Jump to content

Wikipedia:Village pump (technical)

From Wikipedia, the free encyclopedia

This is an old revision of this page, as edited by Iry-Hor (talk | contribs) at 06:36, 9 January 2018 (→‎A request on a generic infobox). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

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

Newcomers to the technical village pump are encouraged to read these guidelines prior to posting here. If you want to report a JavaScript error, please, follow this guideline. Questions about MediaWiki in general should be posted at the MediaWiki support desk.


Archive-date parameter

Very strange spike in views in several articles

A number of low-profile articles about places in the Middle East ('Ain Ghazal, Ayn Ghazal, Al-Bassa, Al-Bassah, Al-Buwaydah, Buwaydah, Al Baqa, Baqa'a refugee camp, Diban) started receiving huge increase in views this year: Pageviews Analysis (Up to 50,000+ each). If you check 'Agent' parameter on the left, you'll see that until June 9 it was "Spider", and then "User". Views for some of those are almost identical day-by-day, and it stops and starts randomly. It's 100% unnatural. What can explain this? --Triggerhippie4 (talk) 23:01, 2 December 2017 (UTC)[reply]

Web crawler maybe? --Malyacko (talk) 00:10, 3 December 2017 (UTC)[reply]
But why these articles? --Triggerhippie4 (talk) 00:18, 3 December 2017 (UTC)[reply]
I have seen these sorts of pattern before, but I can see no sense to them, unless the pages are being used as a covert messaging channel. All the best: Rich Farmbrough, 14:58, 4 December 2017 (UTC).[reply]
It's possible these towns were discussed and linked in an Arabic news article or social media site. External linkage can create traffic spikes that sustain over a couple weeks. It's sometimes possible to track down the source but sometimes not. -- GreenC 15:21, 4 December 2017 (UTC)[reply]
Yes, as Onceinawhile pointed out, it is only the English version which gets these hits.[1], [2] If it had been in the Arabic news, then also ar.wp articles would have gotten hits. Huldra (talk) 21:56, 6 December 2017 (UTC)[reply]
No way. These places are not in the news, numbers are identical for different pages and they suddenly disappear and reappear. Only English wiki pages of these articles are affected. Click on Pageviews Analysis above. --Triggerhippie4 (talk) 18:14, 4 December 2017 (UTC)[reply]
In my experience, when you see a pattern like that it's generally because a celebrity has tweeted a link to the page for some reason or another (there's a particularly extreme example of the phenomenon here). ‑ Iridescent 18:21, 4 December 2017 (UTC)[reply]
That's different. Your example is a developed article about known subject of broad interest (urban legend/internet meme kind of thing). It has at least hundreds of views each day and it changes more gradually. I'm talking about stubs about unrelated non-notable villages that no one tweeted about, and that go back and forth like this simultaneously: Pageviews Analysis. --Triggerhippie4 (talk) 02:57, 5 December 2017 (UTC)[reply]
It's really difficult to tell. For example Al-Bassa is linked from this page and maybe that page is the one getting the traffic due to news etc.. with some viewers clicking through to Wikipedia. Or possibly that page has a maintenance script that has gone haywire, polling Wikipedia for existence. Do you have any theories? -- GreenC 15:14, 5 December 2017 (UTC)[reply]
I suppose it's some technical error. I decided to bring it here because it's seems too odd that articles that always averaged less than 100 views a day, now accumulated more than 10 million views in the last 8 months. --Triggerhippie4 (talk) 17:01, 5 December 2017 (UTC)[reply]
User:Onceinawhile noted the same, back in September, here. As one of the editors who have done lots of work on both articles, Ayn Ghazal and Al-Bassa, I am totally bewildered. I am quite familiar with the fact that whenever a place in the Middle East is in the news, then the page views goes through the sky on that article. But this is something different. For one thing, none of the places have been particularly in the news, lately, AFAIK. Secondly, look at the page views for both articles for the last 90 days: they follow each other closely. It would be interesting to see if the page views came from the same IP address...or if this is just some technical error. Huldra (talk) 20:49, 6 December 2017 (UTC)[reply]
My guess is machine generated because there is no collateral spike in connected articles. For example Ayn Ghazal, which gets 15k-30k hits a day, the lead section prominently mentions Ofer. But Ofer is only getting about 10 hits a day. If they were live people visiting Ayn Ghazal, some percentage would be clicking through to Ofer, but that is not happening: (PageView Ayn_Ghazal|Ofer). Ofer stays at a very low flat rate while Ayn Ghazal jumps up and down with high numbers (higher than Israel itself at times). Can't think how it would make sense unless machine generated. -- GreenC 21:24, 6 December 2017 (UTC)[reply]
I agree that this numbers are false, in one way or another. The question is: is it the Wikipedia page views counting which is wrong, or is some machine sitting somewhere viewing these articles over and over again? Huldra (talk) 21:28, 6 December 2017 (UTC)[reply]

