Jump to content

Wikipedia:Village pump (technical)

From Wikipedia, the free encyclopedia

This is an old revision of this page, as edited by Trappist the monk (talk | contribs) at 19:45, 15 October 2015 (→Good news, bad news). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

 Policy Technical Proposals Idea lab WMF Miscellaneous 
Shortcuts
The technical section of the village pump is used to discuss technical issues about Wikipedia. Bug reports and feature requests should be made in Phabricator (see how to report a bug). Bugs with security implications should be reported differently (see how to report security bugs).

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.


vertical bar in template parameter

Is there a way to escape a | in a template? A page title contains a | and I'm trying to use {{cite web|title = Title with | in it|date= ...NE Ent 13:35, 3 October 2015 (UTC)[reply]

In {{cite web}}, use %7c inside |url= and | in any other parameter when you want to write a vertical bar. – Jonesey95 (talk) 13:48, 3 October 2015 (UTC)[reply]
Alternatively, you can place | within nowiki tags, i.e. <nowiki>|</nowiki>. 2607:FB90:21C9:F62A:0:37:4515:9301 (talk) 17:39, 3 October 2015 (UTC)[reply]
@NE Ent: Very often, a pipe in the title of a web page is used to separate the true title of a web page from the name of the website. They might be in either order. So, a web page that is titled like "Guaranteed Roadrunner Trap | Acme" or like "Acme | Guaranteed Roadrunner Trap" would be split into |title=Guaranteed Roadrunner Trap |website=Acme. If you provide the URL of the web page, I can look at it to advise on the best action. --Redrose64 (talk) 20:54, 3 October 2015 (UTC)[reply]
Also, {{!}} will do the trick. — Mr. Stradivarius ♪ talk ♪ 04:08, 4 October 2015 (UTC)[reply]
Thanks. It's "blah blah blah | Science News" -- I don't what ya'll consider the true title ... I just use what is between the <title> </title> tags on the page. NE Ent 12:31, 10 October 2015 (UTC)[reply]

comparing revisions across two or more history pages without missing some in between

When searching old revisions by comparing an older one with a newer one that necessarily are listed on different pages of the page's history, a searcher ordinarily conducts two or more searches, thereby missing the effect of one intermediate revision listed at the bottom of a history page (or two intermediate revisions at the bottom of two history pages, etc.). This is because the older revision selected on the history page is not itself reflected in the top of the resulting diff, but only implied in the midst of everything still older, thus easily missed or misunderstood. It is possible to solve this by manually constructing a URL, but that possibility is easily missed by users.

Example (frequent editing of the article will quickly make this outdated but nonetheless the model applies):

Perhaps this can be solved with a technical solution.

Nick Levinson (talk) 02:00, 4 October 2015 (UTC)[reply]

You can list up to 5000 revisions at once in a page history. Careful setting of the "From year (and earlier)"/"From month (and earlier)" options will bring the appropriate revision close to the top; or you can set &offset= This will allow you to get both on the same screen. --Redrose64 (talk) 10:43, 4 October 2015 (UTC)[reply]
That's complicated for most users and not explained where most users would likely see it, and I know how to get up to 500 revisions in a page history, not 5,000. Can this be explained with a link from history pages or the procedure technically simplified? Nick Levinson (talk) 00:41, 11 October 2015 (UTC)[reply]
If you click a number of revisions then you get a url where the number can manually be changed to anything up to 5000. I suspect the interface links stop at 500 for performance reasons. I don't know whether it's OK to advertise the 5000 method on page histories. PrimeHunter (talk) 08:40, 11 October 2015 (UTC)[reply]
If we were at WP:VPR, I'd oppose adding even a little to the page history page with so little need for it. Thankfully, we're not. At least you have some way of getting what you need. Maybe an addition to Help:Page history about this would be warranted, and the page history page already has a link to that page. ―Mandruss ☎ 09:55, 11 October 2015 (UTC)[reply]
For comparison, user contributions link to Help:User contributions which already says: "The blue numbers list the number of edits displayed on a page: 20, 50, 100, 250 or 500. The number you select replaces n in the links to the previous or next pages e.g. (newer 100 / older 100). Views of up to 5000 edits per page are possible by modifying the URL." PrimeHunter (talk) 10:28, 11 October 2015 (UTC)[reply]

Shutting off pings from specific users

I really do like the pings in general, and I can apparently shut them off overall, but sometimes I'm dealing with a user who has gotten "ping-happy" and stays that way even after I've asked them to stop doing that. I don't want to go as far as reporting the user, but I sure would like to have their pings to me shut off. Is there any way this can happen? Stevie is the man! Talk • Work 16:09, 7 October 2015 (UTC)[reply]

Probably better place to ask is WT:Echo, but it sounds like a good idea. --Edgars2007 (talk/contribs) 17:30, 8 October 2015 (UTC)[reply]
Thanks. I am continuing to get ping-abused from two particular editors. So I would love a solution. Stevie is the man! Talk • Work 18:43, 10 October 2015 (UTC)[reply]

Revising Special:WantedPages

There is fairly universal agreement, at Wikipedia talk:Special:WantedPages, that this automated list is, in its current form, almost worthless. That's because the high-count items are being generated by either signatures on article talk and other pages, or by templates on article talk pages.

What is useful is Wikipedia:Most-wanted articles. That's built using red links from article pages, only. But it's not automatically generated - it's not a special page.

