Wikipedia:Village pump (technical)

Village pumps Idea lab Miscellaneous Policy Proposals (persistent) Technical

Skip to Table of contents First discussion Bottom of page

See also: MediaWiki talk:Common.css

The technical section of the village pump is used to discuss technical issues about Wikipedia. Bugs and feature requests should be made at Bugzilla (How to report a bug). Bugs with security implications should be reported to securitywikimedia.org. Newcomers to the technical village pump are encouraged to read these guidelines prior to posting here. Questions about MediaWiki in general should be posted at the MediaWiki support desk.
Please add new topics to the bottom of this page.
« Older discussions, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105

Centralized discussion
Proposals	Discussions	Recurring proposals
Discussion on changing guidelines for loss and restoration of bureaucrat tools. Discussion on the format of the Core Contest for 2013, input requested here An RfC on a replacement for naming conventions related to Taiwan/Republic of China which was marked archived without a replacement in its place A proposal to use our citation templates to provide microformat data. An RfC regarding ticker symbols in article leads. An RfC on "intractable conflicts of interest". Invitation to complete a civility enforcement questionnaire. Note: inactive discussions, closed or not, should be archived.
archive talk edit history watch

Blurry graphs

Why is it that all the graphs and all or almost all of the maps on Wik pages are blurry beyond legibility? Photos are crisp and usable, but I have to click on the graphs to get any use out of them. Since I am in a country where internet connection is slow, this means adding about a minute or doing with-out. Can't this be changed? Kdammers (talk) 05:18, 23 November 2012 (UTC)

• Widespread conversion of maps/graphs to PNG: Years ago, there was a massive, rabid, almost tyrannical push or shove to force all maps or lettered graphs into the blurry, fuzzy, hazy, cloudy PNG format which is 5x-8x-20x times slower than the JPEG format. The original JPEG maps or graphs were then rapidly, viciously deleted. The oft-noted reason was for "clarity" of labels in PNG-format images to avoid "image artefacts" in some adjacent-color mismatches, even though the JPEG format tended to sharpen the contrast of dark lines and letters against the background colors. Plus, in cases of questionable labels, then a JPEG image could be quickly expanded to a larger size, while the slow gargantuan PNG-format image took eons of time to slowly enlarge for better readability. And get this: when small JPEGs were changed into cumbersome PNG format, then article pages were switched from mostly text-based data into becoming mostly PNG-based data, far larger than the text data in the page. I must have run hundreds of experiments to confirm that the JPEG images were almost always clearer than PNG-format images when scaled to similar sizes. Meanwhile, GIF-based thumbnails were force-blurred for some months, but eventually returned to clearer thumbnails as in years past. Later, many images were converted into variable-size SVG-format images, which are rendered as blurry, fuzzy, hazy (etc.) PNG-format images, as thumbnails. I guess the rationale has been to always use PNG-format images, just in case a labeled map or graph might use a rare color combination which "bleeds" artefacts into the nearby mismatch colors. However, in practice, most labeled images tend to avoid bleed-prone color edges (such as lime green on tan), so the quick, crisp JPEG format could be used, except for rare color borders. Now, huge JPEGs can also turn blurry when thumbnailed very small, so a reasonable trick has been to excerpt a small-scale closeup (such as downtown map) from a huge multi-megapixel image, and then use that reduced-size closeup to generate the small, clear thumbnail-size images, much clearer than thumbnails based on the huge megapixel JPEGs. Another tactic is to thicken the major labels on maps to 3-pixel letters (or "2.5 pixels" using 1-gray + 2-black pixels in lines). Hence, when 3-pixel letters are thumbnailed as JPEG format, then the sharpening of labels is improved so even small thumbnails have readable labels. It is amazing how simple to create readable maps in lightning-fast JPEGs which are still readable as tiny thumbnails, and quickly enlarged to confirm small labels. Unfortunately, the genocide of JPEG-based maps, now deleted into mass graves, has made the restoration of readable maps a very difficult problem. -Wikid77 (talk) 06:26, 23 November 2012 (UTC)

Wow, I've seldom seen a rant so long and yet so misinfomed. Anomie⚔ 15:45, 23 November 2012 (UTC)

The conversion to slow, blurry PNG-format images has been a colossal failure, which thwarted real improvements to map or graph design; so now we need some real solutions. -Wikid77 03:54, 25 November 2012 (UTC)

These days, when I see a post outside of a WP:!VOTE which begins with a bullet and some boldface, I expect that sooner or later I'll apply WP:TLDR. So, whether it's misinformed or not I couldn't say. I do know this: if you convert an image which is in a lossy format (like JPEG) to any other format (whether lossy or not) there's no way that you can recover the lost information and so sharpen up the image. --Redrose64 (talk) 19:51, 23 November 2012 (UTC)

@Kdammers: Can you cite some examples? What is your approximate screen size (in inches) and resolution (in pixels)? —[AlanM1(talk)]— 08:37, 23 November 2012 (UTC)

Yes: http://en.wikipedia.org/wiki/List_of_countries_by_population, http://en.wikipedia.org/wiki/Gini_coefficient Kdammers (talk) 09:18, 23 November 2012 (UTC)

I have no problems whatsoever. I do believe that the quality of the default PNG handler in Internet Explorer is rather poor. It would be useful to know if that is perhaps the browser that you are using. —TheDJ (talk • contribs) 09:46, 23 November 2012 (UTC)

Have you changed your browser zoom from the default of 100%? (It's easy to do this accidentally and some browsers remember the setting.) That makes text larger but also shows images at a scaled up size. (In most browsers, Ctrl-0 resets the zoom: control and zero.)

Brion Vibber recently implemented a nice enhancement for tablet browsers (where high-DPI and user rescaling are ubiquitous) so that scaled pages display properly rescaled images. But this does not seem to work for old-style zooming on desktop browsers, so a scaling factor greater than 100% there would result in blurry or jagged images.

— Richardguk (talk) 10:34, 23 November 2012 (UTC)

I am using Firefox. I re-set to 100%, and the images seem to be sharper, but the texts are so small that they are not readable. Kdammers (talk) 10:58, 23 November 2012 (UTC)

By "texts", do you mean the article text or the text within the images? Assuming you mean the latter:

Zooming is a bad way to see small image text (as you've discovered!). There are two better ways, depending on how the image is specified:

• If an image is a thumbnail and its size is unspecified in the wikitext (as with the images in List of countries by population), Wikipedia uses a size based on your user Appearance preferences (see "Thumbnail size" under Special:Preferences#mw-prefsection-rendering; I think 220 pixels is the standard width if editors have not changed their user preferences). So you could increase that to something larger.
• However, some images (such as most of those in Gini coefficient) have a particular size specified in the Wikitext (it will say something like [[File:...|123px|...]] to specify the width, [[File:...|x456px|...]] to specify the height, or [[File:...|123x456px|...]] to specify a cap for both, as detailed at Wikipedia:Extended image syntax). If you think the images should be made larger for everyone, simply edit the article text to use a larger number. But avoid widths of more than about 500px, because the images will be too wide for readers with narrow screens.

Hope that helps.

— Richardguk (talk) 14:23, 23 November 2012 (UTC)

Thank You for Your help. I meant the text (e.g., legends) with-in the picture.1) I tried Your suggestion for thumbnails. I set Thumbnail to max, and now the image is blurry but the legend is still minuscule and not readable. 2) I am chary of unilaterally make a blanket size change with-out feedback from other users, in case the problem lies just with me or my set-up. — Preceding unsigned comment added by Kdammers (talk • contribs) 02:04, 24 November 2012 (UTC)

To enlarge an image, don't zoom the browser, click on the image. For the page List of countries by population, clicking the first image takes you here. Click on the image again to get it as large as your browser will allow; if your mouse pointer then changes to a shape like , you can click again to get the maximum enlargement. --Redrose64 (talk) 10:38, 24 November 2012 (UTC)

Assuming that your zoom is now reset to 100%, I think the remaining problem is simply that most of the graphs in the pages you mentioned have been designed for displaying at relatively large sizes (hence with relatively small legend text):

• 

  File:File World population.png

• 

  File:Rank order countries.png

• 

  File:GINIretouchedcolors.png

• 

  File:Economics Gini coefficient2.svg

• 

  File:GiniPlots UK.png

• 

  File:Berg Ostry 2011 Chart 4.gif

SVG format images are better suited to scaling than PNG and GIF, but I don't think that's the main problem in the above cases because the unscaled originals are relatively large.

The article editors probably wanted to avoid cluttering the article with many large images, because readers can always click on the images to see much larger versions on the individual file description pages.

Some of the images would have been better if they had been created with larger legend text (a design issue). But most of the images have detailed graphic content that is inherently unsuited to rendering at small sizes, so the only sensible way to view them is to display them at large sizes in the article or to rely on readers clicking through to the large previews on the file description pages.

In other words, might the remaining problem be that the image text is too small rather than too blurry?

— Richardguk (talk) 11:17, 24 November 2012 (UTC)

File:GINI retouched legend.gif (larger legend; earth tones)

File:GINIretouchedcolors.png (original small legend)

• Autosized images but need larger legends: I have modified article "Gini coefficient" to now wp:autosize the images as "upright=2.20" or "upright=1.40" as scaled 2.20-1.40x times larger than each user's default-image-size setting in Special:Preferences. However, the world-map legend box is still too small to read easily. I am uploading a less-glaring GIF variation of File:GINIretouchedcolors.png with a larger legend box, with the legend border as 3-pixel width, for thumbnail comparisons. The Commons Upload Wizard is very slow, so it will take a long time to upload. More later. -Wikid77 (talk) 16:38, 24 November 2012 (UTC)

As shown in the map images at right, I have uploaded a fast GIF version of the GINI world-map image (3x faster, thinner than the PNG-format image), which now has a larger legend box and bigger lettering enlarged towards 3-pixel letters. However, at typical 220-pixel thumbnail size, the labels were still blurry, so I think that major lettering should use 4-pixel lines to be legible in a typical thumbnail. I worked on map lettering some years ago, so I am just now remembering the issues about using large labels. Anyway, after all these years, I clearly see, rather than the prior push to tediously convert all these images to PNG-format (or SVG) data, there should have been a strong recommendation to use larger labels and lettering in map or graph images. -Wikid77 (talk) 23:52, 24 November 2012 (UTC)