@BBlack (WMF): please could you let us know who is best placed to investigate this? It has been going on for many months now. Perhaps a quick look at the IP addresses of the tens of thousands of daily hits would quickly shed some light on the issue. Onceinawhile (talk) 11:16, 16 December 2017 (UTC)[reply]

I looked into this, specifically 'Ain Ghazal on November 29, 2017. I don't know how much I'm allowed to say, so I'll keep it generalized :) On that day, most of the hits originated from a single city. It is a large city (where we'd expect more viewers), but the ratio was still big enough that I'd say it's inline with other false positives I've seen. Indeed, when I compare desktop pageviews vs. mobile web (or mobile app), you'll see the spike in traffic is strictly on desktop. For the English Wikipedia, generally mobile viewership will be on par if not more than desktop. As such, I would say with some confidence that this article is being continually scraped by an undeclared bot. I suspect the same is true with the other articles you speak of -- just try comparing desktop vs mobile. This is the same trick that The Signpost uses to filter out their WP:TOP25 reports. The issue unfortunately is common. Currently I believe the logic to identify bots and other non-human sources is solely to go by the user agent. Of course if someone were to spoof their user agent to look like it is human, there's no a whole lot we can do. More on that at phab:T123442 and in the Topviews FAQ (I can add the same information to the Pageviews FAQ, since this issue is not limited to the top-viewed articles API). It is possible to dig deeper and get more answers, but I should let you know this sort of research is very tedious and time consuming. The takeaway here I think is to improve bot detection, as it seems every time we investigate a report it is the same story of an undeclared bot. Hope this helps! MusikAnimal (WMF) (talk) 19:12, 16 December 2017 (UTC)[reply]
@MusikAnimal (WMF): thank you for looking into this, and for confirming our suspicions. I didn’t understand all of the technical language in your post, but imagine the results you describe mean that the traffic is mostly emanating from one person’s computer. I just can’t understand what possible motive someone would have to set up a bot crawling over these articles of relatively minor interest.
Pinging @Triggerhippie4, Malyacko, Rich Farmbrough, GreenC, Huldra, and Iridescent: in case you hadn’t seen. Onceinawhile (talk) 21:55, 16 December 2017 (UTC)[reply]
One reason might be inclusion in the top views list. It could have political repercussions, such as using the metric to embarrass Israel about Wikipedia "spotlight" on these villages. Or it might be a script gone haywire and no one noticed. The mystery is probably better than the truth :) -- GreenC 22:17, 16 December 2017 (UTC)[reply]
Whatever it is, they seem to have gone on holiday for the last three days, as the page views have been down to normal since the 28th, Huldra (talk) 23:21, 30 December 2017 (UTC)[reply]
I will say that for that particular article and day that I checked ('Ain Ghazal on November 29, 2017), the "majority" of pageviews did not originate from those universities, unless they were accessing the site via VPNs/proxies that belong to another country. MusikAnimal (WMF) (talk) 14:54, 7 January 2018 (UTC)[reply]

Edits (rather than the text of edits) being imported into Wikipedias of other languages

(Before the heading title above raises false alarms that my account's login credentials have been somehow compromised, they have not. This issue appears to involve some sort of (new?) MediaWiki importation tool.)

I've been editing Wikipedia for a long, long time, around a decade and a half, during which many articles I wrote or contributed to have been copied and pasted and translated into Wikipedias of other languages. That's fine/great; no problem! I'm happy/delighted to share the results or work product of my edits in other languages! But I just encountered something new and different.

I was just given a welcome message to the Bihari Wikipedia by another user, thanking me for my edits there. This surprised me as I don't speak or understand the Bihari languages, so why would I ever have edited there? So, I checked bh:Special:Contributions/Lowellian. And I recognize those edits: I did make those edits, but I made those edits to the English Wikipedia, not the Bihari Wikipedia!