So, a purely technical question: Who has the authority/ability to modify the parameters that generate the Special:WantedPages content? I'm aiming, of course, at getting those parameters changed so that only red links on article pages will count; I'd like to know who has to be convinced that the current content isn't worthwhile. -- John Broughton (♫♫) 21:36, 7 October 2015 (UTC)[reply]

WantedPages is hard-coded into the MediaWiki software as including pages from all namespaces. So there is no parameter to change at the moment. The relevant task is phab:T37758.
Personally I think the query pages need a lot of help, since most of them (not just WantedPages) are pretty worthless on large wikis like English Wikipedia. Unless and until any improvements are made, Labs is the best place for developers to create improved alternatives to the query pages. — This, that and the other (talk) 01:21, 8 October 2015 (UTC)[reply]
It's used by the ContentTranslation feature, so it is very prominently linked at user contribution pages under "New contribution". Gparyani (talk) 18:08, 11 October 2015 (UTC)[reply]
The answer to the specific question you asked is that you need developer-level rights to change that, because it's hardcoded in the software and thus needs a software change to change. (It might also need a change to the database schema, and the like; it doesn't strike me as the type of thing that would be easy to do efficiently with the current software.) Gparyani's response is a good guide as to how you might practically go about getting it changed. --ais523 22:03, 14 October 2015 (UTC)

Is there a way to auto-populate categories through Wikidata/other wiki comparison?

I just created Category:Polish skydivers and Category:Cichociemni, both based on Polish Wikipedia categories (linked through Wikidata). Now, pl wiki categories have quite a few entries, and I expect 20-30% of them are on en wiki already. Is there some way (bot, script) that would automate checking which entries categorized under equivalent Polish categories exist here, and auto-tag them? It seems like something that should be possible to automate. --Piotr Konieczny aka Prokonsul Piotrus| reply here 09:15, 9 October 2015 (UTC)[reply]

If nobody has better solution, I have one, but it's quite crappy. Take Autolist, fill the parameters and get the list of Wikidata items. The URL with parameters for Cichociemni is here (it will take some few minutes to open). Then you can open each article and add category by yourself. Yes, it doesn't actually ease the work very much :D --Edgars2007 (talk/contribs) 10:51, 9 October 2015 (UTC)[reply]
But anyway - 👍 Like the idea. Would really like to use such tool, which at least gets that list of pages, that needs to be categorised. --Edgars2007 (talk/contribs) 14:22, 9 October 2015 (UTC)[reply]
@Magnus Manske: Possible? --Izno (talk) 03:09, 10 October 2015 (UTC)[reply]
Well, you can use PagePile. Generate a list of pages on plwiki in a category in QuickIntersection, then filter the pagepile here, by "converting" it to Wikidata, then to enwiki; try "Load from PasteBin" with ID d6653a0c to see what I mean. You now have a list of all enwiki articles that correspond to plwiki articles on Polish skydivers. Convert that back into Wikidata, then load it in autolist, and set "pages in enwiki category Polish skydivers" as "NOT". You will then get all Polish skydivers from plwiki that are on enwiki but not in the enwiki category "Polish skydivers". The parts are all there, you just need to connect them :-) --Magnus Manske (talk) 12:13, 12 October 2015 (UTC)[reply]
@Magnus Manske, Izno, and Edgars2007: My eyes glazed a bit reading the how to... any chance somebody (pretty please) could hack together a tool which would make it as easy as typing in/linking the two categories (English and Polish)? --Piotr Konieczny aka Prokonsul Piotrus| reply here 02:49, 13 October 2015 (UTC)[reply]
Actually, I think I already wrote this but forgot about it ;-) Is this what you want? --Magnus Manske (talk) 18:21, 13 October 2015 (UTC)[reply]
@Magnus Manske: No, it actually does the opposite - it finds articles, which are in plwiki, but not in enwiki. BTW, if we have already started to talk about the tool - sometimes it isn't working, if there is more than 100 results. See here (fill in "WDQ": tree[211][150][17,131] OR claim[27:211]) - 101st result and so on isn't right... And if I tick "Source Wikipedia", then at page 2 the tick is on "Wikidata". --Edgars2007 (talk/contribs) 19:31, 13 October 2015 (UTC)[reply]

Table won't sort

On List of Xbox Live Arcade games, there is a table of 707 rows. (At least, the article says there are 707 games. I haven't counted the rows.) When clicking the arrow next to the text of the "Release Date" column header, there is a delay of a few seconds, and then the arrow will change, as expected; but the table does not sort.

I don't see this behavior when sorting by the other columns; the table sorts normally after clicking one of those arrows. Tarcil (talk) 00:03, 10 October 2015 (UTC)[reply]

@Tarcil: I've fixed this by adding data-sort-value for each of the rows, as described at Help:Sorting#Specifying a sort key for a cell. — Mr. Stradivarius ♪ talk ♪ 01:20, 10 October 2015 (UTC)[reply]
@Mr. Stradivarius: Thank you! Is there an item on the developers' backlog to automate this for the thousands (I assume) of tables on Wikipedia that don't include this markup? Tarcil (talk) 17:42, 15 October 2015 (UTC)[reply]
Works fine for me. Tvx1 01:20, 10 October 2015 (UTC)[reply]
Is it really necessary to link all those years? That's the reason the sorting breaks (it would not need a sort key otherwise), and it seems excessive to me. nyuszika7h (talk) 16:06, 11 October 2015 (UTC)[reply]

