Wikipedia:Village pump (technical)

From Wikipedia, the free encyclopedia
  (Redirected from Village pump (technical))
Jump to: navigation, search
  Policy   Technical   Proposals   Idea lab   Miscellaneous  
Shortcuts:
The technical section of the village pump is used to discuss technical issues about Wikipedia. Bugs 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.

« Older discussions, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134
Centralized discussion
Proposals: policy other Discussions Ideas

Note: inactive discussions, closed or not, should be archived.

Contents


Knowledge of personal pronoun preference data base[edit]

There is an interest in tracking progress toward gender equality by dynamically updating statistical measure of self-identified "gender pronoun usage preference" on user pages to confirm trend of progress to achieve gender parity. This is likely admin only access to technical user data. Does anyone know how easy/how difficult it would be to gather and then print out these statistics? LawrencePrincipe (talk) 21:16, 13 February 2015 (UTC)

@LawrencePrincipe: Actually, you can access it using the "gender" magic word like so: {{gender:username|he|she|unset}}
So if you enter {{gender:Philosopher|he|she|unset}}, you get he, indicating hat I have my preferences set to "he". I do the same thing to you and see that yours is "unset". – Philosopher Let us reason together. 00:14, 14 February 2015 (UTC)
@Philosopher: On a one-by-one basis that works, though the stats are needed system wide on a weekly and/or monthly basis. That is, out of the total number of new users per month, what ratio self-identified as female and what ratio self-identified as male system-wide. Can it be done system-wide for being dynamically updated? LawrencePrincipe (talk) 00:35, 14 February 2015 (UTC)
Listed at Wikipedia:Database reports/User preferences but out of date. --  Gadget850 talk 00:53, 14 February 2015 (UTC)
@Gadget850: Those stats are quite fine, but quite dated: "User preferences statistics; data as of 23:12, 19 June 2014 (UTC)." Is there a way to update them, or get a month-by-month graph? LawrencePrincipe (talk) 02:01, 14 February 2015 (UTC)
Looks like this was run on the old Toolserver. I don't know of a replacement. --  Gadget850 talk 02:23, 14 February 2015 (UTC)
Note that this data is very unlikely to be useful or meaningful. The preference is designed to be used for gendering language correctly (not an issue in English) and not for demographic tracking; most users will never set it. Andrew Gray (talk) 15:28, 14 February 2015 (UTC)
@Andrew Gray: The gender stats from the UN study from 2010 are the only ones currently is use and are dated. Although not optimal, looking at the "user preference choice of gender" offers a useful approximation. It is not optimal, but useful for gaining some insight into the current status of the gender stats. Are you familiar with the new Toolserver applications and data base access? LawrencePrincipe (talk) 16:02, 14 February 2015 (UTC)
FWIW, this is the WMF Analytics discussion from last year on whether it was likely to be useful data. Andrew Gray (talk) 19:51, 14 February 2015 (UTC)
Hi Andrew Gray; That conversation between Ryan Kaldari, Aaron, and Steven Walling at WMF Analytics was interesting on gender preference data being used in a sub-optimal way of getting gender stats updated from the old 2010 UN data. If anyone has their user id name, then I could ping them to try to find someone who knows the new Toolserver. The link just provided above here by Gadget850 was strongly on target with women editors estimated at 19%, which is actually quite close to the 2010 UN numbers. If anyone knows the new Toolserver well enough, or has the user id for any of the above (Ryan, Aaron, and Steve Walling) then I could ping them. LawrencePrincipe (talk) 20:46, 14 February 2015 (UTC)
I'm not sure what you mean by "knows the new toolserver", but if you have an idea of what the old SQL query was you can try it at quarry - no need for a toolserver/labs account. I would strongly stress that the 19% is probably as much random chance as meaningful data, though. Andrew Gray (talk) 20:56, 14 February 2015 (UTC)
Does someone know how to update this link: Wikipedia:Database reports/User preferences. LawrencePrincipe (talk) 21:22, 16 February 2015 (UTC)
@LawrencePrincipe: The code that generates this report is at Wikipedia:Database reports/User preferences/Configuration. Looks like it comes from the Toolserver era, so it's going to need adjustments, but you should be able to run the SQL queries at least without modifications on Quarry (mentioned above). Perhaps you can get MZMcBride (its author) to make it work. Matma Rex talk 12:46, 17 February 2015 (UTC)
@Matma Rex: MZMcB wrote that for the old toolserver, and what is needed to maintain stats now is if it can be updated/converted to the new toolserver to update the stats. Any way that this can be done? LawrencePrincipe (talk) 01:53, 18 February 2015 (UTC)

(un-indent) phabricator:T60196 is what's blocking us here. --MZMcBride (talk) 20:03, 19 February 2015 (UTC)

@MZMcBride:@Rich Farmbrough: That issue seems to have been addressed by @Rich who appears to be able to find a way to get the gender stats on a monthly basis (see section immediately below). The last update to comments on the phabricator link you just provide was from 14Dec which has not been updated since then. Any chance of getting the month-by-month gender preference stats for the last three or four monthly intervals? LawrencePrincipe (talk) 00:09, 20 February 2015 (UTC)

Section break for Wikimedia research stats[edit]

There was a user survey done by the WMF, the results of which have not been released yet. There is a paper which describes identifying users gender by name as a supplementary process (I forget which paper). There is also some work on gendered linguistic differences on Wikipedia, and the conclusion I took from that was that it is not valid to extrapolate gender tells from general Internet discourse to Wikipedia discourse without supporting evidence. I also analysed some small samples of text on a standard "gender tester" with poor results. However there is no reason that a full text analysis of all talk page contributions could not be done, and it might well lead to a model with good gender predictive power, since there is a valid training set consisting of the gender identified users. All the best: Rich Farmbrough11:46, 18 February 2015 (UTC).

@Rich Farmbrough: That paper you mention would be interesting to locate if you have any leads. The issue of using the new toolserver to count up the number of new accounts which self-identify as female seems to be difficult to accommodate, even though the old toolserver apps did this with ease. The research apps you describe above sound pretty advanced, though the simple counting of new accounts which self-identify as female would offer a first step if someone could figure out the new toolserver. LawrencePrincipe (talk) 15:18, 18 February 2015 (UTC)
The last 5000 new accounts (that have edited) were 9 male, 1 female, 4990 unset. All the best: Rich Farmbrough16:13, 18 February 2015 (UTC).
Thanks, Rich, that is important to keep in mind (also, specifying user gender is only possible since october 2011):
"However there is no reason that a full text analysis of all talk page contributions could not be done, and it might well lead to a model with good gender predictive power, since there is a valid training set consisting of the gender identified users." - Except that this would be freaking creepy! --Atlasowa (talk) 16:37, 18 February 2015 (UTC)
Declared: male = 502573, female=98369 , female = 16.40% of those with declared gender.
Admins: female=56, male= 600.
All the best: Rich Farmbrough17:04, 18 February 2015 (UTC).
@Rich Farmbrough:@Atlasowa: Those are remarkable gender stats, but what does 500,000 male editors refer to? The Economist newsweekly has stated here [[1]] that there is a total of 50,000 editors in 2007, which went down to 30,000 editors in 2014. What does your 502,573 number mean, and is it related to the 2011 date cited by @Atlasowa? Maybe there is a simple explanation. LawrencePrincipe (talk) 17:50, 18 February 2015 (UTC)
References
  1. ^ "The future of Wikipedia: WikiPeaks?". The Economist. March 1, 2014. Retrieved March 11, 2014. 
The vast majority of the 20,555,036 Wikipedia accounts have never edited (or have no edits visible in their contributions). There are several obvious reasons for these accounts, and some not so obvious:
  1. Someone sees some vandalism and creates an account, but it is fixed by the time they get there.
  2. Someone creates an account but finds editing too hard, changes their mind, gets distracted (ooh look a butterfly!)
  3. Accounts created to protect a name (e.g, User:Placeholder, and WP:DOPPELGÄNGER accounts) or to reserve a name
  4. Accounts who try to vandalise and give up because the edit filters won't let them
  5. Accounts that created an article which was later speedy deleted
  6. Accounts where the password was forgotten, or set wrongly
  7. Sleeper socks
and probably a few more.
And the other figures are based on "active editors" which has various definitions, usually in terms of X edits per month or year.
All the best: Rich Farmbrough18:49, 18 February 2015 (UTC).
LawrencePrincipe, the Economist refers not really to "total editors" but to "active editors" (Active Editors = Registered (and signed in) users who made 5 or more edits in a month), see english WP summary stats, 2007: ~50,000 active editors and 2014: ~30,000 active editors. Same trend on other big WP, german WP summary stats 2007: ~9,000 active editors and 2014: ~6,000 active editors. Total (ever) Registered users on enWP: 24,107,661, that makes 500,000 male editors a very small part. Same for admins: Out of 1,363 enWP Administrators, 56 are female and 600 male. --Atlasowa (talk) 19:28, 18 February 2015 (UTC)
@Rich Farmbrough: The comments all appear to be in agreement. The stats would be interesting to see for results for the last 10,000 editors and the last 15,000 editors. Are they all consistent at about 10% gender pronoun usage? @Atlasowa: The Economist article also states that non-English Wikipedia is stable at about 42,000 active editors, up or down 2000 for seasonal variation. The 'active' editor number appears to be a significant one for all the reasons that Rich has stated above. Can the stats and tools you provided be used as reliable data to be published from Wikimedia? LawrencePrincipe (talk) 21:41, 18 February 2015 (UTC)
The Quarry tools are there, and I might be able to make something better. (Well I am able, but I need to brush up my SQL.)
The 5k figure I can potentially get you the 10k and 15k figure. To replicate those figures, it is important to state the time by which the accounts needed to have edited to qualify. Probably simpler just to list the 15k accounts.
I do agree with Steve Walling, though, that the declared numbers are a poor proxy. For example GamerGate may have made females lees willing to self-identify (or it may not.. but it's a potential confounding factor.)
All the best: Rich Farmbrough23:36, 18 February 2015 (UTC).
@Rich Farmbrough: All those stats would be useful since at present there is no new data on gender stats since the 2010 UN study. If you say "it is important to state the time" for the 5k, 10k, 15k then it may be just as easy to double the time interval to indicate how many women registered (approx. 10k level), and the triple the time interval to approx. the 15k level, if that seems reasonable. The actual number for each time interval would tell the story. Regarding your other question about women "GamerGate" concerns, my understanding is that if women were given confidence that self-identifying gender would help them to better achieve gender parity, then they would have every incentive to do so. Historically, a chance to achieve goals is often a high incentive to participate. LawrencePrincipe (talk) 00:31, 19 February 2015 (UTC)
Absolutely, it's the case that Editathons based on improving coverage (achieving gender parity) on women's history, women's literature etc. are very successful in attracting participants (of both genders) I believe for this reason. However the motivation to edit is generally not related to gender parity, and those who are really passionate about this tend to think (perhaps rightly) that their time is best spent encouraging other women and girls to edit.
As far as the numbers go 5,000 editors is about 2 or 3 days worth. But I can take samples at monthly intervals, or some such. All the best: Rich Farmbrough00:52, 19 February 2015 (UTC).
If the monthly stats are a convenient time interval then by all means it would be useful to see it, for a few monthly intervals if possible. It is not clear to me why this data is being called unreliable by different editors. This is the same data source for counting the number of new editors and number of edits per week, per month, etc, with no-one thinking these data counts as being unreliable? Why call them unreliable for counting personal pronoun preference selection on user pages? LawrencePrincipe (talk) 20:37, 28 February 2015 (UTC)
It's unreliable because it's not complete. Making your gender public is optional: those I have met in real life know my gender, but I can (and do) hide it from those who only know me electronically: {{he or she|Redrose64}} → he or she. I expect that there are many editors who have not set their gender simply because they don't know that it is possible for them to do so - and some may have set it without realising that it's not compulsory. By contrast, the number of edits that I make is permanently recorded, and moreover is something that I cannot hide. --Redrose64 (talk) 20:23, 1 March 2015 (UTC)

I have produced a little graph at Wikipedia_talk:WikiProject_Countering_systemic_bias/Gender_gap_task_force#F.2FM_ratio_of_gender_identifying_editors_on_en:WP which shows the gender ratio of new gender identifying editors month by month. All the best: Rich Farmbrough01:16, 4 March 2015 (UTC).

Unlabeled star is meaningless to newcomers and many advanced editors[edit]

At the top of my page there is an unlabeled star. When I hover over it I get a message stating, "Add this page to your watchlist." Why doesn't it bear a label so editors will know what it is supposed to do? I know (now), but for a long time I didn't. While I'm at it, there is a label marked "TW." I am surmising that this is short for "Twinkle," but really I have no idea what Twinkle is. Why must we have such ingroup labels at the tops of our pages? Sincerely, GeorgeLouis (talk) 06:51, 25 February 2015 (UTC)

That star is part of the "Vector" skin (explicitly designed to be compact); it's not found in the Monobook skin - the word "watch" is. עוד מישהו Od Mishehu 08:17, 25 February 2015 (UTC)
Also, there's a tooltip displayed when hovering over the star, which describes its purpose. — Dsimic (talk | contribs) 08:22, 25 February 2015 (UTC)
I would say, what's wrong with the star, when you can hover over it to see what it does? And you must have enabled Twinkle for yourself if you see "TW" there. Twinkle is not enabled for all users, only those who choose to enable it. The "TW" abbreviation is used simply to save space, not as an intentional obfuscation measure. — This, that and the other (talk) 13:03, 25 February 2015 (UTC)
The star icon has become a browser standard: Firefox, IE and others use it for bookmarks. The watchlist icon was chosen as an analogue to bookmarks. --  Gadget850 talk 13:24, 25 February 2015 (UTC)
Chrome and IE use it so IMHO it makes sense for us too, As for Twinkle - It's easier and with the greatest respect I'd imagine 90% know what it is so not really seeing any need to change the star or even consider doing so. But that's just my 2¢. –Davey2010Talk 13:38, 25 February 2015 (UTC)
  • I know very well what they both mean, but I do agree if I didn't I would be upset when viewing from my mobile in desktop mode where I can't 'just hover' over either to see. I'm guessing if I was first editing from a tablet I'd be upset too since it is another touch device without hover. I'd suspect that it would be a good idea to include in our vector.css/js a way to determine if the user is on a touch device and 'use our words' instead of symbols. Just my 2cp on it, and I really don't care anyways except I was a little annoyed by the direction of this discussion. — {{U|Technical 13}} (etc) 13:43, 25 February 2015 (UTC)
  • Technical 13 - "little annoyed by the direction of this discussion" - Well I'm assuming it's me - I've read my comment 4 times and still see nothing wrong with it so could you explain what I've said that's ticked you off?.... –Davey2010Talk 14:08, 25 February 2015 (UTC)
  • For once I have to agree with T13 here. Every other function on the Wikipedia page is represented by words, not an icon (at least until you get into edit mode and use the RefToolbar). It's kind of jarring. Besides, mystery meat navigation is never a good thing. By default, I think that Mw-watch-icon.svgWatch and Mw-unwatch-icon.svgUnwatch would be better. --Ahecht (TALK
    PAGE
    ) 14:22, 25 February 2015 (UTC)

I think GeorgeLouis makes a really good point!