After poking around the page histories some more (and navigation is difficult due to the aforementioned fact that I don't understand Bihari, so I have to do some informed guesswork clicking on links), for example https://bh.wikipedia.org/w/index.php?title=Template:Countries_and_territories_of_Southeast_Asia&limit=500&action=history, I can see that my edits have been somehow "copied" from the English Wikipedia to the Bihari Wikipedia, and in a way different from a normal page move: normally, when a page is moved, the edits are also moved with it so that each edit still only appears in one page history (unless it was a copy-and-paste move, in which case the text of the edits but not the edits themselves get copied/moved). But this situation appears to be something different, in which my edits are now duplicated so that the same edit history appears on both the English and Bihari Wikipedias, accredited to me.

This is fundamentally different from normal copying-and-translation of articles into other languages since that copies the prose/text (the work product of edits) rather than the actual edit history (the edits themselves). This concerns me because those page histories make it look like I have been editing the Bihari Wikipedia when I have never done so. I don't want, for instance, people complaining to me about edits I made on another-language Wikipedia that I had no knowledge of and did not make. (This is different from normal copying/translation of articles into other-language editions, since normal copying/translation doesn't make it look like I directly edited those other-language editions.)

This all appears to be done by some sort of (new?) MediaWiki importation tool? Can someone in the know provide background for what all is happening here?

Lowellian (reply) 22:42, 12 December 2017 (UTC)[reply]

@Lowellian: from bhwiki's log I can see that @SM7: has imported these using the normal transwiki import process. This part is OK and normal. phab:T179832 describes some of the intricicies of imported usernames and how it is being dealt with. — xaosflux Talk 23:32, 12 December 2017 (UTC)[reply]
Okay, I don't really understand the issues described in phab:T179832 or whether they apply to this specific case, but maybe someone can at least clear this up for me: if you look at https://bh.wikipedia.org/w/index.php?title=Template:Countries_and_territories_of_Southeast_Asia&limit=500&action=history, some edits are marked "imported>" followed by the username. My problem is that my edits, at least, seem to have been imported without even that "imported>" designation, as if I made the edits directly to the Bihari Wikipedia, when I did not. My edits should be marked "imported>" if they were imported. Is this something that has been looked at and fixed so that it won't happen in the future? —Lowellian (reply) 23:41, 12 December 2017 (UTC)[reply]
See meta:Help:Import for the feature. It's from 2003.[3] You can select English as interface language at Special:Preferences at a foreign wiki. This may make it easier: https://bh.wikipedia.org/wiki/Special:Preferences?uselang=en. I don't know when an edit will say "imported>". I haven't seen that before so the anomaly may not be that it's "missing" from you but that it's shown for some of the others. PrimeHunter (talk) 00:02, 13 December 2017 (UTC)[reply]
The import process recently got a few enhancements, which is what phab:T179832 was talking about.
When importing edits, the importer must specify the source of the import, and can choose whether imported edits by users who exist in SUL are attributed to the local account for the user or in a fashion like "en>Example" that links to the source wiki from the edit history.
Of course, we also have over a decade of imports that were done under the previous system, where edits were attributed to a local user even if that user didn't exist. A cleanup script is being run on the servers to update the attribution to the SUL user where possible and with a "imported>" prefix otherwise. Since Lowellian's account is in SUL, they were attributed to that SUL account. BJorsch (WMF) (talk) 02:00, 13 December 2017 (UTC)[reply]
Thanks for the explanation. In that case, I would like to strongly make the following suggestion: in all cases when edits are imported, including for SUL accounts, there should always be some sort of marker/designation in the page history that the edit is an import in order to avoid the incorrect implication that the editor made that edit directly to that particular language-edition Wikipedia. It is inconsistent and misleading that only some users' imported edits are marked as imported while other users' imported edits are not marked as such, and unfair to make users look like they were spamming English edits onto Wikipedia editions in other languages when they did not do so. —Lowellian (reply) 02:38, 13 December 2017 (UTC)[reply]
The best place to make suggestions is in Phabricator. The best indication would probably be a tag. But note there's no way to reliably identify all past imported edits. Anomie 14:07, 13 December 2017 (UTC)[reply]
See also Wikipedia:Requests for page importation where we handle requests to copy histories TO enwiki from elsewhere for reference. — xaosflux Talk 23:36, 12 December 2017 (UTC)[reply]

Local accounts attached without a visit (and welcomed without an edit)

Today I have been getting Welcomed on both the Latvian and Kazakh wikis. I have never edited over there. I checked my contribs over there, but nothing shows. Could this be part of the same issue [ the previous topic about imported contributions ]? Bit of a coincidence that they both happened today and it has never happened before. — Insertcleverphrasehere (or here) 07:26, 13 December 2017 (UTC)[reply]