SQL question

I've been trying to find out the total admin actions per month in order to answer a question at WT:RFA, but I've been having problems running my SQL query on Quarry. My query to find the total admin actions for September ran fine, but when I tried to expand that to all months from January 2001, my query took over 20 minutes to execute and was killed. Is there anything I can do to improve the efficiency of this, or am I better off just making separate database queries for each month from an external script? — Mr. Stradivarius ♪ talk ♪ 11:49, 10 October 2015 (UTC)[reply]

Backend is mysql? If log_timestamp isn't indexed, it's likely to be slow regardless of your query. You could try using the IN operator [1] instead of the multiple OR statements. NE Ent 12:29, 10 October 2015 (UTC)[reply]
Thanks for the tips. :) Yes, it's mysql. I've tried using the IN operator and also reducing the number of dates from 1000 to 300, so we'll see if that helps. It's waiting in the queue now, but there are no other queries running - I think I might have lost my right to have it executed straight away after the previous two attempts... — Mr. Stradivarius ♪ talk ♪ 13:40, 10 October 2015 (UTC)[reply]
Actually, I just saw this example query, which is an approach that should be much more efficient. I'll give that one a try. — Mr. Stradivarius ♪ talk ♪ 14:19, 10 October 2015 (UTC)[reply]
You can also sign up for a tool labs account and run the query manually instead of using quarry :) Legoktm (talk) 15:08, 10 October 2015 (UTC)[reply]
I have one, actually, but I wanted to do it on Quarry so that other people could easily see the results and maybe modify the query. In any case it will have to wait, as I'm getting "SELECT command denied" errors at the moment. — Mr. Stradivarius ♪ talk ♪ 15:43, 10 October 2015 (UTC)[reply]
It's because you are trying to run the query on the private "enwiki" database, not "enwiki_p" which is public. I just ran it on enwiki_p here. --Glaisher (talk) 15:55, 10 October 2015 (UTC)[reply]
BTW, you don't have to include USE enwiki_p;. For enwiki it's there by default. Edgars2007 (talk/contribs) 16:56, 10 October 2015 (UTC)[reply]
Ah, that makes sense, thanks. — Mr. Stradivarius ♪ talk ♪ 16:58, 10 October 2015 (UTC)[reply]

Page Padding

Am I delusional or was the padding on the mw-body class recently changed? It seems most noticeable on the Main Page where there seems to be a larger white gap along the left edge than I recall seeing in the past. Dragons flight (talk) 20:21, 10 October 2015 (UTC)[reply]

No recent changes. But in Vector, the padding does change from 1 to 1.5em when the screen width goes above 982px. It has been doing that form some years now. Have you gotten a bigger screen perhaps? -- [[User:Edokter]] {{talk}} 10:57, 13 October 2015 (UTC)[reply]

Page size dumps

Does anyone know if there is a dump file of en.wikipedia which contains some measure of the size of each page, without having to waste the bandwidth downloading all the page contents. Thanks. HYanWong (talk) 14:04, 11 October 2015 (UTC)[reply]

To answer my own question, in enwiki-XXXXX-page.sql.gz ('only' a 1.2Gb download), the penultimate column in the 'page' table is a field named 'page_len', which gives this statistic.HYanWong (talk) 22:09, 11 October 2015 (UTC)[reply]
@HYanWong: What do you need the data for? Instead of downloading a dump, it might be better to run an SQL query on Quarry. — Mr. Stradivarius ♪ talk ♪ 01:50, 12 October 2015 (UTC)[reply]
@Mr. Stradivarius: I want to find out the page size of about 1.5 million pages (all living organisms), which I'm going to combine with page views from http://dumps.wikimedia.org/other/pagecounts-ez/merged/ to assess some measure of popularity of each species or higher level group. Thanks for the tip about Quarry: is it considered reasonable behaviour to ask for such large datasets via SQL? HYanWong (talk) 07:51, 12 October 2015 (UTC)[reply]
I think in this case downloading enwiki-XXXXX-page.sql.gz would be better solution. I once run a query on Quarry, that returned 30,000 pages, it was fine, but you most probably won't get 1.5 million pages returned. But you can try :) --Edgars2007 (talk/contribs) 09:37, 12 October 2015 (UTC)[reply]
Thanks. An additional problem in this case is that the query itself contains 1.5 million page titles (grabbed from the wikidata dump), so even sending the query in the first place will be a pain! HYanWong (talk) 10:03, 12 October 2015 (UTC)[reply]
Hint (but probably not for this case). Put links to pages in some user-space page, then use pagelinks table joined with page table in Quarry. --Edgars2007 (talk/contribs) 10:29, 12 October 2015 (UTC)[reply]
P.s. I would like to document somewhere that the page lengths are present in XXX-page.sql.gz, but not sure where best to add documentation. Suggestions? (edit: maybe best to change the summary in the dump listing to "Base per-page data (id, title, old restrictions, page size etc).", but that's not user editable)HYanWong (talk) 10:05, 12 October 2015 (UTC)[reply]

Quotation marks breaking URL display

I've noticed just recently that there seems to have been a change to how external links are parsed that seems problematic. In particular, note how this:

[http://example.com?foo="bar"&bar="baz" Foobar]

renders as

"bar"&bar="baz" Foobar

i.e. that the "bar"&bar="baz" is included in the displayed URL, which is obviously unintended. This can be manually solved, i.e. by changing " to %22:

Foobar

However, since there are many, many such links that would have to be fixed (in no small part Google Books URLs), this is probably something that needs to be dealt with on the software end of things. {{Nihiltres |talk |edits}} 17:09, 11 October 2015 (UTC)[reply]

This isn't a new thing - I remember running into this years ago when I was making a template (can't remember which one offhand). — Mr. Stradivarius ♪ talk ♪ 17:21, 11 October 2015 (UTC)[reply]
Yes, it's listed at Help:URL#Fixing links with unsupported characters. The example here was also broken when The Internet Archive first archived the page in [2] 14 August 2014. PrimeHunter (talk) 17:36, 11 October 2015 (UTC)[reply]
I'm seeing the issue, in particular, in some citations. I don't recall seeing that one before. Possibly something's changed in one of the involved templates? Possibly something in this change? {{Nihiltres |talk |edits}} 18:44, 11 October 2015 (UTC)[reply]
Could you give us an example of a place where you're seeing problems? — Mr. Stradivarius ♪ talk ♪ 01:13, 12 October 2015 (UTC)[reply]
Nihiltres, the CS1 citation templates were recently (Sep 25 or 26) updated to feature more robust checking of URL parameters in the templates. I recommend reposting your question at Help talk:Citation Style 1 with some examples from articles. It is possible that the enhanced error detection has resulted in a change in the displayed link. It is also possible that the link wasn't working to begin with, and the new display is a side effect that will go away when the link is fixed.
Since the change, Category:Pages with URL errors has increased its population from roughly zero articles, where it had been for over a year, to 13,000+ articles. We have some citation template URLs to fix. All of you VPT watchers are welcome to come over and fix a hundred or so. – Jonesey95 (talk) 19:35, 12 October 2015 (UTC)[reply]

Edit summary maxlength

WP:ES#The 250 character limit claims an edit summary is limited to 255 bytes (not characters). I recall seeing some language to the effect that the limit was around 230 so a few multibyte UTF-8 characters wouldn't overflow. However, there is now a hard limit of maxlength="200", although possibly that can be overridden with JavaScript? Has maxlength been reduced in the last few months? Johnuniq (talk) 03:21, 12 October 2015 (UTC)[reply]

I must be misremembering because looking at some old source for EditPage.php makes it appear nothing has changed for several years, and "maxlength is overridden in JS to 255 and to make it use UTF-8 bytes, not characters" (what is a UTF-8 byte?). Johnuniq (talk) 06:58, 12 October 2015 (UTC)[reply]
Where do you see this maxlength="200"?
The maximum typeable length and the maximum storable length are not the same thing. The maximum storable length has always been 255 bytes. Edit summaries are stored as UTF-8, and a character like the en-dash (–) occupies three bytes when stored in this manner. The maximum typeable length was 200 characters for several years, there was a gadget that could increase this to (I think) 250; but that became redundant when the limit imposed by the MediaWiki software was increased two or three years back. If those 250 or so characters were more than 255 bytes when encoded as UTF-8, the excess would be truncated. The maximum typeable length is presently 255 bytes; if you type an edit summary using pure ASCII characters, you can enter 255 before it stops letting you type; if you type an edit summary using only characters that are three bytes in UTF-8 (e.g. Japanese script), you'll find that it won't let you enter more than 85. So unless you use characters outside the Basic Multilingual Plane that encode as four or more bytes (Unicode code points U+10000 or higher), the practical typing limit is between 85 and 255 characters. --Redrose64 (talk) 08:40, 12 October 2015 (UTC)[reply]
Thanks but I think "maximum typeable length is presently 255 bytes" is only correct if JavaScript is enabled—I can only type 200 characters with scripting off. The maxlength="200" is in the html source of a page with an edit summary and is what tells my browser (when JavaScript is disabled) to not accept more than 200 characters in the edit summary box. If anyone wants to experiment, the box following has one physical line starting with "A" and finishing with "Z". The line is 497 bytes = 200 characters. With scripting off, I can paste the line into an edit summary, but previewing shows the summary truncated. For some reason, when JavaScript is enabled, pasting the line does not work (it is truncated).
A—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×—×Z
Johnuniq (talk) 10:29, 12 October 2015 (UTC)[reply]
Yes, with JavaScript you can enter UTF-8 characters which use up to 255 bytes, which is was the database field can hold. With JavaScript off you can enter 200 UTF-8 characters regardless how many bytes they use, but on preview or save they are truncated to use at most 255 bytes. The situation is confusing for users without JavaScript but I don't know whether it's possible without JavaScript to make an input field which counts bytes and not characters. The 200 limit without JavaScript is a compromise: Large enough for most edit summaries, and limits the number of times an edit summary must be truncated because it breaks the 255-byte limit by having too many multi-byte UTF-8 characters. Languages with non-Latin scripts have more problems with this. PrimeHunter (talk) 10:52, 12 October 2015 (UTC)[reply]
Yes users with Javascript will get a better experience, because this check is not possibly with plain HTML. —TheDJ (talk • contribs) 11:12, 12 October 2015 (UTC)[reply]
How do I turn off JavaScript in Firefox 41.0.1? --Redrose64 (talk) 20:51, 12 October 2015 (UTC)[reply]
Tools/Add-ons - Click on Disable next to NoScript. Then you have to close your browser and restart it for the disable to take affect. — Maile (talk) 20:57, 12 October 2015 (UTC)[reply]
It's not there. I only have two entries: Microsoft .NET Framework Assistant (disabled) and RealDownloader (disabled). --Redrose64 (talk) 22:58, 12 October 2015 (UTC)[reply]
NoScript is a Firefox extension Maile has installed. Most Firefox users don't have it. It has controls to allow or disallow JavaScript for each website. PrimeHunter (talk) 23:07, 12 October 2015 (UTC)[reply]
In about:config there is javascript.enabled. --Edgars2007 (talk/contribs) 21:01, 12 October 2015 (UTC)[reply]