1) Watching articles really, really matters. Watching articles, looking at your watchlist is the engine of Wikipedia! If nobody watches articles, they degrade. Recent Changes patrol is not able to be our only quality control, and page watchers are regularly better qualified to assess the merits of an edit. I would even say that it is erroneous to quantify "user activity" or "editor attrition" by number of new user registrations, or by new users with +1 edits, but that "regular users" is the important metric: "users that watch articles". I don't know how to count those users, but for the trends we can take these old stats 2013/2014 about article numbers (in thousands=k) and number of "superuser" (phantasyname for user with >100 edits per month) and calculate the number of articles "to watch" per superuser (in thousands=k) :

wiki language k articles September 2014 k articles July 2013 user >100e/m (mean January–June 2013) user >100e/M (mean June–August 2014)[1] k articles /superuser July 2013 k articles /superuser September 2014
en english 4594 4270 3281 3056 1,3 1,5
de german 1752 1603 1023 846 1,6 2,1
fr french 1540 1402 785 720 1,8 2,1
ru russian 1143 1020 659 519 1,5 2,2
es spanish 1122 1027 499 488 2,1 2,3
it italian 1140 1044 436 361 2,4 3,2
ja japanese 924 864 344 330 2,5 2,8
zh chinese 784 703 281 305 2,5 2,6
pl polish 1062 976 257 215 3,8 4,9
nl dutch 1789 1672 242 203 6,9 8,8
pt portuguese 838 786 188 152 4,2 5,5
uk ukrain 523 451 140 142 3,2 3,7
sv swedish 1891 1158 124 105 9,3 18
he hebrew 161 148 114 116 1,3 1,4
hu hungarian 266 243 114 104 2,1 2,6
ko korean 287 243 99 94 2,5 3,1
cs czech 303 269 93 77 2,9 3,9
ar arabic 318 235 88 93 2,7 3,4
fi finnish 354 327 81 81 4 4,4
fa persian 420 310 75 71 4,1 5,9
ca catalan 436 405 70 65 5,8 6,7
no norwegian (Bokmål) 429 388 69 63 5,6 6,8
tr turkish 234 213 59 59 3,6 4
vi vietnamese 1108 796 50 51 15,9 21,7
bg bulgarian 165 149 47 31 3,2 5,3

You notice the wikis with lots of botgenerated articles, but this is not my point here: More articles over time and not more users over time = problem to watch articles and maintain quality. In July 2013 every "superuser" would have to watch 1300 articles, to have all articles watched (by all 3281 "superusers"). In August 2014 every "superuser" would have to watch 1500 articles, to have all articles watched (by all 3056 "superusers"). We need watchers!

2) There is another aspect, that may be a problem. New users attention could be diverted by the more prominent colourful notifications (introduced in 2013). And while the notifications are about me-me-me (my edits, my conversations, my thanks, my status, my reverts, ...), the watchlist is about the articles, about actual Wikipedia content. In the old days, there was only the watchlist and the users talkpage - now the notifications could distract contributors from that. I found some research about the introduction of notifications in 2013, which seems to support my concerns that the competition for user attention between the "me-me-me-notifications-bling-icon" vs. "static watchlist-link" has harmful effects on productivity: meta:Research:Notifications/Experiment_1#Summary: Our results suggest that the presence of Notifications effectively increases the amount of activity that new users will engage in (more edits, more edit sessions and more hours spent editing). However, the effect of Notifications on the productivity of new users is unclear. On average, users with Notifications were less likely to make productive contributions to articles, but in our experiment, a few highly productive newcomers made up some of the difference. Our results also show that newcomers with Notifications enabled are more burdensome. They made more edits that were reverted by others and they were more likely to be blocked.

3) I think showing the word "watch" (or not) is the kind of small change that you don't notice if you have clicked the watchstar dozends of times, but that can have a real impact for how new users use Wikipedia and how much they are "dragged in" by the watchlist.

4) I think we should do A/B testing to find out if there is an effect for new users. It would be a very small change, there can't be any harm and the test would be limited to a certain period. What do you think?

5) Another way to make people notice the watch-star could be to display the watch-star on the Hovercard (see beta-features Hovercard, really cool feature). --Atlasowa (talk) 17:52, 25 February 2015 (UTC)

@Atlasowa: Today, my watchlist passed 20,000. Will this help lower the minimum for other people? --Redrose64 (talk) 19:42, 25 February 2015 (UTC)
Ha, I like you :-D --Atlasowa (talk) 21:34, 25 February 2015 (UTC)
Redrose64: Just saw that you're no. 419 of Wikipedia:Database_reports/Most-watched users, you're royalty! I definitely think that people should stop watching Little Barrier Island (~4000 watchers, Wikipedia:Database_reports/Most-watched_pages), go watch a bird! (prefererably botgenerated and from Africa) --Atlasowa (talk) 21:51, 25 February 2015 (UTC)
It's out of date. I now have 208 watchers, according to page information. --Redrose64 (talk) 22:01, 25 February 2015 (UTC)
Thank you! Surely we could replace the star with "Watch" or "Unwatch," as the case may be?? 22:47, 25 February 2015 (UTC) — Preceding unsigned comment added by GeorgeLouis (talkcontribs)
Little Barrier Island is the result of pagemove vandalism (see WP:STOCKS); the Main Page got moved to Hauturu/Little Barrier Island some time back, so everyone watching the Main Page ended up with both pages on their watchlists. Probably a lot of the watchers are no longer active, since we don't currently have a ton of active editors who were active when the event happened. Nyttend (talk) 02:32, 26 February 2015 (UTC)
Interesting to see that nobody notices that the word "Watch" outside of the context of MediaWiki is probably less 'informing' and meaningful (so even more mystery meat) than the 'star' icon is to all people. If I remember correctly, that was confirmed by user testing in 2009 and actually led the Vector team to switch away from that word to a more common representation that people know from bookmarks etc. —TheDJ (talkcontribs) 13:27, 26 February 2015 (UTC)
[citation needed] :-) TheDJ: link? --Atlasowa (talk) 22:18, 26 February 2015 (UTC)
https://usability.wikimedia.org/wiki/What%27s_new,_questions_and_answers#Where_did_the_watch_tab_go? Still looking for the data, though. --Atlasowa (talk) 17:39, 27 February 2015 (UTC)

Wow. Many old-timers only dimly remember our pre-WP days, which is why software is tested on the innocent. So, the whole concept of article watching, is what newbies don't get. Once that's understood, the question of star or word becomes a small one. And we've stumbled into a slightly related topic, the poor interface for mobile. That's one reason why I seldom use my Android phone for editing, and not often my iPad. The idea of an intermediate level of interface, showing as much as my real computer does but not relying on hover box or tooltip (the distinction makes little difference) is appealing but our clever and industrious software developers are already 'way behind the likes of Facebook or GMail in serving the various platforms. It seems a bit much to add to their load. Jim.henderson (talk) 13:49, 26 February 2015 (UTC)

We are getting into some serious and arcane in-groupishness here. How and where do I make an official request to replace the star with a word or (better) a phrase? GeorgeLouis (talk) 17:15, 26 February 2015 (UTC)
phab:. --Redrose64 (talk) 17:17, 26 February 2015 (UTC)

"looking at your watchlist is the engine of Wikipedia" I have over 8000 pages on my watchlist. It's not quite as useful as you might image. :-) Maury Markowitz (talk) 00:23, 27 February 2015 (UTC)

Maury Markowitz, I'm dead serious about "watchlists = the engine of Wikipedia". The individual watchlist is a small cog in the machine, but together watchlists are the engine. Because they feed editor attention. We can see the effect of watchlists in "editing burst" at articles. Even after bots: articles show up in watchlists and are checked and edited by other editors afterwards. Editing is contagious :-) --Atlasowa (talk) 18:14, 27 February 2015 (UTC)

mw:Mobile_Beta/Watchlist/Logging#Conclusions :

Account creation hooks like the watchlist star are an effective way to draw more new users to sign up
Over half (56%) of new users creating an account on mobile came directly from the watchlist star CTA. This number is an underestimate of users coming from the CTA, since we were not logging users who first visited the login link from the CTA and then tapped on Create account from the login page.
New users may not understand or value the watchlist star feature as much as existing Wikimedians
The vast majority of one-time watchlist star users were brand-new Wikimedians who had registered via mobile. However, the vast majority of users who tapped the star more than once were existing users who already had a Wikimedia account.
Very few users are unwatching pages from their watchlist view
Either most users are not finding their watchlist view, or the unwatch-from-watchlist feature is not useful for those who do find it.
Overall, the watchlist star hook + account creation has proven to be a very effective new account creation funnel, helping arrest the year-over-year decline of new account creations
Since the full deployment of account creation and watchlist star to the mobile web, new account creations have held steady at about 800 global registrations/400 English Wikipedia registrations per day. These additional account are pushing the hourly registration rates up to 2012 levels.

Maybe we should have:

  • Mw-watch-icon.svgWatch
  • Mw-unwatch-icon.svgUnwatch and above:
  • Watchlist icon.svgWatchlist

And when you click watch, the star starts *spinning* AND above the watchlist icon should *blink* or *glow*, so that you see these things are connected. --Atlasowa (talk) 17:53, 27 February 2015 (UTC)

Automatic capitalization of title-initial Unicode characters[edit]

Has the automatic capitalization of initial letters in titles been changed recently? ɱ is no longer autocapitalized to . Compare this to ɔ, which correctly becomes Ɔ. Gorobay (talk) 15:23, 25 February 2015 (UTC)

Examples, please. Your links are redirects to articles on the characters. --  Gadget850 talk 15:26, 25 February 2015 (UTC)
The capitalized form of ⟨ɱ⟩ is ⟨Ɱ⟩. Therefore, ɱ should be the same page as , but it is not. Similarly, redirects to Insular S, but its lowercase form () does not. Similarly, is an article, but its lowercase form does not exist. That means that some characters are not being capitalized properly. Gorobay (talk) 15:49, 25 February 2015 (UTC)
Look under the title of the target articles — go to ɔ and you correctly see (Redirected from Ɔ) (because it capitalises the letter), but for some weird reason, going to ɱ provides a message of (Redirected from ɱ). Or click the link, to examine the redirect directly; you'll see an uppercase Ɔ and a lowercase ɱ for the page titles. Nyttend (talk) 17:17, 25 February 2015 (UTC)
Couple of years back ɱ (talk · contribs) had a lot of trouble because the system suddenly started uppercasing their username to  (talk · contribs). I believe they couldn't edit their user page or access their contributions, and other things were not possible. --Redrose64 (talk) 17:21, 25 February 2015 (UTC)
It seems that the lowercase user was renamed to uppercase in early 2013, but the lowercase user still seems to exist with zero edits on other projects. Very confusing. --Stefan2 (talk) 17:26, 25 February 2015 (UTC)
Found it, Wikipedia:Village pump (technical)/Archive 105#Link in history says user does not exist. --Redrose64 (talk) 19:34, 25 February 2015 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── Capitalization only works with character pairs encoded in Unicode 4.0. I guess some MediaWiki config file thinks it’s still 2003. However, Utf8Case.ser (the version in MediaWiki 1.25wmf17) maps ⟨ɱ⟩ to ⟨Ɱ⟩, so it should still work. What else controls capitalization? Gorobay (talk) 20:32, 25 February 2015 (UTC)

I tested this on my local test installation of the MediaWiki software and it works fine (the page title "ɱ" automatically transforms to "Ɱ"). There must be something wrong with the way WMF wikis in particular are configured. — This, that and the other (talk) 23:44, 25 February 2015 (UTC)
The function uc (which I assume is the right function) looks for PHP’s built-in mb_strtoupper before falling back to Utf8Case.ser. Wikipedia currently uses HHVM 3.3.1, which includes mb_strtoupper. HHVM 3.3.1 does not support the latest version of Unicode. What version of PHP or HHVM are you using? Gorobay (talk) 04:32, 26 February 2015 (UTC)
Gorobay, you can find a list of (most) version numbers at Special:Version. Whatamidoing (WMF) (talk) 23:47, 26 February 2015 (UTC)
Right, that’s how I knew Wikipedia uses HHVM 3.3.1. I was asking This, that and the other for specifics about their local test. Gorobay (talk) 02:20, 27 February 2015 (UTC)
I use Zend PHP version 5.5.15. Looks like we need to get someone to look at WMF's build of HHVM. I filed phab:T91056 to track the issue. — This, that and the other (talk) 11:22, 27 February 2015 (UTC)

Tables just changed (using monobook)[edit]

I use monobook, and the default cell height and padding seem to have changed. The font size of the text appears to be the same, but the cells are definitely bigger. Where could a change have been made that would affect this for me? -Niceguyedc Go Huskies! 01:59, 26 February 2015 (UTC)

It's the same for me on Vector. Also, the bullets on bulleted lists are smaller (like a middot) and appearing to be at the bottom of the line instead of the usual larger size centered vertically on the line. Imzadi 1979  02:25, 26 February 2015 (UTC)
OS and browser please? -- [[User:Edokter]] {{talk}} 10:21, 26 February 2015 (UTC)
The bullets are now SVG instead of PNG. There is one known issue where Webkit on iOS does not size the bullets correctly. Webkit bug report. -- [[User:Edokter]] {{talk}} 10:28, 26 February 2015 (UTC)
I'm using Safari version 8.0.3 on MacOS X Yosemite (10.10.2). Firefox (v. 14.0.1) does not have the bullet size issue, however the bullets are still slightly off center vertically, putting them slightly too low, but not as low as in Safari where the are at the bottom of the line. In Chrome (v. 40.0.2214.115), the appearance is the same as Firefox: right size but slightly too low on the line. Just for kicks, I also looked in Opera (v. 12.12 ), and it's the same as Chrome and Firefox. The bullets look the same in Safari on an iPad running iOS 8.1.3 as they do on Safari running on the desktop. Imzadi 1979  15:46, 26 February 2015 (UTC)

@Edokter: I'm using Windows 7 and Firefox 36. I don't actually notice anything different with the bullets, just the extra size of rows in tables. -Niceguyedc Go Huskies! 14:06, 26 February 2015 (UTC)