This is exactly what the previous topic is about. These welcome messages are being triggered by your edits being imported over to other-language-edition Wikipedias without your knowledge, making you look like a new editor to welcomebots, so they start leaving you welcome messages. Everyone complaining here, please add your voice to that topic if you have complaints about how this process is happening. It is wrong that edits from the English Wikipedia are being imported over to other-language Wikipedias without any indication that those edits are imported. —Lowellian (reply) 21:16, 13 December 2017 (UTC)[reply]
It happened on the Punjabi Wiki as well... What is going on here? — Insertcleverphrasehere (or here) 10:38, 13 December 2017 (UTC)[reply]
Insertcleverphrasehere, local wiki accounts get attached to your global account the first time you visit that wiki while logged-in. All three of lv:, kk: and pa: were attached to your account this morning. On some wikis you automatically get a welcome message the first time you log in. So I don't think this has any relation to the topic of imported edits (and have therefore separated this topic from that section). --Pipetricker (talk) 11:41, 13 December 2017 (UTC)[reply]
Turned out I was wrong about that. --Pipetricker (talk) 10:56, 17 December 2017 (UTC)[reply]
I wonder why all three happened today though? Must be the rollout of some kind of welcome bot. — Insertcleverphrasehere (or here) 11:47, 13 December 2017 (UTC)[reply]
As Pipetricker said: "local wiki accounts get attached to your global account the first time you visit that wiki while logged-in. All three of lv, kk and pa were attached to your account this morning." According to Special:CentralAuth/Insertcleverphrasehere you visited a large number of wikis for the first time today. Many users have been confused by such welcome messages in foreign languages. I have thought about suggesting at meta to ban welcome messages to users who have no edits at a wiki and didn't create the account there but just had it created automatically by viewing a page. PrimeHunter (talk) 11:55, 13 December 2017 (UTC)[reply]
Well, I definitely didn't visit all of those wikis today. I did however submit a SQL Query on Quarry for the first time, which required me to authorise it on Mediawiki, which is probably related? Still strange though. — Insertcleverphrasehere (or here) 12:04, 13 December 2017 (UTC)[reply]
That's unrelated. I have done no such thing recently and was welcomed on three wikis also. :) --Izno (talk) 12:23, 13 December 2017 (UTC)[reply]
Izno, are you saying the SQL query is unrelated to the attachment of a number of local accounts at about the same time, despite Insertcleverphrasehere not having visited those wikis? --Pipetricker (talk) 14:04, 13 December 2017 (UTC)[reply]
Yes. --Izno (talk) 15:41, 13 December 2017 (UTC)[reply]
Actually, it probably is related. The process of attributing those old imported edits to the SUL account includes creating the local account for the attribution to belong to, which would trigger the welcome message on wikis that do automated welcome messages on account creation. Anomie 14:10, 13 December 2017 (UTC)[reply]
Probably related to the topic of imported contributions, that is. --Pipetricker (talk) 16:46, 13 December 2017 (UTC)[reply]
I'm seeing this too. Of the 32-and-still-rising wikis that Special:CentralAuth/Cryptic says I attached in December 2017, I've only ever been to two - Gujarati Wikiquote and Persian Wikivoyage, and those only afterward, because I can'tcouldn't get the notifications for the welcomebot messages to go away no matter what I try. —Cryptic 14:50, 13 December 2017 (UTC)[reply]
(In case that crops up for anyone else - what finally worked was logging in directly on those wikis and checking off the notification at Special:Notifications, not just the dropdown on the sidebar.) —Cryptic 15:21, 13 December 2017 (UTC)[reply]
I also got two or three notifications. Emir of Wikipedia (talk) 14:55, 13 December 2017 (UTC)[reply]
CentralAuth shows many accounts being added. Emir of Wikipedia (talk) 15:01, 13 December 2017 (UTC)[reply]
I notice now that ilo.wikipedia.org was attached to my account tonight, when I was asleep and my computers were turned off. It's not in my browser history previous to that, and I have no contributions there (and have received no welcome). --Pipetricker (talk) 15:10, 13 December 2017 (UTC)[reply]
My wuu.wikipedia account was created 45 minutes ago when my computer was turned off. There is clearly some current process creating accounts at other wikis without user action, maybe releated to work on phab:T179832: "Handling of imported usernames". I don't currently have edits registered at wuu:Special:Contributions/PrimeHunter but maybe there are old imported edits somewhere which have not been added to my contributions. PrimeHunter (talk) 16:05, 13 December 2017 (UTC)[reply]
Yes, there is some current process, as I stated in the section above. See phab:T181731 for the task about it. I note that the script is already done with all the "small" wikis. The wikis still in progress or pending are dewiki, enwiki, eswiki, fawiki, frwiktionary, hewiki, huwiki, itwiki, jawiki, kowiki, metawiki, nlwiki, nowiki, plwiki, ptwiki, rowiki, ruwiki, svwiki, thwiki, trwiki, ukwiki, viwiki, wikidatawiki, and zhwiki. BJorsch (WMF) (talk) 00:40, 14 December 2017 (UTC)[reply]
If the script is still pending for larger wikis, then those future runs should be stopped and held off until the problems people are complaining about are fixed. As a starting point, per User:Anomie's suggestion above, it should tag imported edits of SULs as imports instead of misleadingly implying that they were normal edits to Wikipedias of that particular language-edition. —Lowellian (reply) 06:33, 14 December 2017 (UTC)[reply]
Something should be done. See Wikipedia:Village pump (miscellaneous)/Archive 57#Rogue bot or what on other language wikipedias where I raised this. Andrewa (talk) 10:18, 14 December 2017 (UTC)[reply]