• Expanding letters to 5-pixel width: For the major map label ("GINI Coefficient"), the letter size was increased to use 5-pixel lines in labels with letters separated by 5-pixel gaps, and even that size is barely legible on smaller thumbnails (plan 6-pixel lines/gaps). The effect is that words must be displayed in over-size letters to be legible in thumbnail proportions. Hence, in some cases, the major labels should be superimposed as larger text onto the map, or have a larger-font, closeup image version, or else repeat some of the legend labels inside the caption area of an image. A single map cannot have both large-size text and tiny details. Due to cramped space, I changed the GINI-map legend to 1-column (was 2) and narrowed the map 5% to enlarge labels 5.3% (100/95). -Wikid77 03:54/08:45, 25 November 2012 (UTC)
• Summary of results: We acknowledge that the (thousands of) map/chart labels are blurry in thumbnails, and should often have been redrawn larger, years ago (as a design choice). Meanwhile, zooming the browser is not much help, but a user should right-click blurry charts to enlarge in background while reading an article, then later view each enlarged image-tab to see both the larger labels as well as more chart/map details. Although some large PNG-format maps might grind for almost a minute, many PNG charts/maps will right-click within 25 seconds on a slow-dialup line. Very slow maps could have large-label closeups stored in quick JPEG format (for rapid right-click). When tiny thumbnail maps are spotted in articles, then perhaps edit to wp:autosize charts/maps as 40% larger by "upright=1.40" (adding "frameless" when not "thumb" style images). When even 40% (or 50%) larger does not help, then repeat/recap the tiny labels inside the caption of the map/chart (see new captions in "List of countries by population"). In an ideal world, all over-wide charts would be narrowed 5-20% to show labels as 5.3%-25% larger. Also, all full-screen maps/charts would show major labels in 4-pixel (or 6) letter lines (4-pixel gaps) for legible thumbnails, but beware that very-large labels overpower an enlarged map, so also consider the large-label closeups as 2nd images, or use Template:Superimpose to show large labels (live) over modest labels as needed for the topic showing the map/chart. Large labels are often the enemy which overwrites finely detailed charts, so the choice requires editorial judgment, not unlike planning map use in other-language Wikipedias. Long-term, I think there were also plans to quicken the rendering of PNG images which were displaying in excessive high-precision format, as 3x slower than GIF image thumbnails. Anyway, I think we have enough techniques now to solve many of these blurry graphs. -Wikid77 (talk) 08:45, 25 November 2012 (UTC)

"We acknowledge" - you mean, you acknowledge. I can't see anyone else here concurring with your tl;dr diatribes. — Hex (❝?!❞) 16:18, 25 November 2012 (UTC)

• Thanks for noting the lengthy topics. Well, actually, I was concurring with several other opinions above (across the whole thread), but I do understand your point about all the details, because computer graphics is a highly complex subject, and perhaps there should be a separate Village Pump section just for maps, graphs and photographic-display problems. For example, we have not even mentioned mouse-over enlargement or quick map-legend images, nor legends in alt-text, nor pre-loading of enlarged maps as hidden images to allow instant right-click to show pre-loaded images. Ironically, what I wrote above is less than half of what should be noted about the problems with blurry graphs, to also include guidelines for graph line segments, map symbols, histograms, bar graphs, and standardized templates to ensure larger, legible labels. Anyway, thanks for raising the issue, because I would not want people to think that the segments I wrote above could cover even half of the techniques which we should consider. This is a multi-year problem, with many facets. -Wikid77 23:00, 25 November 2012 (UTC)

I didn't make any point like that at all. Are you replying to me, or to yourself? Out of all your verbiage above, the only thing that has emerged that anyone else has even remotely concurred with is that images displayed at a small size in articles could have larger text. Well excuse me, but no shit, Sherlock. — Hex (❝?!❞) 13:46, 30 November 2012 (UTC)

• You are too kind in claiming I had quickly devised all the above techniques by myself, to make labels more legible, with no help from others in the discussion, as if I were a solitary genius who could think so fast, while others merely advised "have larger text". No instead, upon re-reading the above, it was the others who first noted how browser-zoom is not much help, and right-clicking images not only expands for larger labels but also enlarges crucial map details as well. Plus, others had noted the use of small labels was a "design choice" which emphasized that entire maps should be relabeled, rather than expect Wikipedia to auto-enlarge labels for better clarity. Anyway, I thank you for the compliment, but others here are also very intelligent, whether "Sherlocks" or not. -Wikid77 (talk) 20:33, 2 December 2012 (UTC)

Looking up deleted pages by articleid

The usual way to look up pages by articleid is e.g. http://en.wikipedia.org/w/index.php?curid=37733142. However, this does not work for pages which have been deleted. Is there a way for admins to find out what page an articleid maps to, if the page has been deleted? -- King of ♥ ♦ ♣ ♠ 23:13, 24 November 2012 (UTC)

Nope. In fact, if the page is re-created or undeleted, it gets a new page ID. Graham87 14:18, 25 November 2012 (UTC)

That manual was out of date - article ids should be kept if the page was deleted in mw1.11 or later, but to my knowledge there's no way to access them using that without undeleting them. Anything deleted prior to 1.11 will wind up with a new id, though. -— Isarra ༆ 18:16, 27 November 2012 (UTC)

Nope, unfortunately that's not the case. I've just tested this at User:Graham87/sandbox, which used to have a page ID of 11316888. Now its page ID is 37781837. Graham87 12:04, 29 November 2012 (UTC)

Tracked in Bugzilla
Bug 26123

Graham87 12:34, 29 November 2012 (UTC)

Ach, I should have known better than to expect mediawiki to do what would have made sense based on the schemas... -— Isarra ༆ 23:09, 1 December 2012 (UTC)

Search list not updating

Yet again, the search index has failed to update, this time since 22 November.
Help:Searching#Delay_in_updating_the_search_index says this should be reported here. These backlogs frustrate us WikiGnomes in our tidying up, and we get a huge backlog of spelling, grammar and other mistakes to correct when it is eventually updated. Any idea when it might be updated?
Arjayay (talk) 13:58, 25 November 2012 (UTC)

Two days later, and no change - It still hasn't been updated since 22 November.
Some sort of feedback and indication that something is being done would be appreciated.Arjayay (talk) 09:26, 27 November 2012 (UTC)

Providing an example of an article that doesn't show up would be helpful as a first step. --Malyacko (talk) 08:01, 28 November 2012 (UTC)

I'm not sure what help "an example" is, when the thing is clearly stuck, however:-
My usual check search is this search for recieve (sic). There are about 50 "allowable" misspellings, which cannot be corrected (use of the word in web-site titles, quotations etc.) and there are usually 10-15 new misspellings every day. There are currently (see timestamp) 61 matches, and there have been 61 for the last 5 days. The latest date of any of these matches is 22 November. Unless every editor has suddenly learnt to spell correctly, there should be matches after 22 November - probably 50-75 new ones by now, giving a real total of over 100.
I can also demonstrate the lack of removals, as well as lack of additions. Search results do not always appear in the same order, but somewhere near the top you will currently find Baker Boys (2011 TV series), where I corrected the 3 misspellings of received on 23 November, in this diff [[1]]. However, despite being removed on 23 November, these still appear in the "current" (actually the 22 November) search.
Arjayay (talk) 11:47, 28 November 2012 (UTC)

As another example, a search for 'Yan Tan Tethera' on the Reference Desk [2] finds Wikipedia:Reference desk/Language. The discussion in question has not appeared on RD/L since the transcluded archive page was removed on 26 November 2012 [3]. What it does not find is the transcluded archive page itself which has existed since 24 November [4]. You can see an example where it is finding both for older content e.g. [5] where the transcluded page was created on 21 November [6] and would have been removed from the main RD/L about 24 November.

Or to give another example from the reference desk, it finds the first two questions from November 23 on the science desk [7] [8] but not the next two [9] [10]. You can see Wikipedia:Reference desk/Archives/Science/2012 November 23 for all the questions from November 23. (The transcluded page was created more recently but I'm pretty sure the questions are from when the signed posts are dated [11].)

Based on the above, it seems in case of the science desk the last index was sometimes between 00:36, 23 November 2012 (UTC) and 06:35, 23 November 2012 (UTC). (This is a wider timeframe then I hoped, it should be easy to further narrow if necessary using the reference desk or highly active discussion pages like ANI.)

Nil Einne (talk) 12:48, 28 November 2012 (UTC)

My timings would agree with these, albeit mine are less precise - the Search Index usually updates early hours UTC, thus recording the edits (mistakes) made on 22 Nov, but not the edits (corrections) I made from 12.52 UTC on 23 November. Arjayay (talk) 16:32, 28 November 2012 (UTC)

Just to note that the search list has still not updated. - Arjayay (talk) 08:48, 29 November 2012 (UTC)

A thought. Does the problematic search use data which is held not in the main en.wp data storage, but on a different set of hardware? Toolserver is presently reporting a thirteen-day replag for data from meta:. Separate tests demonstrate that this replag is longer than ten days, because I have 21 edits on meta:, of which the four made on or since 19 November are not being reported by X!'s edit counter. --Redrose64 (talk) 13:57, 29 November 2012 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘Never mind the possible different set of hardware, which I very much doubt has anything to do with the search index problem. The index was updated after updates I made on 22 November, but not after changes I made on 23 November, and this matches Arjaya's observations, so a ten-day delay in some other process seems unlikely to have caused a 6-day problem with the index. But why, after four days, hasn't anyone provided any kind of answer to Arjayay's question? Arjayay is not the only editor who is being stymied by this, and the quality of WP articles is suffering. Is there a better place than this to report the problem? Chris the speller ^yack 04:19, 30 November 2012 (UTC)

Yes, the best place to report it is WP:bugzilla. MBisanz ^talk 04:34, 30 November 2012 (UTC)

The index has finally been updated, so I will not pursue further. Chris the speller ^yack 13:43, 30 November 2012 (UTC)

New feature needs testing

mw:Extension:TemplateSandbox is a new extension that should come in handy for users editing templates (and Lua modules, once that is enabled here); I know I've wished for it several times already. It has now been enabled on test2 and mediawiki.org for testing. The extension has two modes:

• Special:TemplateSandbox allows you to place sandbox templates and Scribunto modules as subpages of some other page, and then view arbitrary pages or wikitext using those sandboxed templates/modules in place of the live versions.
  • Note it doesn't try to use "/sandbox" subpages, because then there would be no way to test sandboxing for just one set of templates without also picking up every other template's "/sandbox" subpage. You can, of course, create a redirect from Special:MyPage/sandbox/Template:Citation needed to Template:Citation needed/sandbox if you want.
• At the bottom of the edit form in the Template and Module namespaces, a new box will appear that allows you to preview other pages as they would appear if you saved the template/module you are currently editing. For example, you could open Template:Asbox and make some changes in the edit form, then enter "Sharpsville Area School District" in the preview box to see how Sharpsville Area School District would look with your changes.