@Imzadi1979, Niceguyedc: The table cells for .wikitable have had their padding increased... doubled even. I found only this commit, but cannot find a accompanying task that discusses this change. (The bullets only affects Vector.) -- [[User:Edokter]] {{talk}} 14:43, 26 February 2015 (UTC)
Yep, the previous padding was pretty tight. You can see examples at testwiki:Wikitable padding. This change shouldn't have broken anything. Please provide links here or file Phabricator tasks if you see issues. --MZMcBride (talk) 15:50, 26 February 2015 (UTC)
While we are on this, where is the repository for shared.css? The catalog still links to SVN, and I can't find main on git. --  Gadget850 talk 16:02, 26 February 2015 (UTC)
[2] this? It's on mediawiki core repo. Glaisher (talk) 16:09, 26 February 2015 (UTC)
Thanks! Since I keep referring to it, created {{shared.css}}. --  Gadget850 talk 16:20, 26 February 2015 (UTC)
Is this a permanent change? I'm working with tables all the time on wikipedia and am quite upset by this change. Bigger tables don't even fot on a page anymore. Personally I think the tables harder to read and it looks untidy. Isn't there any way too make the table cells the size you want? Jahn1234567890 (talk) 00:45, 27 February 2015 (UTC)
@Edokter, MZMcBride: I also agree that the change should be reverted - Jahn1234567890's note on its untidy look is well take. It is also harder to follow visually, especially when zoomed in - as I have to because of vision difficulties - as the distance between the text and the lines is increased. Finally, vertical vs. horizontal whitespace is now unbalanced, as dmeonstrated in the small table at List of Governors of Iowa#Living former governors. – Philosopher Let us reason together. 01:48, 27 February 2015 (UTC)
Noted this through the notably wider infobox element, which contains a 6-column table. (see Uranium). Considerably widened. 1. Not a bad thing per se, except that relative image size now changed (i.e. it did not change absolutely), and there might be a guideline saying something about ideal infobox width ("33% body pagewidth max"?). The infobox now has fewer unforced line-wraps, but its whitespace is growing and growing. I hope this is within grand page-design philosophy. 2. Then I went to check a table that needs all 100% page width: {{Periodic table}} in Periodic table. No effect so OK, because it has cellpadding= set (without this, the wider padding is breaking bad, undesired & useless). This leads to these points: a) are there similar tables (using & needing 100% width) that are gravely affected this way, because they are without cellpadding set?, and b) if I remember correct, cellpadding= is deprecated. Though this depends on it. 3. Btw, is the padding in the infobox enlarged too (mostly visible with section headers)? Conclude: nothing broken so far for me, some questions left. -DePiep (talk) 08:14, 27 February 2015 (UTC)
{{Periodic table}} should not be affected because it's not using .wikitable, not because of the obsolete attributes being used. I did replace those attributes with proper CSS. -- [[User:Edokter]] {{talk}} 09:34, 27 February 2015 (UTC)
I get this. (Had to squeeze out some ws). -DePiep (talk) 10:45, 27 February 2015 (UTC)
What is this central point where deprecated code (like this) is listed? I only meet such statements off and on. -DePiep (talk) 20:28, 27 February 2015 (UTC)
Wikipedia:HTML5. SiBr4 (talk) 21:09, 27 February 2015 (UTC)
My biggest problem with the table change is that this affects every single table on wikipedia. I don't understand why it was changed in the first place. The previous table size was just fine. And if you found that padding too small you could adjust the width by yourself. Now you have no option at all. Jahn1234567890 (talk) 14:08, 27 February 2015 (UTC)
Yes. Why not make it 8px? -DePiep (talk) 20:28, 27 February 2015 (UTC)

Hi Jahn1234567890. You can trivially reduce wikitable padding in your personal user CSS. In general, I think the increased padding looks better. The cited example above (List of Governors of Iowa#Living former governors) looks good to me. --MZMcBride (talk) 20:47, 27 February 2015 (UTC)

Is there a central place where this whitespace philisophy is discussed? 'I think .. looks better' is not enough to me. -DePiep (talk) 20:52, 27 February 2015 (UTC)
I dunno about a central place. testwiki:Wikitable padding is where I did my testing. I don't think anything broke and there seems to be general consensus that the increased padding (0.3em 0.4em instead of 0.2em) is an improvement. There are quite a few studies that suggest that adding whitespace improves readability. Maybe that would be a better argument? --MZMcBride (talk) 20:58, 27 February 2015 (UTC)
I'm lost. "I dunno", "don't think anything broke", "there seems to be general consensus" (where, where, WHERE I repeat). If this is the way classes are set, why are we talking anyway? Who cares. (and let me note that you still did not address my specific points). Make it 8px, I say. -DePiep (talk) 21:04, 27 February 2015 (UTC)
@MZMcBride: For results tables like this Dave Marcis the change makes tables like this look really untidy. Personally I'm unhappy with this change because with results tables in general this change makes those tables unnecessary big. But how would reduce wikitable padding in my personal user CSS? I haven't worked with this before. Jahn1234567890 (talk) 21:28, 27 February 2015 (UTC)
Before I forget. Why was the table padding changed? It was just fine the way it was. I mean this change effects every wikitable on wikipedia. I think doubling the padding is a way to big change Jahn1234567890 (talk) 22:02, 27 February 2015 (UTC)

(un-indent) Hi. gerrit:188311 contains most of the relevant discussion. DePiep: If you want to propose 8px, a Gerrit changeset or Phabricator task is probably what you want. :-) Also, the English Wikipedia's technical village pump is just one of many venues to discuss technical changes. Jahn1234567890: You'll probably want to add the following code (or similar) to User:Jahn1234567890/common.css:

table.wikitable > tr > th,
table.wikitable > tr > td,
table.wikitable > * > tr > th,
table.wikitable > * > tr > td {
    padding: 0;
}

Hope that helps. --MZMcBride (talk) 22:28, 28 February 2015 (UTC)

The concern is not just with Jahn1234567890, there are tables on wikipedia which are now unnecessarily huge for everyone. If there is a consensus to make this sitewide change (other than just MZMcBride and "a friend of his"), could a new class such as wikitablecompact be created that uses the old 0.2em padding? While affected tables could be rewritten to not use the wikitable class altogether (manually setting the style for each individual cell), it would be an major pain and create unnecessarily confusing code. --Ahecht (TALK
PAGE
) 18:25, 1 March 2015 (UTC)
Hi Ahecht. Describing the increased padding as huge seems a bit extreme. I considered having wikitable variants, such as wikitable-compact, but the use-cases seemed to be pretty limited. If you'd like such variants available locally, we can modify MediaWiki:Common.css. This approach would require proposing alternatives and gaining local consensus to support the additional class(es) here. --MZMcBride (talk) 21:24, 1 March 2015 (UTC)
@MZMcBride: The Padding increase is'nt huge to you perhaps, but for a person like me who works with tables all the time it is more than a huge change. All the results tables I made during my stay on wikipedia are now way to huge and untidy when they should be compact. I mean all the work I have done. Thats feels like a slap in the face to me. I honestly have no idea why the padding whould be changed like that. A lot of people are far from happy with this change. And to be honest I haven't seen a lot people who are happy with this change. There are way to many tables that are affected by this (and not for the best most of the time). There has got to be some way to work this out, because I think it is far from unreasonable to keep this change when there is so much objection against this change. Jahn1234567890 (talk) 21:54, 1 March 2015 (UTC)
Any design or user interface change is going to elicit some negative feedback. A few people have complained and there are various options available, including per-user and per-site customizations, as explained above. --MZMcBride (talk) 22:38, 1 March 2015 (UTC)
  • MZMcBride, you keep saying that this change was discussed, and referring to the gerrit page. Well, only developers read gerrit pages. This wasn't discussed with the people or projects that it affects. There's really no difference between a change that affects hundreds of thousands of pages made by WMF staff and changes that affect hundreds of thousands of pages made by volunteer developers: both should be discussed with the affected projects and communities before implementation. Basically, what I see here is one volunteer developer coming up with what he thinks is a cool idea, and having the connections to get it committed and distributed across WMF projects, without having discussed the proposal with anyone outside of the developer stream. That's pretty much the type of action that, when taken by WMF staff, you yourself have decried. I'd have expected better from you. Risker (talk) 22:57, 1 March 2015 (UTC)
    • Hi Risker. Theoretically any change to MediaWiki core affects millions of pages... I don't see your point.

      The increased padding looks better on the whole. If you have proposals for discussing changes like this across 800-plus Wikimedia wikis (in addition to the thousands of non-Wikimedia MediaWiki wikis), please share. If you have specific technical problems to be addressed, please share.

      As it is, you seem to be arguing that a slight padding increase required extensive community consultation, which seems to be pretty insane. Do you really believe that's how development works or should work? --MZMcBride (talk) 23:04, 1 March 2015 (UTC)

      • No, MZM, what I expect is an analysis of the impact of the change and a method of informing groups that will be affected by the change of how it will affect them, and what they can do to mitigate the change. This is especially important when it comes to subjective changes that have not been requested anywhere by anyone and are, essentially, the wish and desire of a single developer. Risker (talk) 23:14, 1 March 2015 (UTC)
        • Well, I am spending volunteer time responding to questions and concerns being raised here. :-) In this case, the change was so small/slight as to not warrant much in the way of notification or contingency plans. There was no expected breakage and the impact to users was considered negligible. --MZMcBride (talk) 23:26, 1 March 2015 (UTC)
        • Hello @MZMcBride: Thank you for your reply. I Figured out how to work on my own CSS page. But somehowe I can't get the skin like it was before the padding change. Could ypu help me out? Jahn1234567890 (talk) 23:22, 1 March 2015 (UTC)
      • I'm sorry, but I disagree witht he change as well. 0.2em to 0.4em is doubling, not "slight", and it shows. Any change with such a visual impact should have been discussed, at least in Phabricator. The perceived improvement is subjective and there was no premise that prompted this change. -- [[User:Edokter]] {{talk}} 23:26, 1 March 2015 (UTC)
        • No need to apologize, disagreement is fine. And the nature of Cascading Style Sheets allows individual sites or individual users to customize their experience by overriding the default padding or creating derivate CSS classes. There was a premise that prompted this change and it's explained in the commit message of gerrit:188311. --MZMcBride (talk) 23:32, 1 March 2015 (UTC)
          • So, shall we set it back to 0.2em on this wiki in Mediawiki:Common.css then? -- WOSlinker (talk) 09:19, 2 March 2015 (UTC)
            • I am very inclined to do so. I will also raise a task and seek to revert it. -- [[User:Edokter]] {{talk}} 10:09, 2 March 2015 (UTC)
          • The premise of "it looks better" is not much of an argument, and considering the flack it has invoked, it was poorly researched for impact. You cannot send user to their personal CSS to 'undo' a change of this nature. What's even more troubling, that if you want to undo it in a table or template, it requires inline CSS for every table cell. -- [[User:Edokter]] {{talk}} 10:09, 2 March 2015 (UTC)

(un-indent) Err, "flack" is a pretty generous term. Maybe a half-dozen people have commented, which is probably about equal to the number of people who reviewed the Gerrit change. It might've been best to not make this edit yourself. Do you know if any other wikis have noticed or cared about this change to the wikitable padding? If the objections are specific to the English Wikipedia, a local override is perhaps for the best. Or perhaps 0.2em 0.4em. :-) --MZMcBride (talk) 05:17, 4 March 2015 (UTC)

Cremation or burial?[edit]

Why can't I delete Cremation or burial?? When I choose delete it just reloads the page. I can delete other pages but not this one. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 07:28, 26 February 2015 (UTC)

Not quite sure. I just did and it worked fine. Something with your browser or personal Javascript maybe? Seraphimblade Talk to me 08:11, 26 February 2015 (UTC)
That's strange I deleted another page with no problem. I'll just put it down to sunspots as they get blamed for almost everything in the Arctic. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 08:19, 26 February 2015 (UTC)
I'm guessing it's a problem with '?' which usually indicates a query string in url's. MediaWiki encodes '?' in pagenames as %3F to avoid confusion but maybe your browser is still confused. Which browser? Does the same happen for Quo vadis? PrimeHunter (talk) 19:30, 26 February 2015 (UTC)
I was using Chrome Version 40.0.2214.115 on Ubuntu 14.04 but I just tried in Firefox 36.0 on the Quo vadis page and got the same result. This time I checked the url and noticed it was https://en.wikipedia.org/wiki/Quo_vadis%3F?action=delete. I'll have to try later on Windows. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 23:01, 26 February 2015 (UTC)
@CambridgeBayWeather: That really is odd, since I'm using Ubuntu 14.10 with Chromium 40.0.2214.111. Are you using Twinkle or any other scripts, or just the regular Mediawiki delete button? Seraphimblade Talk to me 01:22, 27 February 2015 (UTC)
It's indeed an issue with '?' but probably not in your browser. Your link https://en.wikipedia.org/wiki/Quo_vadis%3F?action=delete doesn't work for me either. How do you reach that link and what is your skin? I'm in Vector and click the "More" tab and then "Delete". That gives me https://en.wikipedia.org/w/index.php?title=Quo_vadis%3F&action=delete which works. Your url structure works on page names not containing '?', for example https://en.wikipedia.org/wiki/Quo?action=delete. It may be a MediaWiki bug that a query part cannot be added to a url of form https://en.wikipedia.org/wiki/...%3F. It also fails for https://en.wikipedia.org/wiki/Quo_vadis%3F?action=history as non-admins can test. The "View history" tab gives me https://en.wikipedia.org/w/index.php?title=Quo_vadis%3F&action=history which does work, but I think https://en.wikipedia.org/wiki/Quo_vadis%3F?action=history should also have worked, like https://en.wikipedia.org/wiki/Quo?action=history works. MediaWiki apparently chokes on the %3F? combination where %3F is an encoded '?' ending the pagename, and '?' is the start of a query part. Or maybe there is some general url rule I'm unaware of which disallows '%3F?' in url's. PrimeHunter (talk) 01:49, 27 February 2015 (UTC)
The %3F? combination works on this wiki with an old MediaWiki version: http://mersennewiki.org/index.php/Why_participate_in_GIMPS%3F?action=history. http://mersennewiki.org/index.php/Special:Version says MediaWiki: 1.5.8. PrimeHunter (talk) 02:24, 27 February 2015 (UTC)
The problem is not restricted to '?' at the end of the pagename. https://en.wikipedia.org/wiki/%3F_Nycticebus_linglom?action=history should have produced the page history of ? Nycticebus linglom. PrimeHunter (talk) 02:56, 27 February 2015 (UTC)
User:PrimeHunter. To get the non working link I just followed the original Quo vadis? link you provided. I then clicked on the delete and it reloaded the page with the https://en.wikipedia.org/wiki/Quo_vadis%3F?action=delete non working link. I'm using Vector as well. I notice you said that you clicked on the "More" tab and then "Delete". On the "More" I see the only option is "Purge". The option to delete is on the "Page" tab and says "Delete page". Also the https://en.wikipedia.org/w/index.php?title=Quo_vadis%3F&action=delete works for me as well. I looked through my deletion log and found this, I did see at least one more with the ? at the end, and this with the ? in the middle. So at one time I was able to delete pages with the ? at the end. The other links on your 01:49 post all work the same for me as they did for you. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 03:19, 27 February 2015 (UTC)
The "Page" tab is not part of MediaWiki. It is made by "Add Page and User dropdown menus to the toolbar" at Special:Preferences#mw-prefsection-gadgets. The gadget is disabled by default. I have tried to enable it and get the same non-working url as you on the Page tab. The gadget uses MediaWiki:Gadget-dropdown-menus.js which uses MediaWiki:Gadget-dropdown-menus-vector.js for Vector users, but the bug is in MediaWiki and should be fixed there. Question marks in pagenames are rare and I'm not sure it's worth coding the gadget to work around the bug by using url's with /w/index.php?title= instead of /wiki/. PrimeHunter (talk) 03:33, 27 February 2015 (UTC)
OK that was it. I removed the gadget and was able to delete https://en.wikipedia.org/w/index.php?title=User:CambridgeBayWeather/Test3%3F&action=delete. I found that with the gadget enabled it is possible to move the page and not leave a redirect. That deleted the ? page and then of course I was able to delete the page without the ?, https://en.wikipedia.org/w/index.php?title=User:CambridgeBayWeather/Test%3F&action=edit&redlink=1. Given that there are several things I like about the gadget and the very few times that the ? comes up I'll leave it enabled and use the move workaround. Thanks for all your help. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 03:48, 27 February 2015 (UTC)
Moves are made with a special page and not a query string, so the bug does not interfere. Protect and Purge on the Page tab are affected. It probably also affects several other scripts and tools. I'm not proficient with Phabricator. Can somebody see if it's there, and add it if not? PrimeHunter (talk) 04:18, 27 February 2015 (UTC)
I'm not seeing a problem with the Protect from the page tab. It worked fine yesterday. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 04:58, 27 February 2015 (UTC)
On a pagename with a question mark? The bug is only known to affect that. PrimeHunter (talk) 05:11, 27 February 2015 (UTC)