I see those things as honestly unnecessary. It causes a one-time notification on a per-wiki basis where any of your edits have been imported. Given that there are only 20 wikis left, spending time now would only prevent finishing the task while waiting for the extra work that is already planned-for down the road. --Izno (talk) 12:26, 14 December 2017 (UTC)[reply]
The problem is not really the welcome notifications. That's just a side effect / symptom of the ultimate cause and real problem, that we are being attributed, without our permission, to edits that we did not make. —Lowellian (reply) 20:58, 14 December 2017 (UTC)[reply]
We made our edits to articles which according to the licences we agreed to can be copied provided they are attributed. So it seems to me we agreed to the copies being attributed to us. (Maybe I misunderstood what you mean, but any way if it isn't a technical issue the tech Village pump isn't really the place to discuss that.) --Pipetricker (talk) 23:26, 14 December 2017 (UTC)[reply]
Attribution has multiple components: who wrote something, what was written, when it was written, and where it was written. The problem is that imports without tagging misrepresent one of those components: where it was written. These attributions aren't accurate because they are edits made to a specific wiki that are now being attributed as edits to a different wiki. And this is certainly a technical issue: it started out as a technical question asking why certain unusual behavior was being noticed, and it has continued as a technical discussion that includes how to, on a technical level, get those edits properly attributed; see the comments below where Xaosflux works out a technical solution to add tagging. —Lowellian (reply) 23:40, 14 December 2017 (UTC)[reply]
When you make an edit, above the edit box there is a notice containing the sentence "Work submitted to Wikipedia can be edited, used, and redistributed—by anyone—subject to certain terms and conditions." So by making the edit, you implicitly gave permission for reuse. Checking those T&Cs, I see that attribution allows several alternatives; one of them is "or c) a list of all authors". By crediting each individual edit to the person who made that edit, even if that edit was on another wiki, sufficient attribution is given. --Redrose64 🌹 (talk) 11:21, 15 December 2017 (UTC)[reply]
No, it isn't, and the legal language being quoted here only says that reuse and redistribution is allowed with proper attribution, which actually supports my point. Wikipedia allows reuse and redistribution with proper attribution, and the problem here is that the attribution is wrong. Redistribution copies text without claiming that the text was originally written directly to the copy. An attribution that claims someone wrote something somewhere that they did not write it (directly to the copy, instead of somewhere else that is then reproduced by the copy) is a fundamentally incorrect attribution, particularly when, as in this case, the copy is only a partial, non-identical copy, being in a different language. Copying over text from one wiki to another is not the same thing as claiming someone wrote directly to a wiki that the person did not. —Lowellian (reply) 22:32, 19 December 2017 (UTC)[reply]
  1. I see you have twice listed four alleged requirements for proper attribution. The policy at Wikipedia:Copyrights#Re-use of text says that attribution under CC-BY-SA has exactly one requirement, namely that you must "provide credit to the authors". It further says that "a list of authors" (even if just plain text), is adequate. Do you believe that the policy is wrong? Can you find anything in Wikipedia:Text of Creative Commons Attribution-ShareAlike 3.0 Unported License that requires, for example, that attribution include "when it was written"?
  2. Why does it actually matter to you that your revision has been copied from the enwiki database over to another one, without saying "copied from this other database" on it? Do you think that someone (who?) will think you've done something disreputable, as a result of omitting such a label? If you're familiar with the 5 Whys analysis, then I'm very interested in the "fifth why", or the underlying problem.
WhatamIdoing (talk) 20:19, 20 December 2017 (UTC)[reply]
Exactly, attribution must "provide credit to the authors". That is the problem: when a credit claims someone wrote something somewhere that they did not write it, then that credit is wrong, since the actual editor is the import script, not the person claimed by the import script. Consider the following example: Alice adds the text "lorem ipsum foobar" to an article. Bob copies that "lorem ipsum foobar" text and then pastes it into another article. Alice may be the originator of the text, but the edit history would correctly show Bob, not Alice, as the editor who added it to the second article. If that text is disputed on the second article, then Bob, not Alice, is accountable for adding that text on the second article. But here, the import script is breaking that standard and instead giving the false appearance that Alice was the editor who added the text to the second article, so that Alice becomes the person who is then held accountable.
False attribution is false attribution; it should be corrected without any further justification being needed. I would think that accuracy and truth are fundamental to the goals of Wikipedia as an encyclopedia. That said, at the very least, these wrong credits also give the false appearance that the author is spamming a wiki with text in the wrong language for that wiki. On top of that, different wikis have different stylistic guidelines regarding things like formatting or titling conventions. This gives the false appearance that the author is ignorant or willfully ignoring those guidelines and conventions. Furthermore, what if text added is controversial? The author could then potentially be accused of, for example, POV-pushing on a wiki that the author has never even edited.
Lowellian (reply) 22:03, 20 December 2017 (UTC)[reply]