"jump up to ^"

Sometimes when I look at articles and see the citations at the bottom, instead of just having the "^" caret, it will say before it, linked in blue) "jump up to ^", next to every citation. It's ugly and quite redundant. Is this a Wikipedia issue or is it on my end? I use Firefox on a Macmini. 108.54.167.196 (talk) 15:14, 12 October 2015 (UTC)[reply]

WP:Village pump (technical)/Archive 140#Jump up!. --Edgars2007 (talk/contribs) 15:20, 12 October 2015 (UTC)[reply]

Florence Rush

I added an infobox to Florence Rush and the entire article disappeared (diff), I've reverted for now, can someone please take a look? --The Vintage Feminist (talk) 15:52, 12 October 2015 (UTC)[reply]

 Done. You had left in the opening part of an html comment before the death date and age template, The Vintage Feminist, but not the matching close, so everything after that, including the template close, was swallowed in the comment, leaving only an unclosed template call. DES (talk) 15:59, 12 October 2015 (UTC)[reply]
Oops, okay thanks. --The Vintage Feminist (talk) 16:08, 12 October 2015 (UTC)[reply]

Tech News: 2015-42

Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.

Recent changes

  • The database size lists have been updated. These control special page update frequency and which wikis use global abuse filters. [3]

Changes this week

  • The new version of MediaWiki will be on test wikis and MediaWiki.org from October 13. It will be on non-Wikipedia wikis from October 14. It will be on all Wikipedias from October 15 (calendar).
  • You will be able to upload images to Wikimedia Commons using the visual editor. When the image is uploaded it will be added to the article you're editing. [4]
  • Pages that show citation error messages will automatically be placed in a hidden category. [5]

Meetings

  • You can join the next meeting with the VisualEditor team. During the meeting, you can tell developers which bugs you think are the most important. The meeting will be on 13 October at 19:00 (UTC). See how to join.

Tech news prepared by tech ambassadors and posted by bot • Contribute • Translate • Get help • Give feedback • Subscribe or unsubscribe.

16:29, 12 October 2015 (UTC)

Blank image

File:GrahamCube.svg has started rendering blank for no apparent reason. It used to be ok in the article. Clicking through to the image it is still there and no new versions have been uploaded. SpinningSpark 21:37, 12 October 2015 (UTC)[reply]

I've fixed the file by adding the type="text/css" attribute to the <style> element. Unlike in HTML, that attribute seems to be required for CSS stylesheets to work in SVG. SiBr4 (talk) 21:51, 12 October 2015 (UTC)[reply]
Funny that it used to be ok though. Something must have changed somewhere. SpinningSpark 22:04, 12 October 2015 (UTC)[reply]
Yes, the documented behaviour is that in the absence of any other indication, the default is "text/css". --Redrose64 (talk) 23:26, 12 October 2015 (UTC)[reply]

Article Feedback Activity Log erased

Some entries in Special:Log have only a single word after the user page, talk page, and contribs links. Those entries were in the Article Feedback Activity Log, and had some text that got erased after AFT became obsolete on March 3, 2014. The entries that were in the Article Feedback Activity Log should be completely erased. GeoffreyT2000 (talk) 01:55, 13 October 2015 (UTC)[reply]

Because the Article Feedback extension was uninstalled, its log entry formatter is no longer available, so these logs are presented in a pretty useless raw form. The relevant Phabricator task is phab:T64722. — This, that and the other (talk) 06:10, 13 October 2015 (UTC)[reply]

Cascading protection glitch

Tracked in Phabricator
Task T115586

Are users here who are template editors but not admins able to edit Template:Center? I ask because I appear to have caused a glitch which makes it cascade-protected from a deleted page. Here's how it happened:

  1. I notice that when I edited Template:Center, there was a notice above the edit box that said it was cascade-protected from File:Onion Powder, Penzeys Spices, Arlington Heights MA.jpg. I decide that this protection isn't serving any purpose, as the file is no longer visible on the main page.
  2. I delete File:Onion Powder, Penzeys Spices, Arlington Heights MA.jpg without unprotecting it. I thought this would remove the protection and make pages fetch the file from Commons, all in one edit.
  3. I reduce Template:Center to template-protection.
  4. Template:Center is now displayed as template-protected, but when editing it the notice that says that it is cascade-protected is still there. The difference is that instead of File:Onion Powder, Penzeys Spices, Arlington Heights MA.jpg being listed as a source, there are no sources listed.