────────────────────────────────────────────────────────────────────────────────────────────────────I can fix this for MediaWiki:Gadget-dropdown-menus-vector.js, but I do not maintain the non-vector version if that is what you are using. Because of the same bug all internal links involving a page name with a ? will not work, so it's worth fixing for the interim I think. MusikAnimal talk 06:23, 27 February 2015 (UTC)

Should be fixed now. Thanks for the report! MusikAnimal talk 07:05, 27 February 2015 (UTC)
This should be fixed in core though... -- [[User:Edokter]] {{talk}} 12:38, 27 February 2015 (UTC)
Thanks User:MusikAnimal that works now. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 16:18, 27 February 2015 (UTC)

Save button placement[edit]

I frequently find that after editing, in trying to hit the "save" button, I will inadvertently hit the adjoining link to CC-BY-SA-3.0 License. Can we somehow provide a little safe space around what is arguably the most important widget on Wikipedia? LeadSongDog come howl! 13:53, 26 February 2015 (UTC)

@LeadSongDog: What I did was to add
div#editpage-copywarn { display: none; }
to Special:MyPage/common.css. This hides the block of text from 'By clicking the "Save page" button' down to 'a hyperlink or URL is sufficient for CC BY-SA 3.0 attribution.' --Redrose64 (talk) 14:24, 26 February 2015 (UTC)
Most users probably have the links in MediaWiki:Wikimedia-copyrightwarning on the first line to the right of the buttons in the edit window. I don't think we should add spaces in the interface for everybody just to avoid that some users get a link in the vicinity of a button. PrimeHunter (talk) 15:18, 26 February 2015 (UTC)
That's why I suggested personal customisation. But MediaWiki:Wikimedia-copyrightwarning isn't to the right of the buttons, it's directly above, in both Vector and MonoBook. If LeadSongDog doesn't want to hide that block of text, they can increase the gap below it with a rule like
div#editpage-copywarn { margin-bottom: 2em; }
or perhaps
input#wpSave { margin-top: 2em; }
- again, put the chosen rule in Special:MyPage/common.css. --Redrose64 (talk) 15:30, 26 February 2015 (UTC)
Thank you for the suggestions on simplifying my life, but of course personal customization does nothing for the vast majority of users who are not logged in. Can we not simply use the same space presently above the copyright warning, but put it below it instead? The problem is particularly egregious when working in Safari on the iPad with fat fingers.... It probably runs afoul of some wp:accessibility principles too. LeadSongDog come howl! 16:11, 26 February 2015 (UTC)
The problem would only affect users working on 4:3 resolutions as far as I can tell, on 16:9 and 16:10 resolutions there is no link close to the save page button. Sam Walton (talk) 16:17, 26 February 2015 (UTC)
Perhaps, but 1024x768 is supposed to be our most basic viewport size. Looking at the WCAG, one finds that we've missed G178: Providing controls on the Web page that allow users to incrementally change the size of all text on the page up to 200 percent. Though that wouldn't be the ideal solution to this particular issue, it really should be addressed. More to the point, though, is G155: Providing a checkbox in addition to a submit button. We fall afoul of that (on every viewport size) simply by making the irrevocable agreement on "Save page" not require a dismiss checkbox. LeadSongDog come howl! 17:03, 26 February 2015 (UTC)
LeadSongDog, can you clarify your request? Do you want an extra blank line underneath the copyright notice, or do you want the copyright notice underneath the Save button?
Also, changing anything about the copyright statement requires approval from Legal, so we'd have to check with them before anything could be done. Whatamidoing (WMF) (talk) 00:01, 27 February 2015 (UTC)
Not suggesting any change to the notice, just looking for some space above the "Save page" button to separate it from the CC BY-SA 3.0 License link. If vertical space is an issue, reducing the gap below "Watch this page" or below the "Save page" would do it. LeadSongDog come howl! 01:20, 27 February 2015 (UTC)
I'm not a lawyer and don't work for WMF but I guess the legal concern is that By clicking the "Save page" button, you agree to... should be close to the "Save page" button so contributors are clearly warned of what they agree to. The text also starts above the button for me. Earlier I meant that the part containing the links is not near the Save button on my screen where the line wraps after the last link on "GFDL". On narrow screens/windows or with large fonts the line may wrap earlier and place a link near the Save button. PrimeHunter (talk) 02:40, 27 February 2015 (UTC)
Well that's the real argument for having a separate checkbox for "I understand" before enabling the save page button, but that's a rather larger change than 2 or 3 en of white space. LeadSongDog come howl! 03:10, 27 February 2015 (UTC)

Colors in vector skin?[edit]

Hello, I need some help please (Windows XP, FF 36, Vector skin). For some reason all colors within my Wikipedia windows are brighter and/or messed up. For example: WP:PR usually had a light green background color for me, but now is very light gray, almost white. The documentation box in template:cite web was light green, but is now very light blue. Project banners on talkpages have been turquiose / light greenish and are now bright yellow. The Wiki browser page as a whole "looks" brighter. Any idea, where to look for such settings or how to reset them to the "old" status? (I tried to reset brightness and contrast on my monitor and graphic settings, but with no success). GermanJoe (talk) 14:36, 26 February 2015 (UTC)

I have Win 8.1, FF36 and Vector. I see no changes. {{documentation}} still has the same light green/blue background.      --  Gadget850 talk 15:01, 26 February 2015 (UTC)
Thanks for checking, Gadget850. I did a complete color recalibration in my monitor setup and that somehow fixed the problem. I still don't know, how the settings got corrupted to begin with, but never mind - it's OK again :). GermanJoe (talk) 15:20, 26 February 2015 (UTC)
Did you upgrade graphics drivers (either manually or through Windows Update)? I had this problem about 6 years ago on XP with new NVidia drivers defaulting to some whacky color settings. --Ahecht (TALK
PAGE
) 20:36, 26 February 2015 (UTC)
Only one minor non-graphical update yesterday (that I know of), but I'll check this more closely. Thanks for the tip. GermanJoe (talk) 04:39, 27 February 2015 (UTC)

Strings[edit]

Can somebody give me a link about this kind of patterns: %s%((.-)%) (example from Template:Title disambig text)? I would like to understand them and not ask questions "how to do that or that" :) --Edgars2007 (talk/contribs) 14:42, 26 February 2015 (UTC)

The template says {{#invoke:String|match|{{{1}}}|%s%((.-)%)||-1|ignore_errors=true}}. That means it uses the match function of Module:String. The documentation at Module:String#match has three links under "For information on constructing Lua patterns". PrimeHunter (talk) 15:06, 26 February 2015 (UTC)
So obvious. Thanks! --Edgars2007 (talk/contribs) 15:33, 26 February 2015 (UTC)
Sorry, I tweaked pattern in the template, and then protected it, as I noticed it was used in Template:Disambiguation which has more than 100,000 transclusions. Hopefully that was something along the lines of what you were looking to do... — Mr. Stradivarius ♪ talk ♪ 03:14, 27 February 2015 (UTC)

@Mr. Stradivarius: I'm assuming you didn't bother testing that one.

  • Original on "Foo (bar)": {{#invoke:String|match|Foo (bar)|%s%((.-)%)||-1|ignore_errors=true}} -> bar
  • Your version on "Foo (bar)": {{#invoke:String|match|Foo (bar)|^.*%s%((.-)%)||-1|ignore_errors=true}} ->

You broke it. Reverted. And for the record, the "-1" was already there to get the last match. Dragons flight (talk) 04:04, 27 February 2015 (UTC)

@Dragons flight: Oops. Sorry, yes, I should have tested that properly. Thanks for catching it. — Mr. Stradivarius ♪ talk ♪ 04:18, 27 February 2015 (UTC)
No biggie. Though, I did get a kick out of the fact you broke that template and then immediately protected the broken version. Dragons flight (talk) 04:28, 27 February 2015 (UTC)
Does that qualify for WP:STOCKS? I can see the caption now - "Mr. Stradivarius for the when-it's-broken-it-stays-broken award"... — Mr. Stradivarius ♪ talk ♪ 04:45, 27 February 2015 (UTC)
In case you are curious, the problem appears to be that the use of a match selection index (in this case, "-1") is incompatible with the use of the start anchor tag "^".
  • {{#invoke:String|match|Foo (bar)|^.*%s%((.-)%)|||ignore_errors=true}} -> bar
That's a pretty subtle thing to miss, but does show the value of double checking the result. Dragons flight (talk) 04:47, 27 February 2015 (UTC)

Is there a such parser function/magic word?[edit]

I'm not sure exactly where to ask this, so I thought this would be the best place: Is there a magic word/parser function that automatically detects the target of a redirect (and returns that value) if the page which the magic word/parser function is placed is a redirect? I'm just wondering since there's an edit I would like to perform on {{Db-redirnone}}, but I can only do it properly if that magic word/parser function exists. Steel1943 (talk) 20:56, 26 February 2015 (UTC)

I don't think so, unless H:MW is out of date. --Redrose64 (talk) 21:05, 26 February 2015 (UTC)
(edit conflict) No magic word AFAIK, though there is a Lua script that does that. SiBr4 (talk) 21:08, 26 February 2015 (UTC)
@SiBr4: Thanks for letting me know about that Lua module. However, I tested it out, and I cannot use it for my intended purpose since I cannot use the parser function {{{FULLPAGENAME}}} for the "redirect-page-name". I'm going to ask this on the module's talk page. Thanks again for finding this. Steel1943 (talk) 21:55, 26 February 2015 (UTC)
Never mind SiBr4 ... it work as I need it to (after a 2nd test). Thanks again! Steel1943 (talk) 21:59, 26 February 2015 (UTC)
Neither using the page the module is on as parameter nor using magic words fails by itself. However, placing any text above the redirect breaks the redirect, so the module returns the current page name. Therefore you'll need something else for the deletion template to work. SiBr4 (talk) 08:42, 27 February 2015 (UTC)

Moving over a salted page (redux)[edit]

I'd like to revive this thread.

Summary: When an admin moves a page to a salted target, there is no warning that the page is salted. (I tested it myself and saw no warning.) I'm not sure if the target remains salted.

Case in point: [3], [4]

A warning box would have made User:Necrothesp aware of an old AfD. He then would likely have made others aware of that at the new AfD.

Can this be fixed? Bug report? Thoughts? Anna Frodesiak (talk) 01:09, 27 February 2015 (UTC)

To your q "I'm not sure if the target remains salted", no - because salt can only be applied to the title of a non-existent page; once a page is moved to that title, it's no longer the title of a non-existent page. --Redrose64 (talk) 11:08, 27 February 2015 (UTC)
Hi, Redrose64. I see. And if it is deleted again? Is the redlink then salted again for non-admins? Anna Frodesiak (talk) 21:51, 27 February 2015 (UTC)
No, it's not; and I've just confirmed it at User:Cryptic/test1 (edit | talk | history | links | watch | logs). —Cryptic 22:02, 27 February 2015 (UTC)
Thank you, Cryptic. So, how do we get this fixed. This is the second post about this, and it seems to be getting just as ignored as the first one. Anna Frodesiak (talk) 22:06, 27 February 2015 (UTC)
Posted at phab:T91129
Anna Frodesiak (talk) 22:34, 27 February 2015 (UTC)

Bullet points[edit]

Wp bullet issue.png

When you look at the bullet points in for instance this AFD Wikipedia:Articles for deletion/New Downs - Are they extremely tiny ?,
As you can see from the picture they're absolutely tiny for me so I'm not sure if something here's changed or it's something to do with my laptop/chrome settings?,
Thanks, –Davey2010Talk 03:33, 27 February 2015 (UTC)

They appear at the normal size for me. — Mr. Stradivarius ♪ talk ♪ 03:35, 27 February 2015 (UTC)
There is some discussion of this at the top of #Tables just changed (using monobook). – Philosopher Let us reason together. 03:40, 27 February 2015 (UTC)
Thanks Mr. Stradivarius,
Philosopher - It actually appears to be a Chrome issue I think ?, I tried IE and they're normal but when I try Chrome logged out it still seems tiny, The font settings are all normal, I'll wipe everything and see if that does the trick, –Davey2010Talk 03:46, 27 February 2015 (UTC)
Forgot to say I'm using Vector, Anyway nope all still the same, Well it looks like I partially have the Monobook issue...Just in Vector Face-grin.svg, Ah well thanks anyway ) –Davey2010Talk 04:44, 27 February 2015 (UTC)
It's a Webkit issue; Chrome fixed this last year (in Blink), so that should not be a problem (unless you have a really old Chrome version). -- [[User:Edokter]] {{talk}} 09:43, 27 February 2015 (UTC)
Hi Edokter - Nope I'm using the latest version as far as I know (34.0.1847.116), Thanks, –Davey2010Talk 17:35, 27 February 2015 (UTC)
Davey2010, that is old! The latest version is 40. -- [[User:Edokter]] {{talk}} 17:47, 27 February 2015 (UTC)
Yes, as I noted at #New CSS in Firefox above, Chrome 40.0.2214.115m is current version. --Redrose64 (talk) 17:51, 27 February 2015 (UTC)
Edokter Eh seriously?, I've been updating chrome the moment the 3 bars go green so how on earth could I have gone from using updated versions to old ?.....Unless without me knowing Chrome hasen't been updating at all for quite some time?, I really don't no!, Anyway thanks for solving the mystery ) –Davey2010Talk 18:13, 27 February 2015 (UTC)
By the looks of it my browser hasen't updated itself since April 2014 ..... Great!, Anyway thanks all for your help :) –Davey2010Talk 18:24, 27 February 2015 (UTC)
Safari 8 on OS X Yosemite (Wikipedia home page)