Is there an example of "false attribution"? You might copy the URL of a diff from a history page on another Wikipedia. Lots of fake books are sold by scammers who have copied various articles. They usually include a list of authors from the history pages to comply with the re-use procedures, so your user name may be printed by them. It's not false attribution. Johnuniq (talk) 23:45, 20 December 2017 (UTC)[reply]
Let's have a look at a page that Lowellian created - Template:Undated (Revision history) and the same at bh.wikipedia - bh:टेम्पलेट:Undated (Revision history). Other people made edits, and checking the lists together, there is a remarkable correspondence, apart from one or two users like Enterprisey / APerson which can be explained as a username change back in May 2016. Oh, I see that I'm in both lists too. Did I make this edit? There's no denying it, guilty as charged. Did I also make this edit? Well, it says that I did, but I don't recall ever making that edit on that wiki. However, if that edit is going to be credited to anybody, I'd rather that it was credited to me and not to somebody else pretending that it was their idea. Am I kicking up a stink? No. --Redrose64 🌹 (talk) 23:57, 20 December 2017 (UTC)[reply]
If these attribution issues don't bother you, that's fine, but please don't just dismiss others' concerns as "kicking up a stink". I'm not the only one who has raised concerns about the behavior of these imports. Why is it so terrible to have an imported edit be credited to both the original author and the import script via an automated tag? Automated tagging doesn't hurt you in any way, while it improves the situation for those who do care about proper attribution. That's all I'm asking for: that the import script should place a tag indicating, in addition to the original author of the text, that the edit is an import, so that there is a way to distinguish between edits made directly and edits that are imported. —Lowellian (reply) 22:49, 21 December 2017 (UTC)[reply]
Do you see the imported edit as saying not only "Lowellian wrote this", but actually "Lowellian personally placed this content on this wiki"?
I don't think that's what's intended. The point is to identify who owns the copyright, not which software database you originally placed the content in. (In your example, the copyright holder would be Alice, not Bob; if Bob copied Alice's text from one article to another, then Bob needs to include an edit summary that identifies either who originally wrote the text or which article it came from). WhatamIdoing (talk) 18:02, 21 December 2017 (UTC)[reply]
Having the implication be "this user personally placed this content on this wiki" is not intended, but it is what is happening. That is why it is so important to have some sort of tag indicating that it's an import rather than something written directly. —Lowellian (reply) 22:49, 21 December 2017 (UTC)[reply]
Who do you think interprets it that way? Not you, of course, because you already know that's not what happened. Not the admin who imported it, or the regulars at any wiki where importing edits is typical, because they already know how it works. So who do you think would be both aware of these imported edits (i.e., not >99.9% of readers) and honestly confused by it? WhatamIdoing (talk) 01:28, 22 December 2017 (UTC)[reply]
The only reason that I even know about these imports is because of those welcome alerts that drew me into this issue. I've been an admin over a decade and still didn't even know about of this import mechanism until I started asking questions in this discussion, which makes it very likely that >99.9% of editors aren't aware there is an import mechanism and upon seeing those untagged edits while examining edit history (which will become ever more common as those wikis grow in the future and more and more editors join and edit) would think they were regular edits and not know that they were imports. This issue is fresh in the minds of the editors here in this discussion right now, but we cannot expect editors outside this discussion to be aware of it years from now. That's why we need a tagging mechanism, so there will be not be any long-term future confusion. —Lowellian (reply) 23:01, 22 December 2017 (UTC)[reply]
If you're a regular editor at a wiki where this is done, then you will have seen this in dozens of articles at your wiki. If you don't know, you'll ask, and one of the other regular editors will tell you.
This has been done for years at the German Wikipedia. It looks like you've got about 10 imported edits there for every "real" one you've made directly. The numbers are very similar for me. Apparently, neither of us have ever been asked about those imported edits. I don't think that I've ever seen a question or complaint about an imported edit, and I assume that's true for you, too, or you'd have already known that it was possible to import edits. I therefore think that it's reasonably safe to assume that the likelihood of some hypothetical future editor yelling at us for doing something "wrong" on another wiki is pretty close the actual experience of it never having happened before in all these past years. WhatamIdoing (talk) 07:10, 23 December 2017 (UTC)[reply]
Actually, that you're saying that this has been done for years at the German Wikipedia actually raises another issue about there being no way from the edit history to tell when the imports were done. But to get back to the original issue, only a very small number of my edits have thus far been imported to the German Wikipedia, concentrated on a few articles. It is not reasonable to conclude from a small number of imports that that there aren't going to be complaints as the number of imports grows. And I will reiterate that fundamentally, attribution should be correct without needing to be justified. Tagging imports helps the situation for those of us who care about this attribution issue and has no effect on those who do not care, so there's no reason not to do it. —Lowellian (reply) 17:22, 23 December 2017 (UTC)[reply]