I'm guessing that this will have to go to Phabricator, but if there's a way we can resolve it here, that would be nice. — Mr. Stradivarius ♪ talk ♪ 11:09, 13 October 2015 (UTC)[reply]

To me it shows the "template protected" message now. Not a template editor or administrator.Jo-Jo Eumerus (talk, contributions) 11:11, 13 October 2015 (UTC)[reply]
I think that the glitch may have defeated the protection level detection code that makes the "template protected" message, as it relies on the CASCADINGSOURCES magic word producing some text, but if there are no sources then the output will just be blank. I am also guessing a similar thing happened with the Lua code that checks for cascading protection to see what colour the padlock icons should be. — Mr. Stradivarius ♪ talk ♪ 11:19, 13 October 2015 (UTC)[reply]
Also, I should add that after I deleted File:Onion Powder, Penzeys Spices, Arlington Heights MA.jpg I restored it, then tried cascade-protecting it and then unprotecting it again, but it didn't seem to make any difference. — Mr. Stradivarius ♪ talk ♪ 11:12, 13 October 2015 (UTC)[reply]
I placed the "Template editor" right on an alternate account of mine and loaded the edit page. I did get the page (I didn't actually try to change anything there), along with the warning that only admins can edit it due to cascade protection. עוד מישהו Od Mishehu 11:37, 13 October 2015 (UTC)[reply]
For me (TE) it displays the "cascade-protected" warning alongside the template-protection one, though it seems I am able to edit the template normally: the edit box is pink and unlocked, and there are "save"/"preview"/"changes" buttons, unlike when I try to edit a fully protected page. SiBr4 (talk) 13:39, 13 October 2015 (UTC)[reply]
Phabricator would be the place for this, yes. Checking the database, what's going on is basically what you guessed: the deletion of the cascade-protected page should have deleted the protection entry (and other links table entries), but it did not. Unfortunately undeleting the page would have no effect, since the links table entries are tied using the page_id value and undeleting assigns a new page_id rather than reusing the old page_id (T60986 seems to be commenting on that situation). I've fixed this specific issue by deleting the problematic database row, and I don't see any others on enwiki. See T115586 for tracking the underlying cause of the issue. BJorsch (WMF) (talk) 13:08, 15 October 2015 (UTC)[reply]

WP:FLAGCRUFT malicious?

I have recently inserted the WP:FLAGCRUFT js code, and found out it could be malicious. Simple question here: is it really that bad or is there really nothing to worry about here? --JB82 (talk) 14:49, 13 October 2015 (UTC)[reply]

What do you mean by "found out it could be malicious"? If you mean you see MediaWiki:Jswarning on your css and js pages then it's a standard warning always shown on those pages regardless of content. PrimeHunter (talk) 15:41, 13 October 2015 (UTC)[reply]
Adding JavaScript code to your user JS pages can do pretty much anything to your account (e.g. cause it to make edits, etc.). As such, it's best to ensure that any scripts you place on the page are written by someone trustworthy (or alternatively by an administrator; there's no loss of security there because administrators can edit other people's user JS pages anyway). The message there is just a scary warning to try to prevent people from installing dubious scripts they found outside Wikipedia. --ais523 21:55, 14 October 2015 (UTC)

Searching for a template following a string

I tried asking a question at AWB tasks, but that may not be the best place (or maybe it was TLDR).

I want to use AWB to improve some references. I can find each of the references because they are all named references, but the field to be replaced is not necessarily unique, mainly because the access date might vary. My goal is to search for a particular named reference, then locate the immediately following cite web template, and replace the contents of that template with a value.

I'm not quite sure how to do the specification in AWB (or if there is some better way to do it).

I've started the creation of a table to identify the replacement strings here (The table currently has only five entries but there will be dozens when I complete it).

Can someone tell me if it is possible to search for a template following a string, or if there is a better way to approach this?--S Philbrick(Talk) 13:50, 14 October 2015 (UTC)[reply]

Is the opening reference tag always in the form <ref name="name">? Are you replacing the {{cite web}} template in toto? then:
find:
(\<ref\s+name\s*=\s*"name"\s*\>)\{\{\s*cite\s+web[^\}]+\}\}
where "name" is the reference name
replace:
$1{{cite web |...}}
where ... is the full content of replacement {{cite web}} template
No doubt, this could be optimize but should should work as a starting point. If the {{cite web}} template contains another template, this regex will fail.
—Trappist the monk (talk) 14:06, 14 October 2015 (UTC)[reply]
It's better to use [^\{\}]+ instead of [^\}]+ in cases like this. That way, if the template you're replacing does contain a nested template, the regex doesn't do anything, instead of replacing the wrong things and breaking stuff. SiBr4 (talk) 14:19, 14 October 2015 (UTC)[reply]
Thanks for the very quick responses. Yes, the opening reference tag is always in that form. Yes, I will be replacing the template in toto. I do not believe I have any nested templates but I will either use the second suggestion for code or watch carefully.--S Philbrick(Talk) 14:44, 14 October 2015 (UTC)[reply]
I tried it and it worked (at least in the first case). Thanks.--S Philbrick(Talk) 15:00, 14 October 2015 (UTC)[reply]

Good news, bad news

@Trappist the monk:

The good news is when I set up AWB and ran against the list of World University Game references I got over 50 hits which is about what I expected.

The bad news is that when I tried to repeat for Pan Am games I got only two hits, while expecting over 50. If I had zero hits I would assume I set up something fundamentally wrong but I get two which suggests the format is correct.