I've got the same problem with Safari 8.0.3 (latest version) on OS X 10.10.2 (see screenshot on the right). The bullet points have about the same size and position as full stops. SelfishSeahorse (talk) 13:40, 1 March 2015 (UTC)

Linking to IPV6 user accounts[edit]

[[User|2602:304:b1ae:8970:2186:8b44:f7ad:ef34]] results in this 2602:304:b1ae:8970:2186:8b44:f7ad:ef34. The text is the IP, but the link goes to User. Oiyarbepsy (talk) 05:43, 27 February 2015 (UTC)

Yes, that's how piped links work. Did you mean {{user|2602:304:b1ae:8970:2186:8b44:f7ad:ef34}}, which produces 2602:304:b1ae:8970:2186:8b44:f7ad:ef34 (talk · contribs)? Regards, Orange Suede Sofa (talk) 05:48, 27 February 2015 (UTC)
Or perhaps you meant [[User:2602:304:b1ae:8970:2186:8b44:f7ad:ef34]]User:2602:304:b1ae:8970:2186:8b44:f7ad:ef34 - colon, not pipe. --Redrose64 (talk) 11:10, 27 February 2015 (UTC)
Doh, now I feel like an idiot. Oiyarbepsy (talk) 19:43, 27 February 2015 (UTC)

Template:Main - appearance (or lack thereof) on mobile browsers[edit]

It seems that {{main}} is hidden on mobiles, due to site CSS. I suggested personal CSS to unhide it, but apparently that doesn't work. Please discuss at Template talk:Main#Appearance (or lack thereof) on mobile browsers. --Redrose64 (talk) 16:11, 27 February 2015 (UTC)

More generally, all hatnotes (specifically, all elements with the hatnote CSS class) are hidden unless the display width is large enough. This means important navigation information is missing for mobile users who don't think to (or can't) rotate the display to a landscape configuration. Hairy Dude (talk) 16:26, 27 February 2015 (UTC)

username change, and now I can't log in!!![edit]

I asked for a username change from EmeraldRS to Emerald-wiki. It says the change has been carried out but I can't log in with my old or new username. I get a message with the new username saying my account is being renamed or merged. Is there some step that got missed? Help! 80.176.153.231 (talk) 17:35, 27 February 2015 (UTC)

The rename was an hour ago [5] and your account has nearly 1000 edits in total. Try again later. PrimeHunter (talk) 17:52, 27 February 2015 (UTC)

I have waited, it is now about 15 hours since the rename, but still I can't login. Are you sure it is just a matter of waiting and not that something else needs fixing? 80.176.153.231 (talk) 08:49, 28 February 2015 (UTC)

I think it should have worked by now. What exactly happens if you try to log in as Emerald-wiki? If you get a message then quote it precisely. Can you log in at meta: or wikisource:? PrimeHunter (talk) 13:01, 28 February 2015 (UTC)

When I try to log-in the message says "Login error/ Your account is currently being renamed or merged./ View the status", clicking 'View the status' shows the Global rename progress with everything 'Done' except de.wikipedia.org which is 'Queued' and has been for approaching 24 hours. I get exactly the same thing at Meta, Wikisource and Commons. I have never used de.wikipedia.org and, not speaking German, probably never will. I hope you can help. 80.176.153.231 (talk) 14:23, 28 February 2015 (UTC)

Apparently the global renaming process is still waiting for de.wikipedia to be completed. That's one of the downsides of global accounts: if one local instance is delayed or interrupted I suppose you can't use your account at all. It looks like you'll have to wait for the German queue to finish. De728631 (talk) 14:36, 28 February 2015 (UTC)
Okay, thanks, I'll wait. I changed my username once before, that time it happened so fast it surprised me - within 5 mins of submitting my request it had happened, I hadn't even got as far as logging out. This time it is really, really slow! 80.176.153.231 (talk) 14:40, 28 February 2015 (UTC)
  • @Keegan: is such a long delay expected? –xenotalk 14:57, 28 February 2015 (UTC)
  • Note that there is already a user on deWP called Emerald-wiki, which was created on 10 January 2015 but hasn't made any edits. Is that his newly renamed global account, or is it an unrelated account holding up the rename? A few random other wikis I've checked don't have a user by that name, though esWP does. SiBr4 (talk) 15:15, 28 February 2015 (UTC)

It seems the hold-up has been identified. Can I take over the other Emerald-wiki on DE or choose another username - bearing in mind that I can't login anymore?80.176.153.231 (talk) 15:22, 28 February 2015 (UTC) I should say that the Emerald-wiki on DE in January is nothing to do with me. I was EmeraldRS or my previous username, which unwisely was my real name 80.176.153.231 (talk) 15:26, 28 February 2015 (UTC)

I don't know the reason for the de hold-up but I definitely think the de account is yours, created when you were EmeraldRS and part of the rename. Special:CentralAuth/Emerald-wiki shows it belongs to the global account, and was created at a time where you were also created at other wikis. If you are logged in at another wiki and visit any page at a Wikimedia wiki which is part of the global login system then your account is automatically created there. de:Special:Logs/Emerald-wiki confirms it was an automatic creation. https://tools.wmflabs.org/quentinv57-tools/tools/sulinfo.php?username=Emerald-wiki also says there are no unattached acccounts. PrimeHunter (talk) 15:41, 28 February 2015 (UTC)

I'll just keep waiting then!!! I wish I did know German to ask a question on one of their help talks!80.176.153.231 (talk) 15:48, 28 February 2015 (UTC)

Hi there (thanks for the ping, xeno). We'll get that fixed for you. Sorry for the hold up, that's not normal. Keegan (WMF) (talk) 03:36, 1 March 2015 (UTC)
Yes check.svg Done - Legoktm unlocked the account and hopefully fixed the issue. You should be able to log in to your account. If you have any future problems getting locked out, please feel free to contact me directly. You can find my contact information on my userpage. Again, sorry for the trouble. Keegan (WMF) (talk) 03:58, 1 March 2015 (UTC)

Thanks everyone. Got up this morning and the first thing I tried was logging in to Wiki - straight in, no problem - great!! Richard Emerald (talk) 10:41, 1 March 2015 (UTC)

If subpages in mainspace are "disabled", why aren't they disabled?[edit]

Wikipedia:Subpages tells us that "in main namespace (article namespace) ... the subpage feature has been disabled in the English Wikipedia". So how could a page such as Barbara Borts/Temp get created? It's the second one of these I've come across in the last few days (editors apparently find it counter-intuitive to work on a draft in talkspace). Might it be an idea to actually disable mainspace subpages if people aren't supposed to make them? Or limit the capability to admins or something? Justlettersandnumbers (talk) 17:54, 27 February 2015 (UTC)

"disabled" does not mean that pagenames with '/' are impossible. It just means they are treated like any other mainspace pagename, and the features associated with subpages do not apply. For example, Barbara Borts/Temp has no automatically added link to Barbara Borts at the top. Compare to Talk:Barbara Borts/Temp which does have a link to Talk:Barbara Borts. The talk page is a subpage but the mainspace page is not. This is admittedly confusing. PrimeHunter (talk) 17:59, 27 February 2015 (UTC)
I don't know how common mistaken "subpages" in mainspace are but it would be possible in MediaWiki:Titleblacklist to limit creation of '/' in mainspace to admins. There are many valid reasons for '/' in article titles, like OS/2 and 9/11 Commission, so it would inconvenience many non-admins. PrimeHunter (talk) 18:08, 27 February 2015 (UTC)
Thanks for the reply. Yes, I see (and knew) that names with a slash are possible; I too am unsure how often the mistake is made, I just happen to have seen it twice in a short space (both times experienced editors rewriting copyvio pages). It may be better to live with that level of background noise. But what is easily fixed is the documentation: if subpages are not disabled, we shouldn't really be saying that they are. If what we actually mean is "mainspace pages containing a slash character do not become subpages; instead, the whole name is treated as the pagename", then that might be a better way of explaining it. Justlettersandnumbers (talk) 20:59, 27 February 2015 (UTC)
Go forth and update that documentation. --Izno (talk) 22:11, 27 February 2015 (UTC)
But they are disabled. Subpages are a feature which can be enabled or disabled in each namespace with mw:Manual:$wgNamespacesWithSubpages. It's called enable and disable in the MediaWiki documentation and those are the common and perfectly fine names for features which can, well, be enabled or disabled. It's used about hundreds of MediaWiki and Wikipedia features. Wikipedia:Subpages is not a technical help page but an editing guideline in project space. It starts: Except in main namespace (article namespace), where the subpage feature has been disabled in the English Wikipedia, subpages are pages separated with a "/" (a slash) from their 'parent' page.
Click "subpage feature" to learn more about the feature. That's how we work. A page may have some info about something and link to a page with more info. Wikipedia:Subpages#Slashes in article titles also explains that articles can have slashes in the title. PrimeHunter (talk) 23:41, 27 February 2015 (UTC)

Infobox subheader shift?[edit]

In Jacques Brel, my eye gets the impression that the infobox header (the colored bar with text 'Jacques Brel') is asymmetrical: lefthand whitespace looks like 3px, righthand ws 2px. I'm using Firefox 35 (not 36 yet) now. Possibly related: #Tables_just_changed_.28using_monobook.29. Is it just my eye, or is this real? -DePiep (talk) 20:49, 27 February 2015 (UTC)

Not sure, was the dress white/gold for you today or black/blue ? —TheDJ (talkcontribs) 21:33, 27 February 2015 (UTC)
Team white and gold! --MZMcBride (talk) 22:15, 28 February 2015 (UTC)

In Jacques Brel, my eye gets the impression that the infobox header (the colored bar with text 'Jacques Brel') is asymmetrical: lefthand whitespace looks like 3px, righthand ws 2px. I'm using Firefox 35 (not 36 yet) now. Possibly related: #Tables_just_changed_.28using_monobook.29. Is it just my eye, or is this real? -DePiep (talk) 22:13, 27 February 2015 (UTC)

I can't login to Wikimedia[edit]

I tried nominating a Wikimedia user page for deletion due to WP:G11, however, I'm failing to login because everytime I refresh it never logs me in, instead it shows me the Central login message. How can I fix this? --ToonLucas22 (talk) 22:02, 27 February 2015 (UTC)

The Wikimedia Foundation runs a lot of wikis and some of them don't allow account creation or login without permission. Please post a link to the page you want deleted and the page where you try to log in. PrimeHunter (talk) 23:14, 27 February 2015 (UTC)
This page I want deleted. --ToonLucas22 (talk) 23:19, 27 February 2015 (UTC)
That's... weird. Special:CentralAuth/ToonLucas22 looks ok. Have you tried going directly to meta to log in (m:Special:UserLogin)?
(I've marked that page for deletion myself, and created it blank here. I'm sure we'll be seeing lots, lots more block and deleted-userpage evasion of this sort in the future.) —Cryptic 23:37, 27 February 2015 (UTC)
You didn't post the requested link to the page where you try to log in. The wiki of the user page is called meta or Meta-Wiki and not Wikimedia so I wonder whether it's another wiki you try to log in at. Please post the url of the login page. PrimeHunter (talk) 23:51, 27 February 2015 (UTC)
I try to log in to m:Special:UserLogin. --ToonLucas22 (talk) 00:11, 28 February 2015 (UTC)
OK, that's meta where your account was created in December and should work. Does Help:Logging in help? Note that browsers can have site-dependent settings. Can you log in at commons: which is also on wikimedia.org? PrimeHunter (talk) 00:26, 28 February 2015 (UTC)
Nope, I can't log in there either. I refresh, still gives me Central Login message. --ToonLucas22 (talk) 00:36, 28 February 2015 (UTC)
Try others at Wikipedia:Wikimedia sister projects#Sister projects and see if it's limited to wikimedia.org. Try another browser if you can. Look for site-dependent cookie and security settings in your browser. Or just drop the whole thing if you only need to edit Wikipedia. PrimeHunter (talk) 02:43, 28 February 2015 (UTC)

Print/Export as pdf does not render tables[edit]

I've seen some chatter in the archives about the failure to render infoboxes, but it seems to me that the problem is much more widespread than that. I have looked around and every WP article with tables that I have checked have had the tables excluded in the .pdf print/export version. That is to say, after looking around, I have yet to find ANY tables that are rendered in the .pdf generated by print/export. Yes, there are many articles without tables, nevertheless, many articles include critical information in the articles and the lack of tabular information renders (pun intended) the .pdf nearly worthless. This is a widespread problem that IMHO should be prioritized relatively high. Examples are easy to find:

Go to your own favorite corner of WP and if you find tables, you will find this bug. YBG (talk) 05:08, 28 February 2015 (UTC)

Known issue. See T87499 which was merged into T73808. -- George Orwell III (talk) 06:05, 28 February 2015 (UTC)
Great! Much appreciated. YBG (talk) 06:16, 28 February 2015 (UTC)
Help:Download as PDF --  Gadget850 talk 14:49, 28 February 2015 (UTC)

AfD stats tool not recognising votes for me[edit]

Is there any reason why this is only counting my noms, and not my vote at AfD? --Fauzan✆ talk✉ mail 16:04, 28 February 2015 (UTC)

My contributions in Wikipedia:Articles for Deletion/* --Fauzan✆ talk✉ mail 16:13, 28 February 2015 (UTC)
Which (diff) vote are you referring to? Nakon 04:02, 1 March 2015 (UTC)
Nakon, I mean the voting matrix shows only my nominations, and not the 50 other votes at AfD. The tool is not recognizing my votes at AfD, which it should. --Fauzan✆ talk✉ mail 08:23, 1 March 2015 (UTC)

NCMEC template[edit]

I recently created a template for a link to the National Center for Missing and Exploited Children, which would be used on missing persons pages, but I can't get the URL to work. I created it after I saw Template:Find a Grave. Here is my template page: Template:NCMEC. Here is a page that it is used on: Michaela Garecht and here is a URL to a poster: [6]. I'm hoping someone could help with this issue, as the formatting works perfectly with my template for unidentified persons on the website (see Template:NCMEC UID. Thanks, --GouramiWatcher(?) 17:04, 28 February 2015 (UTC)

I don't know the website. Is there a reason the template says NCMU instead of NCMC the second time? Your example works if it's changed to NCMC. PrimeHunter (talk) 17:26, 28 February 2015 (UTC)
"NCMU" is used with the unidentified people. The missing children's URLs have the "NCMC." --GouramiWatcher(?) 20:12, 28 February 2015 (UTC)
But why does Template:NCMEC only choose a url with NCMC if there is an author, date or accessdate parameter, and another url with NCMU if there isn't? None of those parameters sound like they have something to do with unidentified people versus missing children. It seems a very odd system, and I think it's an extremely poor idea to make an external link template which changes the url if an accessdate is added. That means a user may test the link, find it works, add an accessdate, and that action may break the link without the user rechecking it. Additionally, the NCMC case is coded wrong and needs fixing but I would like to first understand the idea behind the whole NCMC/NCMU choice. PrimeHunter (talk) 21:51, 28 February 2015 (UTC)
Changed NCMU to NCMC as per PrimeHunter and it seems to pull the poster just fine then, but the concerns above remain & are on point.