You've had about 200 edits imported to the German Wikipedia so far, which the median editor would likely not describe as a small number, and about as many again across some other wikis (one at tewiki, two at arzwiki, three at gdwiki, orwiki, and lmowiki, six at mlwikisource, eight at hiwiki and simplewiki, 10 at elwiki, 12 at the testwiki, 15 at maiwiki, 20 at bhwiki, 24 at knwiki, 34 at mlwiki, and 52 at the English Wikibooks, for anyone who's keeping count).

And, as the edits are attributed to you, I suggest again that the attribution is already correct. It appears to be your personal preference for attribution to say which wiki you happened to be using at the time your original edit was saved, but I have found no actual requirement to do so. I'll cheerfully admit that I'm wrong on that point when you quote me a line from the license that says that attribution must include information about which sub-domain of which website you first posted your copyrighted content to, rather than just your name.

This is where the disconnect seems to be. You seem to have gotten the idea that there is only one "correct" way to provide attribution, and that this One True Way™ requires four specific details, and the current system provides only three of them. I can find no source that supports your belief that anything more than your username is required. For that matter, I can't even find out whether your alleged requirement of "where it was published" refers to the physical location where it was published, the legal/copyright location where it was published, or the name of the larger work that it was included in. (There is, after all, a big difference between "printed in China", "published in the United States", and "published as part of the Anthology of Something", and (given the realities of printing costs) all three of those might be simultaneously true for any given work.) So at the risk of sounding rather Wikipedian about this, please cite your sources. I've cited mine: the license doesn't mention any requirement beyond identifying (in some fashion) the human who originally created and licensed the work in question, and which work is so licensed. WhatamIdoing (talk) 03:49, 27 December 2017 (UTC)[reply]