Please report any bugs, either here or in Bugzilla (convenience link). Thanks. Anomie⚔ 21:58, 26 November 2012 (UTC)

I hope more people try this tool because it's pretty useful for anyone who has to edit templates. Go ahead and try it on mediawiki.org: hit Edit on the sample "thank-you" template. On the edit form, under the Save button, you'll see "Preview page with this template" -- try User_talk:Example_user to see how your changes would look. Or, go to mw:Special:TemplateSandbox and give it the "sandbox prefix" User:Sharihareswara (WMF)/sandbox and tell it to "render page" User_talk:Example_user . You'll see the ugly changes I made to the template in my sandbox. Sumana Harihareswara, Wikimedia Foundation Engineering Community Manager (talk) 05:25, 29 November 2012 (UTC)

Update: It's now available on most of the non-Wikipedia WMF wikis (any that are on 1.21wmf5). Barring any last-minute changes of plan, it will be available here on Monday when 1.21wmf5 is deployed here. Anomie⚔ 20:04, 29 November 2012 (UTC)

Can't go directly to feedback pages

Tracked in Bugzilla
Bug 42479

Windows 7, IE 8, Monobook. Ever since the "View reader feedback" button began appearing on talk pages, I've had a problem with it. If I leftclick it while holding down Ctrl, it opens in a new tab just like any other link would, and I can read the feedback just like I should. However, if I leftclick it while not holding down anything, the page just sits there as if I'd not done anything, rather than going to the feedback page. Any idea what's going on, and what (if anything) can be done? Nyttend (talk) 02:48, 27 November 2012 (UTC)

Most pages don't have the link so here is an example: Talk:Google. For me in IE9, the first left click doesn't work but a second left click works. I don't mean a double-click. The two clicks can be long apart. It's the same logged out and logged in with Vector. PrimeHunter (talk) 04:28, 27 November 2012 (UTC)

The first single-left-click works fine for me on Firefox 17.0/Vista/Vector. Switching to IE9 produces the same behavior as what PrimeHunter reported. Sounds like an IE problem. jcgoble3 (talk) 05:01, 27 November 2012 (UTC)

Opera has the same problem as IE9. The first click works in Firefox, Safari and Chrome. PrimeHunter (talk) 05:10, 27 November 2012 (UTC)

Do you use Internet Explorer's "Compatibility Mode"? --Malyacko (talk) 12:27, 27 November 2012 (UTC)

In IE9 the feedback link behaves the same with and without compatibility mode. However, I noticed that with compatibility mode enabled I cannot change tab in preferences (MonoBook and Vector tested) unless I open the preferences tab in a new browser tab or window. Two clicks don't work there. Without compatibility mode a single click on the tab works fine. PrimeHunter (talk) 13:11, 27 November 2012 (UTC)

Your bug in IE "compatibility" mode with the preferences is already fixed in 1.21wmf5, which is scheduled to be deployed here Monday. You can test it now on test.wikipedia.org or mediawiki.org if you'd like. The AFT bug is now in bugzilla as bug 42479http://bugzilla.wikimedia.org/show_bug.cgi?id=42479. Anomie⚔ 16:56, 27 November 2012 (UTC)

(added after collision) your 2nd concern (preferences) is reported in Bugzilla:41792. is fixed in 1.21wmf5, which, according to the roadmap, will be deployed to enwiki on December 3rd. you can test the fix now on mw:Main page - mediawiki is already on wmf5.

as to the problem with the feedback: can you please turn on the "Display notification about every script error" settings in IE9 (under Tools => Internet Options => Advanced") and see if you can get a more specific error report? it also may help if you add to the tail of the address line of the browser either "?debug=1" or "&debug=1" (depending whether or not a question mark already appears earlier on the address line). as to Bugzilla:42479: this may or may not be related to the issue you see - i could not decipher this bug report. peace - קיפודנחש (talk) 17:07, 27 November 2012 (UTC)

Neither of those ("Display notification about every script error" or "?debug=1") produced any errors at all. jcgoble3 (talk) 19:14, 27 November 2012 (UTC)

after some troubles, i was able to reproduce. at least for me, the behavior is more interesting: when pressing the control for the 1st time it does not respond, and then on a 2nd click (you have to take enough times between the 2 clicks as to not make it a double-click), it goes to the feedback page. this behavior is not exactly what the other 2 editors reported, but for me, at least, it is 100% consistent, at least in Talk:Google. i did not fully understand Bugzilla:42479, but the little i did, it looks like the right diagnosis. קיפודנחש (talk) 22:28, 27 November 2012 (UTC)

I have flagged this report to Oliver and Fabrice from that team. Philippe Beaudette, Wikimedia Foundation (talk) 08:45, 2 December 2012 (UTC)

Thanks, PB :). Nyttend et al, thanks for flagging this for us: I'm glad to see that the devs are on top of it already. If you encounter further bugs with the Article Feedback Tool, the AFT talkpage or my talkpage are always open :). Okeyes (WMF) (talk) 13:50, 2 December 2012 (UTC)

Still have location map problem

Per HTML 5 snafu - pushpin points moved south, there was a problem, and it reads as if there has been a solution, at least form those using the template Location Map. I see suggestions is may take some time for fixes to propagate, but that was in September. Some of the locations on this map are clearly wrong. Spokane and Bridgeport look OK, but the other three are wrong. Am I missing something, or is this supposed to be fine now? (FireFox) --SPhilbrick(Talk) 18:08, 27 November 2012 (UTC)

Nothing to do with that. The linebreaks were adding extra height to the thumbnail frame, causing the relative dimensions between image size and image frame not to match. The pogs are positioned relative in the image frame, so if they don't match, the pogs are in the wrong place. —TheDJ (talk • contribs) 20:13, 27 November 2012 (UTC)

Thanks. That looks better.--SPhilbrick(Talk) 21:47, 27 November 2012 (UTC)

• Fixes needed to support MSIE browsers: Although the map markers are fine with the Firefox browser, there are still severe dot-dislocation problems (off 3%-7%), in the placement of many markers, for the Internet Explorer browsers, but I think other people gave up trying while thinking the problem would get fixed by "expert" editors. For example, College Station (TX) marks as south of Galveston (i.e., underwater), and Baton Rouge floated down the Mississippi, alongside cruiseships in the Gulf of Mexico. If it were a map of Britain, I suspect London would mark in Dunkirk. For the fix, I got as far as creating Template:Location_map/simple, which works for only 1 marker. However, while the old versions of Template:Location_map_many still map correctly, that template was also trashed to use {Location_map+}, likely with the same peculiar, complicated div-section shift, so I will work to fix that as well. Meanwhile, I have always wanted to add better map features, so rather than just fix the "dislocated dots", I will add some new features to make the map markers shorter to write. There are just so many other problems, and new templates, and so it has been difficult to update all of them in a few months. However, the fixes have been made at a rapid pace, so we can solve the dislocated-dot problem soon. -Wikid77 (talk) 16:35, 28 November 2012 (UTC)
• Ah IE, quelle surprise... Some version info might be useful... Not that I own Windows, so it's up to someone else, but logging information like that is good practice. —TheDJ (talk • contribs) 13:05, 30 November 2012 (UTC)
• Firefox (etc.) needs "line-height:0" to match IE: Maps can be displayed to fit any browser, but the div-section attribute "line-height:0" is needed because Firefox or related browsers allow line-height to override <div> "height:0;" and have placed a map marker in the middle of a tall character box, which lowers the marker by 5-9 pixels. This is a very obscure incompatibility between browsers (as well as only a few pixels lower), and so that is why it has gone unfixed for years. However, newer Template:Location_map~ is way off in IE alignment. I think the alignment problems occur in IE 6, IE 7, IE 8 and IE 9, but I am not sure of the exact version numbers for each one. Anyway, I have created Template:Location_map_all to work on all browsers, but also provide new mapping features to expand our mapping technology, and simplify alignment of map labels for better legibility in thumbnail maps/graphs. More later. -Wikid77 (talk) 15:47, 1 December 2012 (UTC)

Size of breaks / Source display problem

Hi. Some month ago something with the size of breaks was changed across Wikimedia projects, that makes a lot of my Wikiversity articles unusable. (I did not get any answer in the Wikiversity, so I ask here.) When I created my pages it was possible to make square matrices using graphics of the single rows. It looked like this. Now it looks like this:

See this page as an example. Is there any way to fix this problem? Greetings, Lipedia (talk) 11:04, 28 November 2012 (UTC)

Here is your code in <p style="line-height:0px;">...</p>:

It looks fine to me in Firefox. I don't know whether there is a risk of images on top of each other in other browsers or circumstances. The images have height 5px, but line-height:5px; gives space between them for me:

PrimeHunter (talk) 13:01, 28 November 2012 (UTC)

That worked, thank you! By the way: There is another strange problem. In the "Demonstration in Matlab" box at the end of this page is only a part of what I wrote between the source tags. Do you know any solution? Lipedia (talk) 17:06, 28 November 2012 (UTC)

Experimentation indicates it happens when there are more than ten consecutive lines of numbers in the assignments. If the 11th line only has a few numbers then the display works. Maybe there is a 1024-character limit for assigment operations in source tags, or something like that (just guessing). PrimeHunter (talk) 17:39, 28 November 2012 (UTC)

Something definitely has a limit around 1000. 994 characters witn no newlines works here:

x =
1      010       020       030       040       050       060       070       080       090       100       110       120       130       140       150       160       170       180       190       200       210       220       230       240       250       260       270       280       290       300       310       320       330       340       350       360       370       380       390       400       410       420       430       440       450       460       470       480       490       500       510       520       530       540       550       560       570       580       590       600       610       620       630       640       650       660       670       680       690       700       710       720       730       740       750       760       770       780       790       800       810       820       830       840       850       860       870       880       890       900       910       920       930       940       950       960       970       980       9901234

995 characters fails:

bugzilla:41719 is "Code display fails if more than 986 characters of comma-separated values are between curly brackets". The example is User:Camusensei/sandbox. PrimeHunter (talk) 18:36, 28 November 2012 (UTC)

Ya dun' broke the page boy.--The Devil's Advocate (talk) 23:27, 3 December 2012 (UTC)

LaTeX packages

The LaTeX code for this should compile in math mode. The PNG is blurry, especially in Microsoft Internet Explorer.

Mediawiki software apparently allows LaTeX packages to be used. I want to be able to write something like this and have it compile:

Kiefer.Wolfowitz 13:15, 28 November 2012 (UTC)

The software loads the packages; we as editors have no control over it. You can file a bugzilla request to see if they could implement this. (I don't see it as likely as this is a rather limited functional group and because the devs are either volunteers who probably wouldn't care enough to implement this or paid and have other higher priority assignments.) An image of the output of a local file seems to work plausibly, and if you cite the packages and code used in the image, people can recreate it as necessary. --Izno (talk) 15:51, 28 November 2012 (UTC)

Wikimedia software has the ability to run LaTeX packages, which ship with e.g. Ubuntu.