Oh I think I get the premise now, the first incarnation is for use as a citation for inline refs and the second is for standalone usage (still doesn't quite add up however). -- George Orwell III (talk) 22:03, 28 February 2015 (UTC)

Maybe that's the intention. All current uses of both templates are in external links sections and none of them use any of the parameters which invoke the first incarnation. The OP said "the formatting works perfectly" about Template:NCMEC UID, but that template also changed the url (although only by adding '=') if any of author, date or accessdate were added. In all cases I tested, this broke the link. I guess it was simply never tested and '=' should just be removed from the url in both templates, in addition to your change of NCMU, and an addition of a missing pipe. I made these changes [7][8] and both templates now give working links in both incarnations:
PrimeHunter (talk) 00:52, 1 March 2015 (UTC)
Thanks for your assistance! Sorry if I confused anyone! --GouramiWatcher(?) 03:30, 1 March 2015 (UTC)

Table cell contents spilling over.[edit]

When viewing the following table in the mobile version of the site, some text from the first column spills over into the adjacent cell of the second column.

Team Constructor Chassis Power unit Tyre No. Drivers
Italy Scuderia Ferrari Ferrari SF15-T[1] Ferrari P 5
7
Germany Sebastian Vettel
Finland Kimi Räikkönen
India Sahara Force India F1 Team Force India-Mercedes VJM08[2] Mercedes PU106B Hybrid P 11
27
Mexico Sergio Pérez
Germany Nico Hülkenberg
United Kingdom Lotus F1 Team Lotus-Mercedes E23 Hybrid[3] Mercedes PU106B Hybrid P 8
13
France Romain Grosjean
Venezuela Pastor Maldonado
United Kingdom Manor Marussia F1 Team[4] Marussia-Ferrari TBA Ferrari 059/3[5][6] P TBA
TBA
United Kingdom Will Stevens
Flag of None.svg TBA
United Kingdom McLaren Honda McLaren-Honda MP4-30[7] Honda RA615H Hybrid P 14
22
Spain Fernando Alonso
United Kingdom Jenson Button
GermanyMercedes AMG Petronas F1 Team Mercedes F1 W06 Hybrid[8] Mercedes PU106B Hybrid P 6
44
Germany Nico Rosberg
United Kingdom Lewis Hamilton
Austria Infiniti Red Bull Racing Red Bull-Renault RB11[9] Renault Energy F1-2015 P 3
26
Australia Daniel Ricciardo
Russia Daniil Kvyat
Switzerland   Sauber F1 Team Sauber-Ferrari C34[10] Ferrari P 9
12
Sweden Marcus Ericsson
Brazil Felipe Nasr
Italy Scuderia Toro Rosso Toro Rosso-Renault STR10[11] Renault Energy F1-2015 P 33
55
Netherlands Max Verstappen
Spain Carlos Sainz Jr.
United Kingdom Williams Martini Racing Williams-Mercedes FW37[12] Mercedes PU106B Hybrid P 19
77
Brazil Felipe Massa
Finland Valtteri Bottas
Source:[13][4][14][15][16]


Anyone got an idea what's causing this and/or how to solve this. Tvx1 22:24, 28 February 2015 (UTC)

I don't have a mobile device and it looks right for me in both desktop and https://en.m.wikipedia.org/w/index.php?title=Wikipedia:Village_pump_(technical)&mobileaction=toggle_view_mobile. The table has a coding error in {{nowrap|{{nowrap|Mercedes PU106B Hybrid}} which should only have one {{nowrap}}. Does it help to remove that:
Team Constructor Chassis Power unit Tyre No. Drivers
Italy Scuderia Ferrari Ferrari SF15-T[17] Ferrari P 5
7
Germany Sebastian Vettel
Finland Kimi Räikkönen
India Sahara Force India F1 Team Force India-Mercedes VJM08[18] Mercedes PU106B Hybrid P 11
27
Mexico Sergio Pérez
Germany Nico Hülkenberg
United Kingdom Lotus F1 Team Lotus-Mercedes E23 Hybrid[19] Mercedes PU106B Hybrid P 8
13
France Romain Grosjean
Venezuela Pastor Maldonado
United Kingdom Manor Marussia F1 Team[4] Marussia-Ferrari TBA Ferrari 059/3[5][6] P TBA
TBA
United Kingdom Will Stevens
Flag of None.svg TBA
United Kingdom McLaren Honda McLaren-Honda MP4-30[20] Honda RA615H Hybrid P 14
22
Spain Fernando Alonso
United Kingdom Jenson Button
Germany Mercedes AMG Petronas F1 Team Mercedes F1 W06 Hybrid[21] Mercedes PU106B Hybrid P 6
44
Germany Nico Rosberg
United Kingdom Lewis Hamilton
Austria Infiniti Red Bull Racing Red Bull-Renault RB11[22] Renault Energy F1-2015 P 3
26
Australia Daniel Ricciardo
Russia Daniil Kvyat
Switzerland   Sauber F1 Team Sauber-Ferrari C34[23] Ferrari P 9
12
Sweden Marcus Ericsson
Brazil Felipe Nasr
Italy Scuderia Toro Rosso Toro Rosso-Renault STR10[24] Renault Energy F1-2015 P 33
55
Netherlands Max Verstappen
Spain Carlos Sainz Jr.
United Kingdom Williams Martini Racing Williams-Mercedes FW37[25] Mercedes PU106B Hybrid P 19
77
Brazil Felipe Massa
Finland Valtteri Bottas
Source:[13][4][14][15][26]
Does it help to remove all nowrap (may be controversial in an article), or for simplicity replace them by {{identity}} as here:
Team Constructor Chassis Power unit Tyre No. Drivers
Italy Scuderia Ferrari Ferrari SF15-T[27] Ferrari P 5
7
Germany Sebastian Vettel
Finland Kimi Räikkönen
India Sahara Force India F1 Team Force India-Mercedes VJM08[28] Mercedes PU106B Hybrid P 11
27
Mexico Sergio Pérez
Germany Nico Hülkenberg
United Kingdom Lotus F1 Team Lotus-Mercedes E23 Hybrid[29] Mercedes PU106B Hybrid P 8
13
France Romain Grosjean
Venezuela Pastor Maldonado
United Kingdom Manor Marussia F1 Team[4] Marussia-Ferrari TBA Ferrari 059/3[5][6] P TBA
TBA
United Kingdom Will Stevens
Flag of None.svg TBA
United Kingdom McLaren Honda McLaren-Honda MP4-30[30] Honda RA615H Hybrid P 14
22
Spain Fernando Alonso
United Kingdom Jenson Button
Germany Mercedes AMG Petronas F1 Team Mercedes F1 W06 Hybrid[31] Mercedes PU106B Hybrid P 6
44
Germany Nico Rosberg
United Kingdom Lewis Hamilton
Austria Infiniti Red Bull Racing Red Bull-Renault RB11[32] Renault Energy F1-2015 P 3
26
Australia Daniel Ricciardo
Russia Daniil Kvyat
Switzerland   Sauber F1 Team Sauber-Ferrari C34[33] Ferrari P 9
12
Sweden Marcus Ericsson
Brazil Felipe Nasr
Italy Scuderia Toro Rosso Toro Rosso-Renault STR10[34] Renault Energy F1-2015 P 33
55
Netherlands Max Verstappen
Spain Carlos Sainz Jr.
United Kingdom Williams Martini Racing Williams-Mercedes FW37[35] Mercedes PU106B Hybrid P 19
77
Brazil Felipe Massa
Finland Valtteri Bottas
Source:[13][4][14][15][36]
PrimeHunter (talk) 04:56, 1 March 2015 (UTC)
Thanks for mentioning that coding error. That didn't cause the problem however. Actually it did not do any harm at all. If you click on the link to the mobile view and then reduce the width of your browser screen to the minimum you will see the text from the first column spilling over. Using the identity template doesn't solve it. Tvx1 05:38, 1 March 2015 (UTC)
I already tried the minimum width on mobile and it works for me. I get a horizontal scroll bar and no overlap. I will stop guessing. It's too hard when I don't have the problem. PrimeHunter (talk) 05:51, 1 March 2015 (UTC)

Popups not working[edit]

...starting late last week on several different Mac running various fairly-new and fairly-old firefox versions. I've tried disabling the gadget (hovering on a wikilink gives me a yellow box with just the target name) then re-enabling it gadget (reverts back to nothing appearing while hovering on a wikilink) and cache purge/refresh. Same on commons as well as enwp. On a Mac where firefox has not accessed WP in a long time if at all, popups do work. So it seems like "something at some level" has changed to be incompatible with "something else", but I don't know how to diagnose further what/where. DMacks (talk) 04:08, 1 March 2015 (UTC)

It works for me with Firefox in Windows Vista. Does it work if you log out and click https://en.wikipedia.org/w/index.php?title=Example&withJS=MediaWiki:Gadget-popups.js&withCSS=MediaWiki:Gadget-navpop.css? That also works for me. PrimeHunter (talk) 04:25, 1 March 2015 (UTC)
Popups works for me on OSX 10.9.5, Firefox 35. Keegan (talk) 05:42, 1 March 2015 (UTC)
Updated to FF 35.0.1edit: and 36.0 to be sure, and popups are still fine. Keegan (talk) 05:44, 1 March 2015 (UTC)
Removed AdBlock (it hadn't been a problem before, but may as well rule it out), also tried manually importing it in my common.js, still nothing. DMacks (talk) 10:22, 1 March 2015 (UTC)

Wikidata problem[edit]

I have added a new property to an existing item – P527 (name = 'has part', value = barangay) to (Q316370) Santa Fe, Cebu, then added qualifer P1114 (name = 'quantity', value = 10).

When I try to retrieve, using {{#invoke:Wikidata|getRawQualifierValue|P527|P1114|FETCH_WIKIDATA}} it throws an error:
Lua error in Module:Wikidata at line 168: invalid value (table) at index 1 in table for 'concat'.

Is it me or ...?

NB I have successfully used getting qualifier values before, e.g. {{Population census prose}}

--Unbuttered parsnip (talk) mytime= Sun 16:35, wikitime= 08:35, 1 March 2015 (UTC)

I don't know Lua and know very little Wikidata so I could be wrong but it looks to me like getRawQualifierValue in Module:Wikidata cannot handle qualifiers with data type Quantity such as wikidata:Property:P1114. {{Population census prose}} uses getRawQualifierValue for wikidata:Property:P459 which has data type Item. PrimeHunter (talk) 15:58, 1 March 2015 (UTC)
Well it allowed me to add the qualifier to the property, so it's a surprise it doesn't work (not really, this is wikidata we're talking). I'll take my problem over to the wikidata village pump (T) , if there is one. Unbuttered parsnip (talk) mytime= Mon 07:38, wikitime= 23:38, 1 March 2015 (UTC)
I suspect the problem is in Module:Wikidata and have posted to Module talk:Wikidata. PrimeHunter (talk) 23:42, 1 March 2015 (UTC)
I think that my naming the Wikipedia Lua module "Wikidata" may be causing confusion. Whatever happens on Wikidata won't guarantee that the Lua module (which is intended to import data from Wikidata to Wikipedia in a flexible way) will always work. The problem occurs because the module was written before numeric data types existed on Wikidata. Line 168 takes a table of all of the values for that qualifier and concatenates them together with a comma+space separator assuming that the value is a string. As I understand it, you can't concatenate numbers like that, using the table.concat() method, unfortunately. I think I can solve the problem by ensuring that the wikitext gets a string representation of the number but I'll need to do some testing before I can be sure. --RexxS (talk) 00:03, 2 March 2015 (UTC)
Thanks for looking into it. Oddly, the revision time for my above edit [9] says 23:46, four minutes after the 23:42 signature which was made with a normal four ~~~~. I sometimes see a 1-minute difference which makes sense if it takes about a second to process the edit and it was saved near a minute change with signature time and revision time recorded at different stages. I have never seen more than 1 minute. The save went through quickly. My browser history indicates it was 23:46 with an accurate clock. PrimeHunter (talk) 00:08, 2 March 2015 (UTC)
Well, it's not the problem I thought it was. It seems table.concat() does implicitly cast the numbers into strings, so the concatenation method works as it should. From the error message you give, I now suspect that the first item in the table is actually a table, not a single value. I'll go check that and see if I can fix it. --RexxS (talk) 00:24, 2 March 2015 (UTC)
Update: yes the first item is a table, so can't be concatenated with ", " as the call tries to do. --RexxS (talk) 14:40, 2 March 2015 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── The more I've looked at this, the more I become convinced that you can't write a generic call to fetch just the value of a given qualifier associated with a given property because of the possibility of multiple values. For example, if you look at Richard Burton (d:Q151973), he has 3 spouses (P26) listed, two of whom have qualifiers for start time (P580) and end time (P582). What should a call asking for Richard Burton's spouse's start time return? Should it return nil for Elizabeth Taylor? or 3 July 1983 for Sally Burton? or 1949 for Sybil Christopher? Of course we could have it return 1949 and 3 July 1983, but what value would they be without knowing which spouse they referred to? It would be possible to write a call that fetched e.g. the value of first given qualifier from the first given property, but that becomes vulnerable to someone adding another value for the property. How would you want to deal with the situation if Santa Fe (d:Q316370) had 30 sitios and someone added "has part=sitio" with a qualifier of "quantity=30"? Should getRawQualifierValue|P527|P1114 return 10 or 30? Sorry to be negative about the problem, but I can't see a general solution. --RexxS (talk) 14:40, 2 March 2015 (UTC)

