 79.220.44.112 (talk) 04:47, 6 November 2010 (UTC)

Why does monobook have a white background all of a sudden?

This is weird, and it's only in article space. Is it a glitch or just something I never noticed before? Access Denied – talk to me 23:14, 5 November 2010 (UTC)

Which background ? The article namespace content area has always been white. Perhaps you mean the background image behind the content areas ? You might have a broken copy in your browser cache, because it shows up just fine for me. —TheDJ (talk • contribs) 01:16, 6 November 2010 (UTC)

Oh, it's always been white. Must have either forgot or never noticed. Access Denied – talk to me 01:25, 6 November 2010 (UTC)

Well, unless your username is Alpha Quadrant, I changed my background color to match my talk page. --Alpha Quadrant ^talk 04:14, 6 November 2010 (UTC)

How to automate?

There's a fair number of article (perhaps 150 or so) such as Alexis Hornbuckle, with a link to a WBCA All-America list, a scorebook, or MVP award, or some combination. Because the WBCA revamped their site, all the links are dead.

I know the location of the new link. For example the dead link should go to this

Is there a way to make all these changes without doing every one manually? --SPhilbrickT 18:03, 5 November 2010 (UTC)

If you know what the system is (i.e. rules to translate old links to new ones), you could try using AWB. Svick (talk) 21:55, 5 November 2010 (UTC)

OK, I'll check that out.--SPhilbrickT 22:49, 5 November 2010 (UTC)

Thanks, that worked. I've used AWB before, but not for something like this. Good to know.--SPhilbrickT 13:48, 6 November 2010 (UTC)

Help with footnotes

I've copied the following from Wikipedia:Manual of Style (footnotes). I know how to group notes and I know how to group references, but I don't know how to group notes with nested references. Is it possible to explain Claim B and Claim D with the same note AND with a nested reference? Thanks! Location (talk) 03:48, 6 November 2010 (UTC)

Claim A

Claim B

Claim C

Claim D

Claim E

References

• Yes, use the coding {{#tag:ref|Lorem ipsum<ref>dolor sit amet</ref>|name="example"|group="nb"}} and then <ref name="example" group="nb"/> in the second instance.

• consectetur adipisicing elit
• sed do eiusmod tempor incididunt ut labore et dolore magna aliqua

Notes

Lorem ipsum

References

dolor sit amet

Hope that is what you wanted. Rambo's Revenge (talk) 14:44, 6 November 2010 (UTC)

Yes, yes, yes! Thanks a lot! Location (talk) 17:32, 6 November 2010 (UTC)

I’m deeply curious as to the situation(s) in which you’d use a footnote to a footnote. ―cobaltcigs 17:42, 6 November 2010 (UTC)

HA! It's funny that you mention that quote because the example I was working on was Isham Randolph, who has descendants in common with Lewis of Lewis & Clark. Location (talk) 17:57, 6 November 2010 (UTC)

See WP:REFNEST. ---— Gadget850 (Ed) ^talk 18:00, 6 November 2010 (UTC)

Thanks! I read that, however, I couldn't get it to work until Rambo's Revenge provided the working example above. Location (talk) 18:32, 6 November 2010 (UTC)

Search my edit summaries

Is there any way to search my edit summaries for a particular word? DuncanHill (talk) 13:20, 6 November 2010 (UTC)

I might have the ability, what are you looking for exactly? ΔT _{^{The only constant}} 14:09, 6 November 2010 (UTC)

I'm trying to get an ide of how many edits I've made fixing dablinks, so would be searching for "dab" or "dabbed" or "dabbing" or "disambiguated" and the like. DuncanHill (talk) 14:14, 6 November 2010 (UTC)

select select count(*) from revision where rev_user = '1739907' and (lower(rev_comment) like '%dab%' or lower(rev_comment) like '%disambiguated%');
+----------+
| count(*) |
+----------+
|     8846 |
+----------+
1 row in set (4 min 33.22 sec)

Enjoy. ΔT _{^{The only constant}} 14:34, 6 November 2010 (UTC)

Many thanks. DuncanHill (talk) 14:36, 6 November 2010 (UTC)