MediaWiki's example of a LaTeX package fails to compile on English Wikipedia:

English Wikipedia has not flipped the switch that would allow us to use LaTeX packages. Kiefer.Wolfowitz 17:17, 28 November 2012 (UTC)

Then you need to pursue that via Bugzilla, as I already indicated, if it is indeed only to turn the switch on. --Izno (talk) 22:17, 28 November 2012 (UTC)

"Wikimedia > Extension setup" in bugzilla.wikimedia.org, to be exact. :) --Malyacko (talk) 11:55, 29 November 2012 (UTC)

MediaWiki does not allow latex, it allows Math written in a subset of latex. In theory we could squash more into the math tag, but realistically that is not going to happen for the vanilla version of MediaWiki. The tip to add more packages to the rendering can be seen as a 'hack'. But of course we could create yet something else that would allow it. Someone is free to develop a .tex file renderer, it wouldn't be too hard I think if you put some time in it. The biggest problem with any renders is getting it reviewed for security and strengthening it against DoS use. —TheDJ (talk • contribs) 12:59, 30 November 2012 (UTC)

Hi TheDJ!

Thanks for your comment. I would love to have the AMS packages, geometry, and gchords. Otherwise, I shall use GNU Lilypond to create SVG images. Sincerely, Kiefer.Wolfowitz 13:23, 30 November 2012 (UTC)

Thanks to whomever reformatted my example! Kiefer.Wolfowitz 21:38, 3 December 2012 (UTC)

Article Feedback Tool needs sorting out

Can we do something about the ugly, imposing Article Feedback Tool? It clutters up the body of articles making pages less useful for readers. Given how little it is is used, even in such a prominent position, it should be relegated to a link in the toolbox section of pages - if it is to be retained at all. It seems to be a bit of a waste of time IMHO. Note that logged in editors may not see it since it can be turned of in user preferences. -- Alan Liefting (talk - contribs) 19:26, 28 November 2012 (UTC)

The point of the tool is to be used to provide feedback, and it is primarily aimed at users whom are either unwilling or unknowledgable about how to edit. Putting it in the sidebar hence risks not targeting those users and instead targeting experienced editors... who already know how to turn it off on top of that. Also, your "given how little it is used" needs a [citation needed]. --Izno (talk) 22:15, 28 November 2012 (UTC)

Your comments might be better heard by posting them at https://en.wikipedia.org/wiki/Wikipedia_talk:Article_feedback --Malyacko (talk) 11:54, 29 November 2012 (UTC)

Portrait of Père Tanguy#Julien Tanguy and Talk:Portrait of Père Tanguy

Hello,

I am mainly working on fr:Le père Tanguy. And I try to add an information on wp:en. Could you replace that in order ? Sorry for the disagreement. Many thanks Mike Coppolano (talk) 09:21, 29 November 2012 (UTC)

I have fixed the cite error on the page. Does that fix your problem? Keith D (talk) 14:14, 29 November 2012 (UTC)

Many thanks, Keith D. One more : could you a the end of Julien Tanguy bio, insert that text : Octave Mibeau writed a notice [with the ref I've given >>> on Gallica/ see Julien Tanguy talk page] ? Thanks. Mike Coppolano (talk) 17:15, 29 November 2012 (UTC)

I have made this change for you to add the information you requested. Keith D (talk) 19:17, 29 November 2012 (UTC)

Thanks Mike Coppolano (talk) 03:41, 30 November 2012 (UTC)

Embed SVG

I have created an Interactive SVG linkmap of New York where you can hover over a county and click on it to bring you to that county's Wikipedia article. Is there any way I can embed the SVG(not a bitmap of the SVG, the SVG itself) in a page or template? I have tested it on Firefox and Chrome and it works in both.—Kelvinsong (talk) 20:28, 29 November 2012 (UTC)

Ah, but have you tried it in Internet Explorer? If it doesn't work there, it's not really viable. I have tried viewing self-created SVG files directly - four (Chrome, Firefox, Opera, Safari) display it just fine; one (IE7) refuses. --Redrose64 (talk) 22:37, 29 November 2012 (UTC)

Must we always support ancient browsers? Adding these maps to articles doesn't hurt IE 7/8 users, and benefits users of other browsers. It also helps encourage people to switch to free browsers. —Kelvinsong (talk) 22:53, 29 November 2012 (UTC)

I know it is frustrating, but if the reader is using Wikipedia in a public library with computers installed in 2009, they might still have "ancient" Windows 7 with IE 7, and not aware they need to feed corporate planned obsolescence by installing new computers every 4 months. -Wikid77 (talk) 06:35, 2 December 2012 (UTC)

As I noted on the Help Desk: Browser support for SVG is spotty, so the MediaWiki software renders them on a page as PNG. You would have to click through to the image description page then to the actual SVG to get the linking. --— Gadget850 (Ed) ^talk 12:58, 30 November 2012 (UTC)

The image description page can be bypassed with Media: which links directly to the uploaded file, for example Media:Tokyo subway map en jp.svg versus File:Tokyo subway map en jp.svg. But the svg itself cannot be embedded in a wiki page, and Media: without a File: link doesn't provide attribution as usually required by the image license. PrimeHunter (talk) 14:33, 30 November 2012 (UTC)

Attribution isn't a problem—I made the file. Image maps sound like a big compromise here—they give no hover indication, and can be very crude—see the map on this page, where you can be hovering over Connecticut and be taken to the Massachusetts page, and small states like Rhode Island are unclickable, whereas interactive svgs are highly precise because the vector map and the link map are exactly the same. I'd also like to add that IE explorer 8 is only used by ~11% of users.

Also, the Commons won't let me upload the interactive SVG—it says the file is corrupt.—Kelvinsong (talk) 21:03, 30 November 2012 (UTC)

Writing for Wikipedia is harder than writing a trivial iPhone app, and must consider the reach to billions of people, not just a few million here or there. Those people using IE 8 are part of the "anyone who can edit" the free encyclopedia. -Wikid77 (talk) 06:35, 2 December 2012 (UTC)

As a workaround, the <imagemap> extension tag can be used to add overlay links to images. — Richardguk (talk) 13:32, 30 November 2012 (UTC)

Yes, the <imagemap> tag works well with Internet Explorer browsers, and detailed maps can be kept accurate to the exact pixel by setting a <div> "line-height:0" (for Firefox, etc.) when overlaying div-section data onto an image. When a reader can click your map in a hospital, hotel computer room, Internet cafe, public library, or other third-world browser, then that is an exciting accomplishment. Consider the reading of articles with ancient 2009 computers as part of Wikipedia's mission. -Wikid77 (talk) 06:35, 2 December 2012 (UTC)

Note: You can use {{xtag}} to auto-link parser and extension tags. {{xtag|imagemap}} → <imagemap> --— Gadget850 (Ed) ^talk 11:25, 2 December 2012 (UTC)

Stub article template bug

I found a deficiency in Template:Asbox resulting in a recurring grammatical error. The formatting of the template has led to a rather humorous error on, for instance, Template:US-poet-1900s-stub: This American poet-related article born between 1900-1909 is a stub should read as This article on an American poet born between 1900-1909 is a stub, unless of course the article was born between 1900-1909! elvenscout742 (talk) 08:34, 30 November 2012 (UTC)

The documentation at Template:Asbox is clear enough. I have edited Template:US-poet-1900s-stub to improve the wording. How many others need this fix? -- John of Reading (talk) 09:36, 30 November 2012 (UTC)

All US poet stub templates had the same problem. I have fixed them all but not examined other stub templates. PrimeHunter (talk) 16:22, 30 November 2012 (UTC)

Inherited protection template?

Where is {{bug}} getting {{pp-template}} from? I thought it would be the /doc subpage, but it's not. — Hex (❝?!❞) 13:35, 30 November 2012 (UTC)

It is in the documentation, but is coded into {{Documentation/start box}}, which is one of the templates used for building the green box. --Redrose64 (talk) 13:43, 30 November 2012 (UTC)

Oh, of course. Can't believe I missed that. Also, why didn't I just read the Documentation documentation? Nice one, thanks. — Hex (❝?!❞) 13:59, 30 November 2012 (UTC)

Special:WhatLinksHere/Too_Close

There are still more than 30 links to Too Close, now a disambiguation. However, I have changed links in Template namespaces, so I wonder when it will upload a number of links. --George Ho (talk) 15:13, 30 November 2012 (UTC)

Special:WhatLinksHere is likely to remain lagged for a period of days; if you want to force it to update, you can purge the pages that are shown as linking to Too Close but do not, though there is no particular need to. - Jarry1250 ^{[Deliberation needed]} 16:00, 30 November 2012 (UTC)

Actually, purge only updates the purged page itself. It requires a null edit to also update link tables. All the articles I examined at Special:WhatLinksHere/Too Close had already updated the article page, but being in WhatLinksHere means that the link table had not been updated. Help:Job queue#Updating links tables when a template changes gives me the impression that the automatic updating made by the job queue will update the article page and link tables at the same time. Is this false? The link in the navigation templates on the articles were updated by George Ho around 14 hours ago.[12] PrimeHunter (talk) 16:19, 30 November 2012 (UTC)

The job queue is currently showing 88663 so it could be a few hours. --Redrose64 (talk) 17:31, 30 November 2012 (UTC)

Usually fixing the template links will fix the majority of incoming links. I might advise watching the page and returning to it tomorrow or the day after to confirm that there are not extraordinary numbers of links remaining. --Izno (talk) 18:58, 30 November 2012 (UTC)

I was bored, so I null-edited every page on the list. That should have taken care of template links. Those left will need to be checked individually. jcgoble3 (talk) 20:21, 30 November 2012 (UTC)

VisualEditor fortnightly update - 2012-11-26 (MW 1.21wmf5)

Hey all,

Below is a copy of the regular update for the past fortnight on the VisualEditor project so that you all know what is happening (and make sure you have as much opportunity to tell us when we're wrong, as well as help guide the priorities for development and improvement).

The VisualEditor was updated as part of the wider MediaWiki 1.21wmf5 branch deployment on Monday 26 November.

In two weeks since 1.21wmf4, the team have spent their time mostly working on finalising the code and preparing for its deployment in December as a test for users.