I don't understand any of this. Other properties, for instance 'population', are quite happy to have more than one different qualifier ('date' and 'method'), no problem. Other properties, e.g. 'inception' are happy to have (and return) more than one value. I cannot believe that it is not possible to return a qualifier attached to an instantiation of a property, here 'quantity' attached to 'has part'=barangay. Time to bin it all and start again in SQL, or anything proprietary and proven.-- Unbuttered parsnip (talk) mytime= Mon 23:30, wikitime= 15:30, 2 March 2015 (UTC)
Maybe a discussion is needed about what to do with multiple values (an error message and tracking category if the caller didn't specify which value was wanted?), but can it be coded to at least give the value when there is only one? PrimeHunter (talk) 16:03, 2 March 2015 (UTC)
There is no way of specifying which qualifier you want if more than one - you just get them all. But qualifier should be attached to property value, not to property name as implied here. Unbuttered parsnip (talk) mytime= Tue 00:24, wikitime= 16:24, 2 March 2015 (UTC)
Yes, code can be written to return a given qualifier's value if there is only one qualifier present among all the statements connected with a given property. But how would you want it to behave in Wikipedia if somebody added another qualifier in Wikidata? Should it disappear because it's no longer a unique value? or are you willing to risk the value of the second qualifier now appearing in the Wikipedia article instead of the value of the previously unique qualifier?
Qualifiers are definitely attached to a particular name-value pair. So in the case of Richard Burton, the property named "spouse" with the value "Sybil Christopher" has attached to it a qualifier called "start date" with the value '1949'. But again how does that help you if you want a call that returns just the qualifier value? because you don't know the value "Sybil Christopher" when you make the call. On the other hand, if you want a call that returns a list of the all values of a given property along with a list of the values of a given qualifier when present, then that's possible as it could look like this for Richard Burton / property='spouse' / qualifier='start date' :
  • Elizabeth Taylor; Sally Burton, 3 July 1983; Sybil Christopher, 1949; (note that the two dates for Elizabeth Taylor's two marriages to Burton could be added at any time)
But that's different from the problem outlined by Unbuttered Parsnip. --RexxS (talk) 22:56, 2 March 2015 (UTC)
It already returns more than one value for a qualifier, as in 'inception' or 'demonym'. I don't see the point in providing a massive database which can't be interrogated. If it is possible to add a value to a qualifier of a property, then it has to be possible to read it back too. As regards particular case, I can't make any sense of the notion that a qualifier could have more than one 'quantity' value. In the general case, I think one would want the latest. So for population ordinarily you'd want the latest (current) census, not one from 10 years ago. Exceptionally you might want all of them. Because data-sort is not a parser function, it would be incumbent on WD to return a sorted list. -- Unbuttered parsnip (talk) mytime= Tue 10:52, wikitime= 02:52, 3 March 2015 (UTC)
(1) The database can be interrogated. (2) It is possible to read the data back, but if there are multiple values meeting the search criteria, you have to specify what you want read back. (3) There is no reason whatsoever why a property can't have multiple instances of the same qualifier. I already gave you the example of Richard Burton marrying his second wife twice, so the start date of the marriage would appear as a qualifier twice with two different values: 1964 and 1975. If you call for the start date of Richard Burton's marriage to Elizabeth Taylor, which qualifier value would you want returned? (3) The "latest" doesn't make sense in general for all qualifier values which could be numbers or text as well as dates. (4) Yes, sometimes you want just one qualifier value from amongst several - and there's no means of specifying which one - and sometimes you want all of them, but writing a generic module that copes with any Wikidata item won't do what you want. (5) At present there is no call to the mw.wikibase API that returns sorted tables, but sorting can be done in the Lua module, so parser functions are irrelevant. --RexxS (talk) 03:59, 4 March 2015 (UTC)
Reads like a snow job to me. I don't know how the database can be interrogated, because the only module we've been given doesn't seem to work. In particular, it doesn't allow specification of which occurrence of the property, nor of which occurrence of any qualifier. Your reply implies that the qualifiers for different instantiations of a property are jumbled together, and it's pot-luck which gets returned. Sad if true. I didn't specify every qualifier, I was restricting myself to 'quantity'. If a property has a qualifier of 'quantity' then I'd be very surprised if there could be more than one associated value. That's tantamount to saying that the value of the qualifier contains every known number. I can't imagine anyone would update a raw value by adding another instantiation of the qualifier rather than just changing the value, since there is no way of differentiation. As I said, other properties, e.g. 'inception', allow multiple copies of qualifier (a date) and return them without breaking.
Regarding Burton–Taylor, it seems incorrect to have only one wedding tuple with more than one date set, when they married twice – correct form would be one tuple for each marriage with not more than one date set each. -- Unbuttered parsnip (talk) mytime= Wed 12:49, wikitime= 04:49, 4 March 2015 (UTC)

What's going to happen next with Tables?[edit]

As someone who works with wikitables all the time I am very unhappy with the padding change. Most of the time I am working on wikipedia I work on results tables (Most of the time motorsport results tables). Those result tables need to be tight so you can put 20-sometimes 56 or more races in one table. Example: Dave Marcis. Now that the padding has changed (doubled) these result tables are way bigger than they should be. This makes them harder too reed and untidy. I really don't understand why the padding has doubled. This affect every single table on wikipedia. I am really disappointed and upset that this affects all the hard work I have put in to create all these results tables. What's going to happen next? I mean the padding being doubled is'nt a minor change. Particularly when this affects almost all the tables on wikipedia. Is their a possibility that this change is going to be reverted? Jahn1234567890 (talk) 15:59, 1 March 2015 (UTC)

Have you commented at https://phabricator.wikimedia.org/rMW33cfd0bc4a5f009b86c40b0312b4c7f7b006cbc7? --Ahecht (TALK
PAGE
) 18:08, 1 March 2015 (UTC)

Editnotice permissions by namespace[edit]

We are currently restricting editnotice editing via the MediaWiki:Titleblacklist by blocking creation and edits to: Template:Editnotices\/.* <noedit|errmsg=titleblacklist-custom-editnotice>. What technical options can be used to allow creating/editing of notices in the Wikipedia: / Wikipedia_talk: namespaces by standard editors? — xaosflux Talk 16:08, 1 March 2015 (UTC)

I attempted a more-specific whitelist entry, but either have a syntax error or the software needs exact match on WL-->BL. — xaosflux Talk 16:12, 1 March 2015 (UTC)
Why not just remove the title blacklist entry altogether? It's clearly a hack and it has always been very dubious. --MZMcBride (talk) 16:49, 1 March 2015 (UTC)
That is an option, and I plan on opening up a RfC on changing these--but want to know what the technical options are first. — xaosflux Talk 20:06, 1 March 2015 (UTC)
Haha if you do I'll be sure to add "caused an RfC" to my list of...things accidentally achieved, I guess? ResMar 20:09, 1 March 2015 (UTC)
It is the standard way to do this type of change, this isn't like a normal unprotection discussion as it applies to a feature of most every page - I'm sure the arguments are vandals - but there may be something else. — xaosflux Talk 20:11, 1 March 2015 (UTC)
  • It isn't just vandals, it's also intended to avoid displaying off-putting messages to new users (probably other reasons as well, but this one is likely the most important). Editnotices aren't as visible as page content, and furthermore are often hidden by experienced users, so inappropriate changes to these can be easily missed. Lowering the requirement to edit existing editnotices in project namespace to 'autoconfirmed' may be acceptable, provided all the editnotices for project pages that are potentially edited by new users (like the help desk and so) are template-protected (this should number in the dozens). Creation should still be restricted since it is also intended to avoid an unchecked proliferation of editnotices. I've personally seen on numerous occasions editnotices incredibly BITEY, uncalled for or not assuming good faith. So we'd better be conservative in our approach to these user interface messages. Using the title blacklist is a hack for sure but it's all we've got now barring sysop-only. If this is still desired, we may have to rewrite the editnotice template so that it uses a Editnotices/Page/Namespace/UnprefixedTitle format, where mainspace is "main", so e.g. "Editnotices/Page/Main/Barack Obama" or "Editnotices/Page/Wikipedia talk/Help desk". Although it would cost a few hundreds of moves, this way it's going to be easy to make special cases on namespace using the title blacklist, or if we get that in the future, with a protection option that "applies to subpages" (i.e. the non-hackish way). Cenarium (talk) 21:12, 1 March 2015 (UTC)
@Cenarium: From a technical point of view, to change main from using Template:Editnotices/Page/PAGENAME to (Template:Editnotices/Page/Main/PAGENAME or Template:Editnotices/Page/Main:PAGENAME) - can this be done on wiki, or will it require a patch? — xaosflux Talk 14:25, 2 March 2015 (UTC)
Technically, it can be accomplished locally. Requires editing Template:Editnotice load. Dragons flight (talk) 05:05, 3 March 2015 (UTC)
And fwiw, the editnotice will be shown correctly when redirected to a new location. So one can make the moves then update the template to use the new location, it'll still work in the mean time.  Cenarium (talk) 01:53, 4 March 2015 (UTC)

Search not updating[edit]

The CirrusSearch, which normally updates within 10-15 seconds, appears to have stuck
I corrected misspellings of "received" at 12.37 and 12.38 but both sill appear in the search for recieved nearly 6 hours later - Arjayay (talk) 18:18, 1 March 2015 (UTC)

I thought that it updated once a day, around 04:00 UTC, which is some eight hours off. --Redrose64 (talk) 19:59, 1 March 2015 (UTC)
The old Lucene-search updated once a day. The current CirrusSearch should normally update much faster. See mw:Help:CirrusSearch#Updates. I also get the old revision of your diff currently (you posted the same diff twice). PrimeHunter (talk) 21:01, 1 March 2015 (UTC)
This is even worse than when LuceneSearch index was not updated; at least when it got restarted, even if it took a week, it would completely reindex everything. The CirrusSearch update failure has not been seen before to this extent, I think, and something extraordinary will have to be done to make things right. I don't think they expected that something like this could actually happen, and I will be surprised if they have a way to quickly fix it. Chris the speller yack 21:23, 1 March 2015 (UTC)
We have a script for handling outages. Its one of the first things I built. NEverett (WMF) (talk) 18:21, 2 March 2015 (UTC)
Filed as phabricator:T91216. --MZMcBride (talk) 21:31, 1 March 2015 (UTC)
I just filed phabricator:T91217, then saw your post. It looks like you beat me by a nose. I'll update mine as being a duplicate of yours. Chris the speller yack 21:43, 1 March 2015 (UTC)
Now updates to the index are taking place again as articles are edited, but the ones we already complained about are still not updated in the index, as my earlier post predicted. Chris the speller yack 02:05, 2 March 2015 (UTC)
Some old edits have appeared in the updated search, and some haven't. The real one mentioned above at 12.37 hasn't, but the one I should have posted at 12.38 has, although neither article has been altered since. If there is a definite time period that has not updated, is it possible to give all articles edited in that time-frame a null edit to nudge the search along? - Arjayay (talk) 16:36, 2 March 2015 (UTC)
Grabbed the task and will work on it now. Thanks for filing it! We have a script to handle it and I'll kick it off as soon as I find the timestamps to feed it.NEverett (WMF) (talk) 18:21, 2 March 2015 (UTC)
I believe I've reindexed all the effected pages. I'm off to kick off this process on the other ~880 wikis! NEverett (WMF) (talk) 02:07, 3 March 2015 (UTC)
Well, I was surprised; they actually did have a way to quickly fix it. I sampled updates from yesterday to see what was reindexed, and they seem to have caught them all. This turned out better than I expected, and I have more faith in the CirrusSearch group than I did before. Thanks for resolving this. Chris the speller yack 02:56, 3 March 2015 (UTC)

Using qqx as default language[edit]

We can append &uselang=qqx to an url to get the interface messages name, but what can we do when one wants to get the interface message names also when submitting requests ? I don't see qqx in the language selector in preferences. Cenarium (talk) 19:50, 1 March 2015 (UTC)

What do you mean by "when submitting requests" - is this editing a page? If so, the URL will already include a question mark at some point earlier than (but not necessarily immediately before) the action=edit; there can be only one of these, so leave that alone, and append &uselang=qqx to the right-hand end of the URL. --Redrose64 (talk) 19:56, 1 March 2015 (UTC)
The question is about cases where a displayed message doesn't just depend on the current url but also on what you just did, for example move a page. Adding uselang=qqx does not always reproduce the right situation and message. If you want to find the used MediaWiki messages in such situations then there are tips at Wikipedia:Village pump (technical)/Archive 131#What interface page is used for the results after a successful undeletion? PrimeHunter (talk) 20:52, 1 March 2015 (UTC)
Thanks. Yes, it was when I was moving a page. There are some useful tips, though it may make things easier if the non-overriden default messages (for the site language only, like AllMessages) were indexed when searching mediawiki namespace.  Cenarium (talk) 21:55, 1 March 2015 (UTC)
C: Indeed. This is phabricator:T22858. If you have a clone of mediawiki/core.git, "git grep" is probably fastest. --MZMcBride (talk) 22:00, 1 March 2015 (UTC)

Input[edit]

When did we get Smurf buttons for <inputbox>? See Wikipedia:Editnotice#Creating editnotices. --  Gadget850 talk 21:43, 1 March 2015 (UTC)

Hi. This is the "mw-ui-button" CSS class. It was added to InputBox here: <https://phabricator.wikimedia.org/rEINB7ce80c51a5f0c6ee8e78758afb924bc18b00d525>. You'll see some additional mw-ui-related changes at <https://phabricator.wikimedia.org/diffusion/EINB/>. --MZMcBride (talk) 22:06, 1 March 2015 (UTC)
@MZMcBride: But, what are the CSS specifics of class="mw-ui-button"? The reason I ask is that that class does not yet exist over at Wikia. Therefore, I need to add the class details locally to MediaWiki:Common.css. Unfortunately, I could not discern the simple CSS setup from those phabricator pages.Face-blush.svg Thanks!Face-smile.svgSpikeToronto 18:57, 3 March 2015 (UTC)
Hi SpikeToronto. Newer versions of MediaWiki use Less for Cascading Style Sheets. The relevant code for ".mw-ui-button" is here: <https://github.com/wikimedia/mediawiki/blob/0c8327466bebd44a9c0f60cd010307446b9ead63/resources/src/mediawiki.ui/components/buttons.less> (found via this search). --MZMcBride (talk) 05:22, 4 March 2015 (UTC)

Secure link icon[edit]

Why do we have the secure link icon only for red links: Red link example and logged in secure. --  Gadget850 talk 21:47, 1 March 2015 (UTC)

I'm not sure what you mean. Internal red links have never had a secure link icon, as far as I know. I don't see an icon next to red link example using Vector or Monobook over HTTPS. Maybe you can provide additional details about your setup or a screenshot? --MZMcBride (talk) 21:57, 1 March 2015 (UTC)
Must be me, not seeing it logged out. --  Gadget850 talk 22:13, 1 March 2015 (UTC)
Something in my JS. Possibly User:Anomie/linkclassifier. --  Gadget850 talk 22:20, 1 March 2015 (UTC)
I'm not seeing a secure link icon, but I am seeing the small WP:BLUELOCK for that particular link thanks to User:Anomie/linkclassifier. Something random like slkndkfoewgfoiwenbgflksgewe that isn't create-protected isn't showing any icon here. Anomie 23:27, 1 March 2015 (UTC)

[Global proposal] m.Wikipedia.org: (all) Edit pages[edit]

MediaWiki mobile

Hi, this message is to let you know that, on domains like en.m.wikipedia.org, unregistered users cannot edit. At the Wikimedia Forum, where global configuration changes are normally discussed, a few dozens users propose to restore normal editing permissions on all mobile sites. Please read and comment!