An example of my search for string is:

(\<ref\s+name\s*=\s*"2003 Pan Am"\s*\>)\{\{\s*cite\s+web[^\}]+\}\}

The replacement string is:

$1{{cite web|title=Fourteenth Pan American Games -- 2003|date=Feb 20, 2014|url=http://www.usab.com/history/pan-am-womens/fourteenth-pan-american-games-2003.aspx|publisher=USA Basketball| archive-url =https://web.archive.org/web/20150907195503/http://www.usab.com/history/pan-am-womens/fourteenth-pan-american-games-2003.aspx| archive-date =7 September 2015|dead-url=no|accessdate=15 Oct 2015}}</ref>

An example of the page in the list is Rebekkah Brunson; the relevant reference is:

<ref name="2003 Pan Am">{{cite web|last=|first=|date=|title=FOURTEENTH PAN AMERICAN GAMES -- 2003|url=http://www.usab.com/womens/panamerican/wpag_2003.html|work=|publisher=USA Basketball|accessdate=15 Oct 2013}}</ref>

I am a REGEX newbie, but I'm not seeing why that search should not find this string.--S Philbrick(Talk) 17:38, 15 October 2015 (UTC)[reply]

When I put your search and replacement strings into AWB and have it search Rebekkah Brunson, it finds and replaces the middle reference in §References (I did not save). If it isn't broken, I can't fix it. Have you inadvertently set something that you should not have set on the Skip tab? What about the case sensitivity checkbox in Options -> Normal settings?
—Trappist the monk (talk) 19:45, 15 October 2015 (UTC)[reply]

Why can't I see the deleted version of David Godman

Wikipedia:Articles for deletion/David Godman shows this article was deleted by AfD in July 2008. It was recreated in October 2008 but I can't find the deleted versions. Doug Weller (talk) 14:11, 14 October 2015 (UTC)[reply]

It appears that the article was moved to User:Iddli/David Godman. Elockid(Boo!) 14:18, 14 October 2015 (UTC)[reply]
That means then that it wasn't deleted in July, I guess. Pretty promotional. Doug Weller (talk) 15:03, 14 October 2015 (UTC)[reply]
Click "View logs for this page" at top of the page history to see [6] which shows it was deleted in July 2008, and in October 2008 it was restored, moved to User:Iddli/David Godman, and the resulting redirect was deleted. The recreation was in November 2008‎. It was later deleted again and then restored including the October 2008 redirect from the move, which is why the visible page history currently starts there. PrimeHunter (talk) 16:35, 14 October 2015 (UTC)[reply]
Thanks PrimeHunter - I don't think that's a very satisfactory way of handling an article, especially as it lost all its attribution history before October 2008. Doug Weller (talk) 17:38, 14 October 2015 (UTC)[reply]
I've done the hist merge, so all the history is at David Godman now. —SpacemanSpiff 17:46, 14 October 2015 (UTC)[reply]

Apostroph problem

Not sure if this is the correct place to report this. I noticed a redlink in the second section of the HMS Chevron article. I copied the word (Ma’apilim) into the searchbox, and from there created a new article and made it a redirect to where other versions of this word, such as Maapilim go. It's still red. Apparently that specific apostrophe (’) gets substituted to another kind (') from search but not from links in articles. No More Mr Nice Guy (talk) 17:01, 14 October 2015 (UTC)[reply]

The best way to create a new article from a red link is to click on it. No need to use search. Ruslik_Zero 18:13, 14 October 2015 (UTC)[reply]
Generally, we should use straight apostrophes instead of curly ones. You should correct the text in the article. But in my opinion, it would make sense if the software treated straight and curly marks as the same for linking. I think it's a similar problem for capitals. Both cases are treated the same in the search, but in links they're not unless they're at the start of the word (or at least, that's how it used to be). McLerristarr | Mclay1 18:14, 14 October 2015 (UTC)[reply]
@No More Mr Nice Guy: I cannot reproduce this in Firefox. If I enter Ma’apilim in the search box and choose "containing..." (which is now needed to avoid the redirect you created at Ma'apilim) then I get a search results page [7] saying: You may create the page "Ma’apilim". Clicking the red link leads me to a page creation window with the curly apostrophe and not to the redirect you created. What is your browser and what do you see at [8]? PrimeHunter (talk) 22:27, 14 October 2015 (UTC)[reply]
I was using Chrome but I get the same behavior in Firefox. If I hit "containing..." I get the same results as you describe. If I hit enter it takes me to the redirect with the substituted apostrophe. I could create another redirect that would solve the redlink problem, but this behavior is not what a user would expect, I think. No More Mr Nice Guy (talk) 22:54, 14 October 2015 (UTC)[reply]
I still don't understand how you created a redirect with a straight apostrophe after entering a title with a curly apostrophe in the search box. Let's try an example where neither version matches a current page name, like your original case before creating the redirect. If I enter Ma’ap (curly apostrophe) in the search box then I get: You may create the page "Ma’ap". If I click the red link then I go to https://en.wikipedia.org/w/index.php?title=Ma%E2%80%99ap&action=edit&redlink=1 with a curly apostrophe %E2%80%99 in the url. If I enter Ma'ap (straight apostrophe) in the search box then I get: You may create the page "Ma'ap". If I click the red link then I go to https://en.wikipedia.org/w/index.php?title=Ma%27ap&action=edit&redlink=1 with a straight apostrophe %27 in the url. Do you get the same result as me in both cases, or do you get the version with the straight apostrophe regardless whether you entered a curly or straight apostrophe in the search box? I have tested it in Chrome 46.0.2490.71 on Windows Vista and I get the same result as in Firefox. PrimeHunter (talk) 23:14, 14 October 2015 (UTC)[reply]
I get the same result. I don't remember what I did, to be honest. It's possible I didn't notice the curliness at first, I guess, so I typed it in rather than copied it? I'm fairly but not 100% sure I copied it. No More Mr Nice Guy (talk) 23:56, 14 October 2015 (UTC)[reply]

Why do all pages look blank?