A lot of work went into the 'Data Model' component that converts between the information that mw:Parsoid gives us and a structure that the code can edit. This now supports HTML entities (so "&ocirc;", "&#xf4;", and "ô" do not get switched when users don't expect it - 42118), HTML comments (so they don't get accidentally removed - 42124; viewing and editing them is not yet supported), better handling for content that we don't yet recognise (42119), fixing a bug with our integration with Parsoid (42121), and supporting in production the "Change Markers" code that was worked on last iteration (41947).

In the integration work, we removed the "Feedback" link as this would not work well with the code being deployed to multiple wikis (41722) and unfortunately had to add Internet Explorer temporarily to the "blacklist" of browsers that the VisualEditor will not support for December, due to a number of critical issues that the team does not have time to fix (42335) - post-December, the team will work to find a way around the various bugs and inconsistencies in Internet Explorer.

We fixed bugs on handling "alien nodes" (items of content that we don't have a specific handler for yet, like templates), such as the floated alien-covering "phantoms" for selection purposes appearing wrongly in Firefox (42177) and them not appearing at all for items that floated due to a class in Chrome 42134). There was also some nasty bugs that we fixed, including when editing around inline (mid-paragraph) alien nodes (42212) and the data model and the display getting out of sync sometimes when cutting-and-pasting (42219).

Finally, a great deal of work was done to overhaul the link inspector's code and quash a large number of minor bugs with it, such as wrongly replacing links's spaces with underscores (42140).

A complete list of individual code commits is available in the 1.21/wmf5 changelog, and all Bugzilla bugs closed in this period on Bugzilla's list.

Hope this is helpful! As always, feedback gratefully received, either here or on the centralised feedback page.

Jdforrester (WMF) (talk) 20:37, 30 November 2012 (UTC)

"Interaction" drop down menu in Wikipedia tool box

The "Interaction" drop down menu on the left hand tool box section keeps closing up. Every time I go to a new page, I have to open it up. This just started happening today. 216.93.234.239 (talk) 00:22, 1 December 2012 (UTC)

Try to clear your entire cache. PrimeHunter (talk) 01:15, 1 December 2012 (UTC)

It didn't solve the problem. 216.93.234.239 (talk) 05:12, 4 December 2012 (UTC)

The box is supposed to remember its latest state and begin there on a new page. It works for me. What is your browser and version? Did something with your browser change when it started? PrimeHunter (talk) 05:32, 4 December 2012 (UTC)

Firefox 17.0, Windows 8. It's a new laptop that I've only been using for a week, but I swear the old way of things was like that when I first came here with the new computer. 216.93.234.239 (talk) 05:54, 4 December 2012 (UTC)

Changes needed to protection interface

Pending changes went back into use a few hours ago. One of the things those of us who have been working on it forgot to do was to figure out how to remove the two warnings on the protection interface which advise admins not to use it. Anyone here willing/able to look into that? I don't know a thing about mediawiki pages or even how to find them in the first place. Beeblebrox (talk) 02:23, 1 December 2012 (UTC)

Can you tell me what I need to do to see the text that you want changed? Like "click this page and then try to..." MBisanz ^talk 02:32, 1 December 2012 (UTC)

(ec) Oh, I found MediaWiki:Flaggedrevs-protect-legend. Don't know if that's what you're looking for. And the horrible grey box at the top is controlled by MediaWiki:Protect-text. HJ Mitchell | Penny for your thoughts? 02:34, 1 December 2012 (UTC)

I think Steel did this a few minutes ago [13]. --Bongwarrior (talk) 02:36, 1 December 2012 (UTC)

Yep, it is Fixed now. Beeblebrox (talk) 02:39, 1 December 2012 (UTC)

Readability score?

The latest issue of Signpost had an item about a paper on research comparing readability scores of WP, simple and Britanica. Other than pasting an article into a word processor and asking for a readability report, is there an an easy way to get a readability score? RJFJR (talk) 02:59, 1 December 2012 (UTC)

There are some automated tools (I don't recall where, off-hand), but I generally wouldn't bother. There are so many additional considerations that need to be taken into account when considering "readability" that an automated tool (and most of these papers' analyses) simply don't – or can't – take into consideration. – Philosopher ^{Let us reason together.} 13:15, 1 December 2012 (UTC)

Looks like my comment is a readability violation of its own - way too many "consideration"s there. – Philosopher ^{Let us reason together.} 13:21, 1 December 2012 (UTC)

I took down my readability tool as I realized it is more harmful then helpful. Those readability algorithms are a horribly crude and papers failing to mention selection bias and the wide standard deviation. Its inclusion in word processing software is because of the contractual requirements.

What I'd like to see is dictionary based analyses such as a time-to-read (via TTS) by section and word occurrences by topic (that is an adaptive spell check). — Dispenser 23:06, 1 December 2012 (UTC)

Somebody could probably write up a frontend to hook up with GNU style. -- King of ♥ ♦ ♣ ♠ 08:07, 2 December 2012 (UTC)

Statistics

How often and when this page update [14]? Xaris333 (talk) 14:23, 1 December 2012 (UTC)

• Edit-count stats updated mid-month: The statistics are typically updated near, or after, the middle of each month for the English Wikipedia counts:

http://stats.wikimedia.org/EN/TablesWikipediaEN.htm - tables of article edits, talk-page and article counts

The delay in processing data from the prior month has been attributed to a 10-day span to tally the counts; however, there were plans to change to an incremental-count algorithm, which would allow the prior month to be tallied much faster than a 10-day delay. -Wikid77 (talk) 07:04, 2 December 2012 (UTC)

Infobox not showing references in edit mode

I have a query relating to the difference between view and edit for an infobox. Please see thread at [15]. Eldumpo (talk) 19:48, 1 December 2012 (UTC)

Dependency issues

I've notice an oddity in the re-generation of articles when a template (or indeed other page) on which they depend is altered. I understand that when a page is 'touched' (edited or purged mostly), the timestamp of the touch is recorded and all pages dependent on it added to a queue to be rebuilt as resources permit.

Recently, I modified Template:Universities in Taiwan, and can see the page_touched value (via the toolserver) for the article is now 2021129230858. On checking, many dependant articles did not reflect the change, even after 48 hours. Manually performing a null edit on a dependent article does cause it to show my change. Looking at the page_touched value of the seemingly un-updated dependents shows them to now have the same value as the edited template - 2021129230858. See [Pages that link to "Category:Universities and colleges in Taiwan" for some example titles.

I suspect there is a race condition causing dependent pages to be re-evaluated before the change they are being updated to include has been fully committed and/or propagated.

While I would normally raise an issue like this with the MediaWiki bods, my own reading of the relevant source shows no obvious problems, and I cannot re-create the problem on an admittedly very basic test-bed. My suspicion is that the race condition is unique to (or more often triggered within) the more complex setup that serves the live Wikipedia - replication within MySQL or Memcached would be likely candidates.

Can anyone here sanity-check my findings and/or point me to an existing discussion of this phenomenon before I bother the nice people at Wikimedia please? - TB (talk) 20:18, 1 December 2012 (UTC)

Are you talking about the job queue? Graham87 07:23, 2 December 2012 (UTC)

Tracked in Bugzilla
Bug 42614

• Why Job queue seems to require 5 days to reformat 2,000 pages: It seems illogical that the job queue would actually be so slow as to require 5 days to reformat 2,000 articles which use a recently-changed template. Formerly, a template could be changed, and over 400 dependent pages would reformat within 1 hour 10 minutes. When Template:Rnd (numeric rounding) was updated on 5 April 2010, then over 306,000 dependent pages were reformatted within 4 hours, leaving only 1,196 as "stuck" pages. Is there a database race condition which prematurely reformats many of those dependent pages as already "touch'ed" (updated), so the reformatting with the new template revision does not occur (unless forced by a null edit)? This problem has persisted since years ago, when I changed a navbox template used in 23 articles about Vienna, and those 23 articles took days, not minutes, to reformat. -Wikid77 (talk) 08:00/08:06, 2 December, revised 06:06, 3 December 2012 (UTC)

Yes, I am aware of the job queue, but can find no flaw in the way it is implemented by either inspection or testing. Variously through Wikipedia's history we have experienced problems with the timely regeneration of dependent pages for a variety of reasons but there has generally been an identifiable issue. This one eludes me. - TB (talk) 08:32, 2 December 2012 (UTC)

• Run experiments to test race-condition hypothesis: Because everyone is busy and might not have time to respond here with full explanations of current job-queue logistics, we should meanwhile run experiments to test the distributed-database, race-condition hypothesis. Perhaps update a template used in a few dozen articles, then wait a few hours and "count" (by representative sample) the ratio of done-to-not-done with reformatting. Then after several hours (perhaps 12), edit the same template again with a new change, and according to the race-condition hypothesis, then all dependent articles should be reformatted within a few hours with the prior-revision template (a sample of size 30 would show 0 non-reformatted). Some weeks ago, I was sensing that articles were deliberately non-reformatted (pseudo-touched) to reduce the size of the job queue. If every article containing gobs of navspam navboxes were actually reformatted as expected, perhaps the job queue would be delayed by weeks, rather than days to reformat pages using a shared template. Remember some off-topic or remote-tangent navboxes are spammed into over 25,000 articles, none of which are listed in the mega-spammed navbox. I predicted this artificially-created, navspam resource-hog problem in "wp:Overlink crisis" which has been bitterly denied by some editors; however, there have been many articles with more than a dozen bottom navboxes, most of them remote-tangent navboxes. It is part of a wider problem of wp:Datahoarding in articles, where navboxes are only a part of the problem, such as showing 2 entire climate-table boxes for the same town, as if one climate-table was not close enough for temperatures/precipitation on the other side of a small town. -Wikid77 17:27, 2 December 2012 (UTC)
• Re-edit refuted race-condition hypothesis as still no reformat: As I suspected, a test to re-edit a template, 16 hours later still failed to trigger quick reformatting with recent revisions of the template. There was no evidence to support the concept that a recent edit, to a shared template, causes the dependent pages to run a race condition and reformat with the prior-revision template (before the new revision is fully saved); no, instead, even when a template has been edited with an obvious change, and then re-edited 16 hours later with another obvious change, neither edit causes any of the dependent pages to reformat within 2 hours, as has been the case in prior years. Rather than trigger an immediate reformat with a prior-revision template, the reformatting is sporadic, where some articles are still not yet reformatted within 28 hours after the initial change to the shared template. I had suspected the race-condition hypothesis, of pages imagined to quickly reformat with prior-revision templates, was incorrect due to the widespread failure of any (repeat: any) articles to reflect recent template edits, even when templates have been re-edited many hours later. The implications are clear: when a recent template edit must be reflected in related pages, then perform the edit many days in advance, or else crank a script that issues a null-edit (no-change edit) on every page which is needed soon. When a template is changed, it might be days before the dependent pages are reformatted, and some pages might require days longer than others in the list to reformat, as in prior years. -Wikid77 (talk) 23:15, 3 December 2012 (UTC)