Thanks and sorry for writing in English, Nemo 22:32, 1 March 2015 (UTC)

An incorrect redlink[edit]

I was checking something in the Lyndon B. Johnson article, and saw that the link on his father, Samuel Ealy Johnson Jr., was red. Yet, when I hovered over the link, PopUps showed me that there is in fact an article on Samuel. But the link was red, and I still got the "(page does not exist)" title-text. The Samuel article has existed for years, so it's not one of those situations where it was recently created and the backlink colors haven't changed yet. I purged the page, and that seems to have fixed the problem, at least for me. Here's a screengrab. Is this a known issue? I skimmed Phabricator but didn't see anything similar. — PinkAmpers&(Je vous invite à me parler) 02:27, 2 March 2015 (UTC)

The page history of Samuel Ealy Johnson Jr. shows it was moved to that title one minute after Lyndon B. Johnson was last edited, so it was just a normal red link cached at that time. PrimeHunter (talk) 02:50, 2 March 2015 (UTC)
@PrimeHunter: Ah, that explains it. I should've noticed. Thanks. — PinkAmpers&(Je vous invite à me parler) 06:21, 2 March 2015 (UTC)

Signpost app[edit]

Hello all,

As you may know there is a Wikipedia Signpost Mobile App for Android. It's currently borked, but the source code is on GitHub. We're doing a lot of organizational work at the moment and the mobile app is one of the things we'd like to look into: we're looking into technical improvements behind-the-scenes and we'd like to take the opportunity to more tightly integrate mobile access into things.

The Android project is likely to be fixed soonish; would anyone be interested in working on porting the Signpost mobile app to iOS, or know someone that they think might interested in such a project? ResMar 05:40, 2 March 2015 (UTC)

Suppressing Infobox Person[edit]

Hi, could somebody give me the coding in my preferences to suppress Template:Infobox person in an article and to replace a photo if in the infobox with a formatted photo in place set at 250px with the caption given in the infobox?♦ Dr. Blofeld 14:27, 2 March 2015 (UTC)

Question about italicizing[edit]

I came upon the article Vytautas Mažiulis and noticed that italics in the biography section are not displaying properly. The organization name "Milanese Linguistics Society" is not entirely italicized, however the country and phrase that follows it is italicized for some reason. The code looks correct. Is there something that I'm missing here? How can these italics be fixed? Mamyles (talk) 16:15, 2 March 2015 (UTC)

It's because there was an artificial linebreak after "Milanese". I've fixed it (although those organization names probably shouldn't be italicized anyway per MOS:ITAL). Regards, Orange Suede Sofa (talk) 16:21, 2 March 2015 (UTC)
Looks better, thanks. That also clears up why my browser wasn't displaying a space after "Milanese" in the edit window.
I was just bothered by this issue while reading and do not care if the italics are there or not. Feel free to remove if you think they are inappropriate. Mamyles (talk) 16:33, 2 March 2015 (UTC)

Tech News: 2015-10[edit]

16:42, 2 March 2015 (UTC)

How long does it take for a session timeout?[edit]

I've been experiencing trouble with a bot I'm running in Javascript that signs in as a user and does a lot of stuff that takes ~3-4 hours to finish before finally editing some pages with the results. At the end I'm getting a badtoken error for the edits (which I hadn't gotten until recently, actually.. did something change?), so my only guess is that it takes so long the bot is being automatically logged out before it has time to finish. I am going to try one more time with the "keep me logged in" button selected, but just for information's sake, does anyone know the time limit for a session? Is there any way for a user to modify that setting for him/herself? Has that default value changed recently, i.e. in the last few weeks? Thanks!--Dudemanfellabra (talk) 01:19, 3 March 2015 (UTC)

Ok I ran the code again and got the same error, even with the "keep me logged in" button selected, so now I'm stumped. Just to check to make sure I wasn't being logged out after the run, I visited another page after the error and sure enough was still logged in, so that doesn't seem to be the problem.
The code is set up so that it can be run by two accounts--the bot one (User:NationalRegisterBot) and this one, my personal one (for situations like this, mostly). While signed into my personal account, I just tried to run a smaller section of the code that only takes about 10 minutes to run instead of 3-4 hours but which relies on the same routine (literally the exact same JavaScript function) to edit pages, and it worked, successfully editing the page it was supposed to. This led me to believe that maybe there was just something wrong with fetching a token for the bot account, so I tried running the same function that worked on my personal account while signed in with my bot account, expecting it to fail, but much to my surprise, it worked!
Again, the piece of code that worked and the original 3-4 hour piece of code both rely on the same routine to edit pages at the end... the only difference between the two as far as practice goes is the time it takes for the two functions to run. It seems that since the 3-4 hour code takes so long to run, something freaks out with the edit token, causing it to be invalid. To figure out if that was the case--maybe the edit token had expired or something--I made the code fetch the edit token at the beginning of the run and then again at the end of the run so I could compare the two and got the same output for both (42163ed75d03ff490b30c96e9d59716a54f5440f+\, if that's relevant). Because they were the same, it doesn't appear to me that the token expires over the course of execution, so I'm really stumped. How can a function work in one instance and not in another?
All of this started about 2-3 weeks ago, so I suspect something has changed in the API or somewhere else, but I haven't been able to find any information about that. Does anyone know why I might be encountering the error I'm describing?--Dudemanfellabra (talk) 08:59, 3 March 2015 (UTC)
The test fetching the edit token at the beginning and end of the run should certainly have not given you the same token each time; even fetching two tokens one second apart should give you two different tokens (but both valid) since the end of October. Since you got the same token hours apart, whatever you're using to fetch the tokens is apparently caching them rather than fetching a new one the second time.
Ideally, your bot should be able to attempt the edit, and if it gets a badtoken error it should automatically fetch a fresh token and retry the edit.
As for the time it takes, the configuration is currently using the default of 1 hour for $wgObjectCacheSessionExpiry. I don't see any recent changes to the timeout, but I do see there was this recent change to session handling that probably made page views no longer reset the session timer. Anomie 13:05, 3 March 2015 (UTC)

If I were you, i'd rewrite the bot to make use of mediawiki.api.edit module and friends, which do automatic token (re)fetching, uses promises etc and detects api errors. Examples of its usage can be found here. And make sure to add proper logging in the fails of the request, so that you KNOW what is failing instead of having to guess. —TheDJ (talkcontribs) 13:54, 3 March 2015 (UTC)

I see you are using mw.user.tokens and not checking for bad-token failures. That is probably why it is failing. These tokens are valid 'pre generated' tokens, but your script is supposed to be able to fetch a fresh token if your request fails with bad-token for whatever reason. mw.Api knows about these constraints and applies or recovers from them as required. —TheDJ (talkcontribs) 14:06, 3 March 2015 (UTC)

Fixing bulleted list inside an unbulleted list[edit]

This is basically a repost of MediaWiki talk:Common.css#Fixing bulleted list inside an unbulleted list to get some more attention.

When using {{bulleted list}} inside {{unbulleted list}}, it doesn't work because {{unbulleted list}} adds a class to the outer list without a means to revert that inheritance in inner lists. To make stuff less complicated, I propose to make the styles of unbulleted list apply strictly to the direct children of the outer list, so as to prevent the inner lists from inheriting the styles.

Example:

{{unbulleted list
| This
| {{bulleted list|E|F}}
}}
  • This
    • E
    • F

Timothy G. from CA (talk) 03:50, 3 March 2015 (UTC)

Commented there. -- [[User:Edokter]] {{talk}} 09:46, 3 March 2015 (UTC)

Does Wikipedia collect statistics on the up-times and response-times of its servers?[edit]

At times, I have been unable get a page from Wikipedia, and it appears that the servers are either down or temporarily overloaded. When this happens, I do not know if the problem is with my computer, the network, or Wikipedia. I usually diagnose the problem by trying other web sites, but the really does not tell me if anything was wrong with Wikipedia.

My email provider gives me a link to pingdom, where I can easily tell if the email servers were down. Does Wikipedia keep track of down-times, response times, or server loads? Where should I look?

If I had a graph of Wikipedia response-times, I could plan on using Wikipedia when it was lightly loaded.

Thanks.

Comfr (talk) 04:32, 3 March 2015 (UTC)

Lots of data is available at http://gdash.wikimedia.org/ - there is no one graph of some "response-times" though... --Malyacko (talk) 08:32, 3 March 2015 (UTC)
There is some information on status.wikimedia.org, eg. the response times for HTTPS wikipedia. --Sitic (talk) 22:13, 3 March 2015 (UTC)

Orphan tags not showing up[edit]

There is something wrong with Template:Orphan, looking at the edit summary this edit: [21] hid the cleanup tag and as a result the tag has disappeared from 99 of our articles. Is there a fix fir this? - Knowledgekid87 (talk) 04:53, 3 March 2015 (UTC)

@Knowledgekid87: Please see Template:Orphan#Visibility, which explains that the template is only visible for the current & previous month (e.g. not visible on A Gentleman's Kiss), but is always visible within {{multiple issues}} (e.g. visible on AUKcon). The articles are still categorized and still on your report. The template instructions also contains a fix that might interest you. Good luck! GoingBatty (talk) 05:49, 3 March 2015 (UTC)

subst not substituting[edit]

What am I messing up here? Martijn Hoekstra (talk) 10:19, 3 March 2015 (UTC)

Substing does not work inside <ref> or other parser tags. -- [[User:Edokter]] {{talk}} 10:25, 3 March 2015 (UTC)
Thanks! Live and learn. Martijn Hoekstra (talk) 10:28, 3 March 2015 (UTC)

Linking from medal icons[edit]

Hello.

I was trying to make the medal templates to link to an optional target by adding link={{{link}}} in {{gold1}}, so the link can be passed on to the {{sort}} template. It works fine, but when you don't provide a link to {{gold1}}, it links to the gold medal icon itself, instead of not linking to anything at all. What is it that I do wrong?

For an example, check the Brittany Bowe article, section "Results overview", where I've linked the gold medal icon in the last row, left-most column, to an article, and it works, but the other gold medal icons in the same row link to the icon itself. I don't want that, and why does this occurs when the link parameter supplied is empty?

HandsomeFella (talk) 14:42, 3 March 2015 (UTC)

This is result, that you wanted? See my edit. --Edgars2007 (talk/contribs) 15:26, 3 March 2015 (UTC)
Yes, it is. But the parameter is empty by default, so why does one have to add that code to avoid linking to the medal icon? HandsomeFella (talk) 15:50, 3 March 2015 (UTC)
I see now that I had missed that pipe character. Thanks. HandsomeFella (talk) 17:21, 3 March 2015 (UTC)

Table-end templates[edit]

We have a vast number of templates for ending tables, whose content is |}.

I propose that we redirect them to {{End}}, which has the same content, and replace instances with {{End|Foo}}, where "Foo" is the name of the template which starts the table, as in this edit.

Eventually, a bot could do substitution for us. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:10, 3 March 2015 (UTC)

  • support --  Gadget850 talk 20:06, 3 March 2015 (UTC)
  • (ec) Redirecting is fine, but replacing the templates with {end|...} isn't very usefull; it does somewhat confuse the otherwise obvious pairing of templates. -- [[User:Edokter]] {{talk}} 20:08, 3 March 2015 (UTC)
  • Oppose. The templates are from pairs like {{Public art header}} and {{Public art footer}}. The current system gives us the ability to easily change implementation of the pair, for example if something is added to the bottom or it's no longer a table. Your suggestion {{End|Public art table}} could in theory change implementation by testing the parameter value in {{End}} and do something different for Public art table, but it requires that users always remember the parameter and always give it the exact same value. The parameter is ignored by {{End}} so users will see no difference if they omit or misstype it. And {{End}} is fully protected for good reasons but it means most editors would need help to change implementation of low-use cases like the currently unprotected {{Public art footer}}. We would also lose the ability to use Special:WhatLinksHere/Template:Public art footer, for example to compare it with Special:WhatLinksHere/Template:Public art header. Pairs like {{Public art header}} and {{Public art footer}} are also easier to spot in wikitext, especially when they start the line. I see no significant advantages of the proposed system. It would make it easier to fix if wikitext suddenly stopped accepting |} as code for table end but that seems highly unlikely. PrimeHunter (talk) 20:26, 3 March 2015 (UTC)
    • You appear to be considering improbable hypothetical scenarios, such as "[if] it's no longer a table" and "testing the parameter value in {{End}}". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:52, 3 March 2015 (UTC)
      • Eh yeah of course those are hypothetical. Wasn't he just saying that that was the whole point ? To keep that option open ? Such changes have happened in the past. Switches between divs and table structures have been quite common, and I don't think we have seen the last of them yet. I advise against substitution. Redirects could be possible, but might lead to people using the target instead. Indirect transclusion seem appropriate, but will wast more resources. —TheDJ (talkcontribs) 00:27, 4 March 2015 (UTC)
      • The proposal looks like a solution looking for a problem. If there is no problem to solve then there is no reason to destroy the current flexibility and create potential problems. Who knows whether {{Col-end}} will always have a table end? If we replace uses by {{end}} then maybe it will, because we have made it very hard to change it. It currently says <p></p>|} (newlines omitted in code display here), so I guess the current version would actually be excluded from the proposal. Lots of end templates are not an isolated table end or may not be so in the future. {{Multicol-end}} is </div>|}</div>. {{Football squad end}} is |}|}. {{Extended football squad end}} had been |} since 2010 but added code to give or request a source last week. And consider {{S-end}} where the documentation says (I don't know whether it actually works like that currently): "The various succession box templates are considered navigational content and thus are excluded from print in the PDFs and printed books. However, {{end}} can be used to end any table, thus that template cannot be excluded from print. However, {{s-end}} can be excluded without any negative repercussions. Similar templates that are unambiguously the end of tables which should be excluded in print should redirect here, rather than to {{end}}." If we redirect the templates then it's certain and not just a hypothetical that some users will use the targets instead. PrimeHunter (talk) 01:02, 4 March 2015 (UTC)
  • I agree with redirecting end templates to {{end}}, but we shouldn't replace the instances for the reasons that PrimeHunter spells out. If someone does ever need to change one of the end templates, then that would be the difference between a one-edit task and a task requiring bot work. — Mr. Stradivarius ♪ talk ♪ 00:52, 4 March 2015 (UTC)