Tracked in Phabricator
Task T115505

Whenever I go to any page, it looks like there is no content in the page at all (i.e. it looks like a blank page). However, when I click "Edit", I see the wikitext. What happened? Gparyani (talk) 18:06, 14 October 2015 (UTC)[reply]

Something went wrong with the weekly software update. It was immediately reverted and stuff should get back to normal now. —TheDJ (talk • contribs) 18:08, 14 October 2015 (UTC)[reply]
I saw this for a moment: File:Missing_text.PNG. I also saw "Due to high database lag, changes newer than 170 seconds may not appear in this list" on Special:MyContributions. Gparyani (talk) 18:10, 14 October 2015 (UTC)[reply]
Seems to be effecting User pages in particular. Blethering Scot 18:11, 14 October 2015 (UTC)[reply]
A few minutes ago, I started having the same problem in Google Chrome with the idea lab, which I didn't have an hour earlier. I actually posted this answer from Mozilla Firefox. Blackbombchu (talk) 18:13, 14 October 2015 (UTC)[reply]
That problem has already ended for me. Blackbombchu (talk) 18:17, 14 October 2015 (UTC)[reply]
A significant number of pages have been cached as blank during the outage. A WP:PURGE will get it back to normal. I've had to do a revert of a vandalism bot once for this, too. Mamyles (talk) 18:15, 14 October 2015 (UTC)[reply]
(edit conflict) David Godman seems to be affected...was about to use CSD A3 but stopped. Gparyani (talk) 18:16, 14 October 2015 (UTC)[reply]
I've just purged that page. Should be back up. Mamyles (talk) 18:16, 14 October 2015 (UTC)[reply]
Thanks for reporting this. This is currently being handled and the corresponding bug report is phab:T115505. As more information on the reasons becomes available that bug report will receive updates. Sorry for the inconvenience! --AKlapper (WMF) (talk) 18:17, 14 October 2015 (UTC)[reply]
It appears to be affecting every page that has been updated since the problem started. Belarusian presidential election, 2015, EMC Corporation, and Roses Are Red (My Love) are all affected, just to name a few. Portal:Current events also appears completely blank for me.--Tdl1060 (talk) 18:23, 14 October 2015 (UTC)[reply]

The Article Traffic Statistics tool need to be updated badly

Hi my name is SeminoleNation and I have been wondering if Wikipedia plans on implementing a stable way of looking at article traffic statistics. The current http://stats.grok.se/ link is very unstable and frequently crashes and will lose information of multiple days at a time. Thank you--SeminoleNation (talk) 21:51, 14 October 2015 (UTC)[reply]

Hmm, Analytics folks might be the best to ask. You could ask on the Analytics mailing list for example. --AKlapper (WMF) (talk) 10:21, 15 October 2015 (UTC)[reply]

Pages without JS or CSS via Google

Some of you might be interested in this: Especially for slow connections, Google sometimes sends a stripped-down version of a webpage to users, with limited Javascript and CSS. You can see what it looks like here with the article about Oxygen: https://icl.googleusercontent.com/?lite_url=http://en.m.wikipedia.org/wiki/Oxygen

I believe this only happens on the mobile site, but I'm not certain of that. Whatamidoing (WMF) (talk) 00:08, 15 October 2015 (UTC)[reply]

Stats down again

Pageview stats at http://stats.grok.se are down again. Stats have not compiled since October 11. Thus, the following is a summary the datefiles that are currently not compiled.

Template:Multicol

Complete datafiles in need of compiling
  • January 31, 2008
  • February 28, 2008
  • March 1, 2008
  • June 1–2, 2008
  • July 1, 2008
  • July 13–31, 2008
  • October 20, 2008
  • November 15, 2009
  • June 26, 2010
  • September 2, 2011
  • October 20, 2011
  • April 30, 2012
  • February 5, 2015
  • October 12-14, 2015

Template:Multicol-break

Partially complete (between 1 and 23 hourly files have data) datafiles which should be considered for compiling
  • March 3–4, 2008
  • October 21, 2008
  • October 14, 2009
  • October 16, 2009
  • November 22, 2009
  • January 23–24, 2010
  • February 8, 2010
  • June 28, 2010
  • July 5, 2010
  • July 7, 2010
  • July 10, 2010
  • July 23, 2013
  • January 6, 2014
  • August 28, 2014

Template:Multicol-break

dates without datafiles that can not currently be compiled
  • September 23, 2009
  • September 25–27, 2009
  • October 15, 2009
  • July 8–9, 2010
  • December 24–25, 2011

Template:Multicol-end

See update above.--TonyTheTiger (T / C / WP:FOUR / WP:CHICAGO / WP:WAWARD) 04:11, 15 October 2015 (UTC)[reply]

Long pages data needs updating

In the absence of User:Snaevar, how can we get Wikipedia:Database reports/Long pages updated? the data there is now around two months old, and very inaccurate. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:19, 15 October 2015 (UTC)[reply]

Here you are. Should be the same criteria. If (when) you need an update, just say. --Edgars2007 (talk/contribs) 15:01, 15 October 2015 (UTC)[reply]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia:Village_pump_(technical)&oldid=685912021"