I agree with these findings - it isn't a race condition, it's a plain old failure to re-evaluate dependent pages. The fault still fails to manifest on a simple test-bed though. Some optimisation feature on the live Wikipedia environment misbehaving perhaps ? - TB (talk) 09:27, 4 December 2012 (UTC)

My JS broke

It now gives me an error "undefined reference: addOnloadHook". Any idea how to fix this? -- Liliana-60 (talk) 21:20, 1 December 2012 (UTC)

We need a bit more information, like what browser you use and on what page this happens. — Edokter (talk) — 13:25, 2 December 2012 (UTC)

Well I use Opera, and it seems to happen on every page, as all JS-specific features like sortable tables and such are gone. -- Liliana-60 (talk) 20:01, 2 December 2012 (UTC)

The error seems to be caused by the script you load from wikt:User:Liliana-60/monobook.js. — Edokter (talk) — 21:59, 2 December 2012 (UTC)

I figured as much. But how can it be fixed? -- Liliana-60 (talk) 08:10, 3 December 2012 (UTC)

Okay. This doesn't seem to be because of my JS, that is just one symptom. It happens on Commons as well, where I don't actually use my JS. The error message is:

[03.12.2012 14:28:52] JavaScript - http://commons.wikimedia.org/wiki/Commons:Deletion_requests/2012/12/03
Inline script thread
Uncaught exception: TypeError: Cannot convert 'mw.util' to object
Error thrown at line 1, column 0 in http://bits.wikimedia.org/commons.wikimedia.org/load.php?debug=false&lang=en&modules=site&only=scripts&skin=monobook&*:
    window.addPortletLink=mw.util.addPortletLink;

So it seems that the MediaWiki JS broke. -- Liliana-60 (talk) 13:32, 3 December 2012 (UTC)

Cannot reproduce on http://commons.wikimedia.org/wiki/Commons:Deletion_requests/2012/12/03 or http://commons.wikimedia.org/wiki/Main_Page with Firefox's Web Console - I get two JS warnings about getAttributeNode() but that's all. --Malyacko (talk) 20:58, 3 December 2012 (UTC)

How do I code this URL?

For the winners of the European Film Academy awards, how do I code http://www.europeanfilmacademy.org/News-detail.155.0.html?&tx_ttnews[tt_news]=88&cHash=b859c57aafcfa00d80f829301a1e03ec this URL so it doesn't break in the references section? It has those pesky brackets around the tt_news part of the link. The example is cite #22 in the article for Amour. Thanks. Lugnuts ^{Dick Laurent is dead} 11:00, 2 December 2012 (UTC)

See Help:URL#Fixing links with unsupported characters for encoding. Also documented at Template:Cite web#URL. --— Gadget850 (Ed) ^talk 11:20, 2 December 2012 (UTC)

Brilliant - that's fixed it. Thanks! Lugnuts ^{Dick Laurent is dead} 11:23, 2 December 2012 (UTC)

How do search for my new pages?

Is there a way that I can search for new pages that I have created? I don't really want to plough through zillions of items in my contributions to find those that start with N, and I can't help thinking there should be a simple fix, thanks Jimfbleak - talk to me? 13:27, 2 December 2012 (UTC)

Yes, at the bottom of your contribs page you should find a box containing the link Articles created. By default, that is for articles only and also excludes redirects, but you can adjust it for other namespaces by altering the number in &namespace=0 to any one of these. An interactive version is here. --Redrose64 (talk) 14:00, 2 December 2012 (UTC)

Many thanks Jimfbleak - talk to me? 14:11, 2 December 2012 (UTC)

ΒΟΤ

Is it easy to create a BOT? If yes, how can i create one? Xaris333 (talk) 20:10, 2 December 2012 (UTC)

Wikipedia:Creating a bot is a good place to start for information regarding how to create a bot. --J36miles (talk) 20:43, 2 December 2012 (UTC)

I have read ti before but i didn't understand. Xaris333 (talk) 20:57, 2 December 2012 (UTC)

It can be easy to create a bot, and it can also be very difficult. It mainly depends on the task, then on which tool you choose, and of course your technical and programming experience. First step is deciding what task you wish to do though. WP:CREATEBOT is as good of a guide as it gets, and you need to give us some specifics of what exactly it is that you want to do for more concrete suggestions. —  HELLKNOWZ  ▎TALK 21:07, 2 December 2012 (UTC)

WikiLove customization

Hi, I'm working on a WP:Teahouse pilot program with badges for recognition. We're using WikiLove as a delivery mechanism. I'm not sure if it's just a caching issue, but I'm not seeing our preloaded badges show up in WikiLove. The install script we're using is: importScript("User:Ocaasi/WikiLoveinstallscript.js"); ...which imports User:Ocaasi/WikiLoveinstallscript.js to a user's common.js page. It would be awesome if someone could take a peek at the code and see if I'm missing something. Thanks! Ocaasi ^{t | c} 21:07, 2 December 2012 (UTC)

FixedMust have been a caching issue. It's showing up fine now. Ocaasi ^{t | c} 22:56, 2 December 2012 (UTC)

Help

Is there a way to find with articles than are in Category:Football in Spain are not in the Category:Football in Spain task force articles (the Template:WikiProject Football don't have the parameter Spain). (Sorry about my english. I am trying to try this in greek wikipedeia). Xaris333 (talk) 21:43, 2 December 2012 (UTC)

Yes— see Wikipedia:CatScan --— Gadget850 (Ed) ^talk 22:22, 2 December 2012 (UTC)

Doen't work. The first category has articles and the second talk pages of articles. That's the problem. Xaris333 (talk) 23:49, 2 December 2012 (UTC)

Finding URL used directly in article

Is there a way to find a part URL in an article that is not supplied via a template. The "External links search" finds both direct instances and those supplied via a template. The reason is to find articles in which a template can be used for the link rather than direct coding in the page. Keith D (talk) 23:24, 2 December 2012 (UTC)

You can create a bot that searches wikitext directly. Ruslik_Zero 10:08, 3 December 2012 (UTC)

Sandbox edit mode

I use Chrome and Firefox with Vector skin. I want my sandbox to open in edit mode, which "Preferences" suggests is default, but for me it opens in read mode. Is there a simple way to either force edit mode, or add a button to the top right menu to give this option? Jimfbleak - talk to me? 10:21, 3 December 2012 (UTC)

Disable my sandbox and add this to your JS:

mw.util.addPortletLink(
 'p-personal',
 '/w/index.php?title=Special:MyPage/sandbox&action=edit',
 'My sandbox',
 'pt-mysandbox',
 'Go to your sandbox',
 null,
 '#pt-preferences'
 );

--— Gadget850 (Ed) ^talk 13:20, 3 December 2012 (UTC)

Thanks for the script. Unfortunately, when I installed it in my common.js nothing appeared in the top right, and all my other scripts stopped working. Have I missed something? Jimfbleak - talk to me? 16:06, 3 December 2012 (UTC)

Try it now - it appears it was missing a bit of punctuation, which I think I added above. -— Isarra ༆ 16:12, 3 December 2012 (UTC)

Spot on, I tweaked the text to read just "Sandbox" for consistency with the new layout — even I could work out that bit, the rest is like a magic incantation to me ): Thanks again to both of you! Jimfbleak - talk to me? 16:34, 3 December 2012 (UTC)

finding/fixing links to a particular site

Hi. I couldn't find this in the archive or FAQs and I'm sorry if it's a perennial question. A magazine's web site I use a lot for both references and external links has changed its structure breaking every link. Is there a way of identifying all the broken links/references so they can be flagged as deadlink, and ideally fixed? (some may still be on their site, some we'd have to work out the magazine issue and article which hopefully would have the same text). The magazine is Novosti Kosmonavtiki and the broken links are structured like http://novosti-kosmonavtiki.ru/mag/339/04.shtml and http://novosti-kosmonavtiki.ru/foto/gallery_184/index.htm#25

To try and avoid this happening again I've started to webarchive references to good websites. Is there a bot or something that can see that a reference is webarchived and copy across the archived link to other articles with the same reference? Doing it manually is laborious! Thanks in advance for any help. Secretlondon (talk) 15:48, 3 December 2012 (UTC)

To answer the first (simplest) part of your question:

• Special:LinkSearch lists pages that link to a specified website, e.g. http://en.wikipedia.org/w/index.php?title=Special:LinkSearch&limit=500&offset=0&target=http%3A%2F%2F%2A.novosti-kosmonavtiki.ru

There are currently only 126 links to novosti-kosmonavtiki.ru on enwiki, including the links in your post here and multiple links from some pages, so you might be able to fix these manually. But Wikipedia:Bot requests may be able to help if you need a bot to carry out more extensive work.

— Richardguk (talk) 19:54, 3 December 2012 (UTC)

Thanks - checklinks seems to be good at finding archived versions so together with your link above this is easier to fix than I'd feared! Secretlondon (talk) 22:40, 3 December 2012 (UTC)

Traffic increase statistics

About three months ago I collected the number of monthly pageviews from about 150 health articles. I am just checking monthly pageviews again for the same articles and am surprised to see that fairly consistently, all of the articles are getting 20% more traffic. Some articles are getting more, and few are not getting more traffic. Does anyone know the expected projection for traffic growth on Wikipedia or know where I can get reports containing broad commentary on Wikipedia article traffic? I am especially interested if anyone has said that Wikipedia traffic is increasing and if so, what is the rate of increase and to what extent is that rate consistent. There exists Wikipedia:Statistics but that seems like ways of viewing things at any point in time, and I wanted to know about changes over time. Blue Rasberry (talk) 15:54, 3 December 2012 (UTC)

See WikiStats. It is known that English Wikipedia traffic declines during the summer months (Students on summer holiday, more outside activities, etc.), which might explain your particular situation with the health articles. Thanks, West.andrew.g (talk) 16:01, 3 December 2012 (UTC)

Fantastic, brilliant, thanks. I saw your name on Wikipedia:Statistics - thanks for watching this page and fielding questions. Blue Rasberry (talk) 21:01, 3 December 2012 (UTC)

Filter log gets cleared when changing username?

In November, I changed my username from Anonymouse321 to The Anonymouse. I happened to check my current filter log and realized that only entries from after the username change exist (wikilove are the only logs). I also tried checking my old account name's filter log and nothing shows up (it no longer has any contributions, of course). Is this intentional? The Anonymouse (talk • contribs) 17:17, 3 December 2012 (UTC)

bugzilla:25377: "AbuseFilter should use Renameuser hooks (filter history/filter logs need to be updated when a user is renamed)". PrimeHunter (talk) 18:51, 3 December 2012 (UTC)

OK, so there is already a bug report for it. I'll check it out, thanks. The Anonymouse (talk • contribs) 20:29, 3 December 2012 (UTC)

What just happened to the UI?

Tracked in Bugzilla
Bug 42452

I now have giant text at the top of my page which has random words like "Namespaces" "Personal tools" "Views" "Actions" and "Search" at the top of every page. Using Chrome Version 23.0.1271.95 m on Windows 7. Can we go back to this not happening and implement changes to the user interface after (rather than before) they are tested? Thanks! --Jayron32 19:10, 3 December 2012 (UTC)

I saw that for a second, but it was gone the next time I loaded a page. Is it still happening? Writ Keeper ⚇♔ 19:12, 3 December 2012 (UTC)

I have the same issue, glad to know it's not just me. Mark Arsten (talk) 19:13, 3 December 2012 (UTC)

Same problem here. (Windows XP, Monobook skin, FireFox v16) WikiDan61^ChatMe!_ReadMe!! —Preceding undated comment added 19:14, 3 December 2012 (UTC)

There's more about this breakage on wikitech-l. Steven Walling (WMF) • talk 19:15, 3 December 2012 (UTC)

We've got 30 days of this? Secretlondon (talk) 19:32, 3 December 2012 (UTC)

Nope, just clear your browsers cache and you should be good. Ryan Vesey 19:34, 3 December 2012 (UTC)

Thanks, Ryan (Shift + Reload on Firefox). Never a dull moment! All the best, Miniapolis (talk) 20:58, 3 December 2012 (UTC)

And here. Writ Keeper, I've already accused you on ANI. Stay out of your sandbox. Drmies (talk) 19:15, 3 December 2012 (UTC)

I'm having it too (Mac, Chrome, Vector) --j⚛e decker^talk 19:16, 3 December 2012 (UTC)

Same here on Firefox. AutomaticStrikeout (T • C) 19:17, 3 December 2012 (UTC)

Happened here also, and last week this happened on Wikidata. --Stryn (talk) 19:16, 3 December 2012 (UTC)

• Based on my interpretation of this. All the pages need to either be edited or purged. I used ?action=purge to purge the main page and the problem appeared fixed. Can we create some type of botlike thing to go to all of the URL's on Wikipedia and purge them? Ryan Vesey 19:18, 3 December 2012 (UTC)

• Woops, that did nothing. The main page is messed up again. Ryan Vesey 19:20, 3 December 2012 (UTC)

That was apparently my fault in [16]. Very sorry. I have no idea why it caused this. I just wanted the description at Special:Preferences#mw-prefsection-gadgets to match what the gadget actually says after "My" was removed in [17]. It's fixed for me after Reaper Eternal reverted my edit. PrimeHunter (talk) 19:18, 3 December 2012 (UTC)

Are you sure? I thought that was just a documentation page; I didn't think changing it could break anything. Writ Keeper ⚇♔ 19:19, 3 December 2012 (UTC)

I didn't think so either but the timing matches. Do you have the courage to repeat my edit and see what happens? I'm not touching that again. Expect a trout slapping if you actually try it and it destroys Wikipedia. PrimeHunter (talk) 19:29, 3 December 2012 (UTC)

...I kinda want to try. Here goes nothing... Writ Keeper ⚇♔ 19:33, 3 December 2012 (UTC)

Yep, I reinstated the edit, and I haven't seen the problem reappear, even after clearing the (theoretically already empty) cache multiple times. Writ Keeper ⚇♔ 19:39, 3 December 2012 (UTC)

https://wikitech.wikimedia.org/view/Server_admin_log says 1.21wmf5 was deployed to the English Wikipedia 19:05 today. My edit was 19:05 and the interface broke right after. Maybe I'm innocent. Or maybe it was a combination where my edit was the first interface edit to cause something to be reprocessed by 1.21wmf5. 1.21wmf5 has apparently caused similar issues at other wikis where it was deployed 28 November. PrimeHunter (talk) 20:15, 3 December 2012 (UTC)

I suppose, then, that someone needs to go tell the developers to halt the deployment of 1.21wmf5 until the problem is fixed. WikiPuppies _bark ^dig 20:19, 3 December 2012 (UTC)

I'm guessing people have tried this, but if you haven't, maybe clearing your cache will help? It might've been a temporary thing; I don't have a cache on this browser, which might be why I don't see it any more. Writ Keeper ⚇♔ 19:18, 3 December 2012 (UTC)

I use Firefox and did a Shift/reload, and it went away.--Bbb23 (talk) 19:19, 3 December 2012 (UTC)

(edit conflict)(more than I can count)I'm having the same problems as Jayron32 with the same specs, using modern though. It seems to be inconsistent, with pages occasionally displaying correctly at the top but mostly not. I also restored a pertinent comment from Steven (WMF) that Drmies ECed into oblivion. —Torchiest ^talk_edits 19:22, 3 December 2012 (UTC)

I used ctrl/reload on chrome and it went away, but still comes back every time I open a pge. Ryan Vesey 19:24, 3 December 2012 (UTC)

For Chrome, I had to "Clear Browsing Data > Empty the Cache". --j⚛e deckertalk 19:24, 3 December 2012 (UTC)

Fixed it entirely (for now at least). Thanks. Ryan Vesey 19:27, 3 December 2012 (UTC)

Moving up from the thread below. That Shift/Reload has worked for me - and other newly opened pages are coming up clean too. Must remember that one. Peridon (talk) 19:26, 3 December 2012 (UTC)

I have the heading on the left side larger, and random headings across the top. Also the Article title is behind the Article and Talk tabs. This is apparent on first visits to page. Not sure clearing browser cache is the solution as it is happening with pages I have not visited before. Using Chrome 23.0.1271.95 on Windows 7 Pro SP1. --Stewart ^{(talk | edits)} 19:29, 3 December 2012 (UTC)

Just FYI: I think it will help because it's the sitewide CSS and/or JS pages that were borked and need to be cleared, not any article page in particular. Just my hypothesis, though. Writ Keeper ⚇♔ 19:35, 3 December 2012 (UTC)

(edit conflict) (x3) I think someone is about to be sentenced to the village stocks. WikiPuppies _bark ^dig 19:31, 3 December 2012 (UTC)

• I should add that although Shift/reload in Firefox has eliminated the garbage, not all pages look normal. For example, my watchlist page looks fine. However, a revision history page has some parts of the print missing in the upper right-hand corner (page history, etc.). If I recall correctly, I don't use the normal settings. I think I use some kind of script that moves certain actions from the left sidebar to the the spot that's not quite right.--Bbb23 (talk) 19:41, 3 December 2012 (UTC)

Above the link to the "Main page" on the left hand side I realised it now says "Navigation", but the tabs underneath, such as "Toolbox", no longer collapse/expand. So I suspect something is still broken. And despite clearing the cache, the problem still comes back for me on Firefox. Jared Preston (talk) 19:43, 3 December 2012 (UTC)

Cleared my cache, and it worked to fix the problem. Thanks for the tips. However, I am having the problem that the collapse function no longer works on the leftside menu, as noted above. --Jayron32 19:46, 3 December 2012 (UTC)

All javascript seems to have stopped working. Reaper Eternal (talk) 19:50, 3 December 2012 (UTC)

Hmm, my JS is still working fine, but I am seeing the non-collapse issue. Unrelated? Writ Keeper ⚇♔ 19:52, 3 December 2012 (UTC)

The cache purge fixed most pages for me, but my watchlist is still borked after repeated reloads... Sailsbystars (talk) 20:15, 3 December 2012 (UTC)

Same problem. I have big, bold, "Views" and "Personal tools" at the top, and my buttons are hidden. I'm in Firefox and purging the cache did not work. Glad I'm not alone in this...well, kinda. ~Adjwilley (talk) 20:16, 3 December 2012 (UTC)

Figured out my problem, cache still wasn't quite cleared. Now all pages working properly. If at first you don't succeed, purge, purge again. Sailsbystars (talk) 20:43, 3 December 2012 (UTC)

Purge, purge again is good advice. It finally worked. ~Adjwilley (talk) 22:38, 3 December 2012 (UTC)

According to https://bugzilla.wikimedia.org/show_bug.cgi?id=42452#c15 this should work correctly now. If it does not for you: Make sure to reload the page and bypass your cache (see Wikipedia:Bypass_your_cache). If the problem still happens, please report your browser and version and specific pages (web addresses) in case this only happens on some pages. Thanks! --Malyacko (talk) 21:04, 3 December 2012 (UTC)

I am using Firefox 17.0.1. I have cleared my entire cache. The Main page is an example of it not being back to normal. In the upper right-hand corner, to the left of the search box, I have "Page" with a drop arrow and "TW" with a drop arrow. I can see only the top half of those two words, whereas before I could see all of them.--Bbb23 (talk) 21:47, 3 December 2012 (UTC)

I tried to purge the cache but still no change. Using safari on iPad. Monobook. Rcsprinter (natter) @ 21:30, 3 December 2012 (UTC)

I cleared my cache a few hours ago and have been fine since then. Except... didn't the dropdown arrow for Twinkle used to be next to the "TW", not above it and to the right? Just odd for that to be the only glitch I'm getting. Anyone else? — Francophonie&Androphilie^{(Je vous invite à me parler)} 21:53, 3 December 2012 (UTC)

I'm getting that, too. Writ Keeper ⚇♔ 22:05, 3 December 2012 (UTC)

Its doing the same thing on my iPad. How can I clear the cache on my iPad? It's really hard to edit with everything on the page.--Astros4477 (talk) 22:26, 3 December 2012 (UTC)

The only way I know of is to go to Settings → Safari → Clear cache (or whatever it's called). The Anonymouse (talk • contribs) 22:28, 3 December 2012 (UTC)

BTW, I am also seeing the TW issue. Posted at WT:Twinkle#Twinkle menu misaligned. The Anonymouse (talk • contribs) 22:30, 3 December 2012 (UTC)

Purging my cache has helped for me. Chrome; Settings -> Show advanced settings -> Clear browsing data -> Empty the cache. However, it's not appropriate to expect every reader to browse here and follow the instructions to purge their cache. Presumably the underlying problem is still going to be fixed sooner or later? bobrayner (talk) 22:43, 3 December 2012 (UTC)

Some temporary workarounds - I have found (using chrome and 'modern' UI) that the user/talk/prefs/watchlist line and the article/talk/edit/history line of the interface are both just gone. Purging cache, full refresh, etc did not change anything. However, using tab and shift-tab to navigate between links on the page causes them to reappear as normal when you tab to a link on the line. Also, my watchlist, which uses the 'sort by namespace' javascript advertised on the js help page, doesn't suffer from this problem, nor the messed up sidebar font sizes. All other pages seem to, so maybe something that script does is incidentally fixing the problem. And if both of those fail....firefox seems completely fine. --Qetuth (talk) 22:46, 3 December 2012 (UTC)

I've purged my cache, switched skins temporarily, restarted my browser...every normal page is fine now, but all special pages (watchlist, contribs, etc) still have broken menus. Any suggestions? I'm using Modern in Firefox. Nikkimaria (talk) 04:05, 4 December 2012 (UTC)

Click the top left orange Firefox button. Then "History->Clear recent history" Then check just the "cache" box, and make sure "Time range to clear" is "Everything". That's what fixed it for me. —Torchiest ^talk_edits 04:45, 4 December 2012 (UTC)

And on older versions of Firefox, you would go Tools, Options, Privacy, Clear your recent history, then check the cache box and choose Everything as before. Stifle (talk) 10:06, 4 December 2012 (UTC)

My manifestation of this (I suppose) is that columns previously set in em no longer worked, on different machines with different browsers. I don't know if that was fixed, but I changed them to px anyway (but I hate that). Also, the width of infobox islands increased 50% or more.

John of Cromer in China (talk) mytime= Tue 18:39, wikitime= 10:39, 4 December 2012 (UTC)

MonoBook problems?

Has anyone else had severe page layout problems today while using MonoBook? I'm using SeaMonkey 2.14. WikiPuppies _bark ^dig 19:14, 3 December 2012 (UTC)

Yup, see the thread above. Writ Keeper ⚇♔ 19:15, 3 December 2012 (UTC)

I'm getting it too. It's not stopping me deleting things, but it's a damn nuisance finding the tag when the buttons are behind the article title. I'm in Firefox 15 with XP. This page (except in 'edit view') is fine, but the type size in the sidebar has enlarged on ordinary pages. Peridon (talk) 19:18, 3 December 2012 (UTC)

I've tried purging, and that makes no difference. It's coming up on pages I'd not opened before it happened. Peridon (talk) 19:21, 3 December 2012 (UTC)

Have you tried clearing the browser cache, though? Purging is a server thing, IIRC, and I think this is a client browser thing. Writ Keeper ⚇♔ 19:23, 3 December 2012 (UTC)

After purging my local browser cache since the begining of time. GoogleChrome WinXP. Why the hell did they let something so broken ship, and without any sort of notice? Hasteur (talk) 20:29, 3 December 2012 (UTC)

Took a full "empty your cache", not just a refresh or the like, on my mac/chrome, but if that's not working for ya, yikes. --j⚛e decker^talk 21:10, 3 December 2012 (UTC)

Purging the cache (Firefox) seems to have done the trick for me. Fingers crossed. RashersTierney (talk) 21:14, 3 December 2012 (UTC)

It became unusable for me in Firefox 17/Linux today, although it still looked normal in SeaMonkey 2.14. Clearing the cache by going to Edit -> Preferences -> Network -> Cached Web Content and clicking the Clear Now button and reloading the Wikipedia page fixed the messed up layout problem. — Quicksilver^{T @} 02:11, 4 December 2012 (UTC)

Counting template

Is there a template that works in a similar manner to bare external links (i.e. [18], [19], [20]), in other words one that outputs the number of times it has appeared on a given page? This would be particularly useful for automatic numbering of long, frequently updated tables (such as List of countries by population). mgiganteus1 (talk) 19:31, 3 December 2012 (UTC)

Not yet, this is one of those perennial requests; although when we get mw:Lua it may then be possible. --Redrose64 (talk) 20:36, 3 December 2012 (UTC)

• Auto-numbering templates have been begun: There are at least 2 templates which count or "auto-number" their parameters:

• Template:Autocol - quickly formats a list to have auto-numbered entries (or asterisk-bullets), as columnized into "ncols=n" columns across the page.
• Template:Autonumbered_list - highly complex (5-second slow) template to auto-number the rows in a table (max: 50 rows of 5 cells), also wrapped into multiple columns across the page.

I recently wrote the quick Template:Autocol to auto-wrap, and auto-number multiple columns in a lightning-fast manner on any browser, as 1 item per line. Slower Template:Autocol_big allows 65-140 items in an autonumbered table. Now, the extremely complex Template:Autonumbered_list was written way back in 2009, to miraculously number up to 50 rows in a table of up to 5 row cells, which is then wrapped as multiple columns across the page (and get this: it works!). It is so complex, I had to struggle to rewrite it to avoid "wp:Exceeded template limits". However, it is a good example of the future of a table where all rows would be auto-renumbered when any rows are added/removed (likely to be rewritten much faster with a Lua module). However, using {autocol} today is simple and fast:

• {{autocol|ncols=4|n=16|num=y|wrap=y |A|B|C|D|E|F|G|H|I|J|K|L|M|N|O|P}}

1.  A         2.  B         3.  C         4.  D         5.  E         6.  F         7.  G         8.  H         9.  I       10.  J       11.  K       12.  L       13.  M       14.  N       15.  O       16.  P

As noted, people have wanted multi-column auto-numbering, and so it has begun. -Wikid77 (talk) 00:37, 4 December 2012 (UTC)

Edit Filter Question

I've come across a situation where the edit filter has caught a constructive edit and disallowed it. The edit is substantial enough that it clearly needs to be attributed for copyright purposes. What needs to be done to push the edit through so that the attribution for the edit is not broken? Log of disallow [21] Monty₈₄₅ 22:07, 3 December 2012 (UTC)

There is no technical way to reinstate a filtered edit AFAIK, so maybe include their username/IP in your edit summary? -- King of ♥ ♦ ♣ ♠ 22:31, 3 December 2012 (UTC)

Special:BlockList is broken

I can't access Special:Blocklist - the message I get is a heading of "Internal error" followed by a red box with "[2dfb96b1] 2012-12-03 23:32:22: Fatal exception of type MWException". Fun!--Jezebel'sPonyo^{bons mots} 23:33, 3 December 2012 (UTC)

I have the same problem. Yikes. WikiPuppies _bark ^dig 23:35, 3 December 2012 (UTC)

Tracked in Bugzilla
Bug 42664

King of ♥ ♦ ♣ ♠ 23:38, 3 December 2012 (UTC)

Works for me. Peridon (talk) 10:48, 4 December 2012 (UTC)

Layout issue

I was minding my own business (and possibly hit a button), when my layout turned to this. It's not my computer cause when I log out it goes away, and it occurs on my iPhone as well. Please help. Grammarxxx ^{(What'd I do this time?)} 23:47, 3 December 2012 (UTC)

Happened to me too, but bypassing my cache made the problem go away. David1217 ^{What I've done} 23:52, 3 December 2012 (UTC)

Everybody had this for about 5 minutes starting at 19:05. See #What just happened to the UI? It should be fixed hours ago. Clear your entire cache to get the fix. PrimeHunter (talk) 23:54, 3 December 2012 (UTC)

It's fixed now, thank you all! Grammarxxx ^{(What'd I do this time?)} 00:05, 4 December 2012 (UTC)

I'm getting it on every new page I visit until I refresh on each. What am I doing wrong? -— Isarra ༆ 04:44, 4 December 2012 (UTC)

I am getting this too. Firefox 17.0.1, which I was frogmarched into recently.

What gives? - Smerdis of Tlön - killing the human spirit since 2003! 06:04, 4 December 2012 (UTC)

Same. Its in vector and its not any of the scripts I think. Not sure why. ChrisGualtieri (talk) 06:58, 4 December 2012 (UTC)

I'm also still seeing this, depite having refreshed, purged and generally done everything short of colonically irrigating my browser cache. It only seems to affect Firefox, I'm not seeing it in any other browser. Yunshui 雲‍水 08:38, 4 December 2012 (UTC)

Personally, I saw this about 2 days ago, but on Commons only. Now I do not see it on any Wikimedia site. -- King of ♥ ♦ ♣ ♠ 08:47, 4 December 2012 (UTC)

It did it for me, a few minutes ago. It seems to be fixed now. I did change some preferences, so it might be a side effect of one of the gadgets. — Arthur Rubin (talk) 08:51, 4 December 2012 (UTC)

It's now gone for me, after I completely deleted my cache a second time. I'm assuming something didn't work properly last time I did this. Yunshui 雲‍水 08:55, 4 December 2012 (UTC)

Personalized Edittools

Hi all.

There are quite a few personalized "Edittools" scripts floating around, but i think this one is a bit different, and maybe can be more useful to some users.

The main difference is that this tool uses a very simple script that does not contain the edit tools themselves: rather, each user who wishes to personalize their edit tools, creates a specific page in their userpspace (surprisingly, this page is called "Edit tools", or more precisely, Special:Mypage/Edit tools), where they define their own personalized Edit tools.

The tools appear in a box, between the public "Edit tools" and the edit box (aka "textarea"). The script itself is in User:קיפודנחש/personalEdittools.js, and a rudimentary documentation, including installation instructions are in User talk:קיפודנחש/personalEdittools.js.

peace - קיפודנחש (talk) 23:48, 3 December 2012 (UTC)

edit filter problem

I've had a recurring problem with the edit filter flagging posts of mine as spam and not letting me save them. When I report the false positive, people tell me that I'm supposed to be able to press "save" a second time (example) and the message from the filter says something similar. But multiple save attempts don't seem to make a difference, and minor and major adjustments to the wording usually don't help either. Can somebody fix that issue (i.e. back the response level off to "warning" since I think it must now be set to "disallow")? The filter is just too aggressive and makes editing intolerable when it triggers. Thanks. 66.127.54.40 (talk) 01:50, 4 December 2012 (UTC)

Where are my admin tabs?

The layout problems above did not affect Firefox, but persisted in Chrome. I therefore did a complete "clear all browsing data", "empty cache" and logged back in. All my stuff (watchlist, scripts etc) are working, so it knows it's me, but the admin tabs (delete/move/protect) have disappeared. Everything is fine on FF still, so why won't Chrome accept me as a sysop?Jimfbleak - talk to me? 07:41, 4 December 2012 (UTC)

I'm so stooopid, tabs had reverted to dropdown, hadn't noticed. I must get out... Jimfbleak - talk to me? 07:50, 4 December 2012 (UTC)

Wanted: Page purge tool

Is there a script or gadget that will add a "purge page" link to the toolbox? If not, could someone with the sills needed make one, please? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:35, 4 December 2012 (UTC)

There's two filed under "Gadgets > Appearance" in preferences. Andrew Gray (talk) 11:04, 4 December 2012 (UTC)