The sources you cite do not support the argument that untagged imports are providing proper attribution. You've cited a license that says that for redistribution, credit must be provided to the authors. But this isn't a simple case of redistribution, and credit has not been provided to the authors. Redistribution should not change the meaning of the underlying text; if it does, then it is not just redistribution, but new authorship. Let's say Alice adds to the article on the planet Earth the text that "The subject of this article is the third planet from the sun," a true statement. Bob copies that same exact text ("The subject of this article is the third planet from the sun") to the articles on Pluto and on mathematics: those statements have now become, respectively, false and absurd. Should Alice be held accountable as a liar or vandal for text that she originated but whose meaning Bob altered by placing it on a different article? No.
Is Bob or Alice the author of that text on Pluto or television? It is at least as much Bob as Alice; Alice may have originated that text, but by placing that text in a different context that changes its meaning, Bob has also become an author. Similarly, since these import scripts are copying text to different articles in different contexts, the import scripts are also an author, and the attribution is wrong when it only credits the originator of the text, since the text no longer has the same meaning in this different context. This is why imports need to be tagged as such for proper attribution.
Lowellian (reply) 00:51, 30 December 2017 (UTC)[reply]
Again: Where are your sources, the ones that allegedly support your claim that "Lowellian wrote this" (complete with a link to your account, so that everyone knows which Lowellian we're talking about) does not "provide credit to the author"? I see repeated assertions and a whole lot of IDONTLIKEIT, but I also see zero sources behind those assertions. Can you please provide any (single) plausibly reliable source to back up your personal opinion? WhatamIdoing (talk) 20:06, 2 January 2018 (UTC)[reply]
I have absolutely provided a source: the license, which requires credit be provided to the authors. Where are your sources, the ones that allegedly support your claim that the entity posting an edit is not an author? I see repeated assertions and a whole lot of IDONTLIKEIT, but I also see zero sources behind those assertions. Can you please provide any (single) plausibly reliable source to back up your personal opinion? (Note that, for the previous last two sentences, I just reused your words in a different context, that of my argument. In the edit history of this page, this reuse of words is attributed to me, not to you, because their context has been altered as part of this comment that I am posting, demonstrating the principle in question: the editor is an author.)
I don't think you and I are going to reach agreement on this issue, so let's just agree to disagree, since the debate is academic at this point, given that Xaosflux provided below a solution to tag imports.
Lowellian (reply) 20:56, 2 January 2018 (UTC)[reply]

Tagging imports

There's no way to reliably identify old imports to tag the edits, so that's not going to happen. The unexpected creation of accounts should be mostly done since most of the wikis are done. Anomie 13:56, 14 December 2017 (UTC)[reply]

But I'm not asking to identify old imports to tag the edits. I'm asking that future imports should tag imported edits as they are being imported. —Lowellian (reply) 20:56, 14 December 2017 (UTC)[reply]
@Lowellian: See the history of User:Xaosflux/Sandbox3z (for enwiki) and test2wiki:Male user:Xaosflux/Sandbox3z (for a remote wiki) for how NEWLY imported pages will appear in histories. Does the > identifier satisfy what you are looking for? — xaosflux Talk 22:18, 14 December 2017 (UTC)[reply]
@Xaosflux: Yes, that's great, thank you so much, that's pretty much exactly what I'm looking for! :) I don't know how involved you are in the process of these Wikipedia imports; are you in position to actually get that change done immediately for all future imports? —Lowellian (reply) 22:41, 14 December 2017 (UTC)[reply]
@Lowellian: I'm not working on the ticket, so here's the situation: any newly imported revisions will follow that already - its already live. The problem that you are seeing is that pages that were imported to projects previously have no information stored to change them in to this new style. — xaosflux Talk 23:20, 14 December 2017 (UTC)[reply]
@Xaosflux: Given the lack of stored information on previous imports, I see the difficulty there and am satisfied as long as these identifiers apply to all future imports. But I want to ask for clarification on something: these identifiers will apply to ALL future imports, including imports of edits from SUL accounts, right? Because that was the previous issue: imports of edits of non-SUL accounts were already being marked with an "imported>" identifier before the username, but edits of SUL accounts were not marked/identified in any manner. —Lowellian (reply) 23:52, 14 December 2017 (UTC)[reply]
@Lowellian: unless the software changes again in the future, then yes that is the plan for all imports going forward as far as I know, I think it is a good thing. The problem with the other edits is they USED to say your name on them, but in the database they didn't reference your userid, some of these appear to be resolvable, but many were not - in the database cleanup they are now saying "import>NAME" instead of just "NAME", because there is no way for them to know if it should be en:NAME, or es:NAME, or fr:wikisource:NAME, etc. For edits that did match a SUL account from the point of view of the destination wiki they already did match your SUL ID and there is nothing to do. — xaosflux Talk 03:56, 15 December 2017 (UTC)[reply]
@Xaosflux: Unless the person doing the import checks the "Assign edits to local users where the named user exists locally" checkbox. If they check that box, then SUL accounts will be attributed to the local name as was done by the maintenance script. As I told Lowellian in this edit, feature requests are better made in Phabricator than by continuing to demand changes here. Anomie 15:10, 16 December 2017 (UTC)[reply]
@Anomie: thanks for the clarification. I wasn't demanding anything, but will open a feature request to include an edit type identifier. — xaosflux Talk 15:57, 16 December 2017 (UTC)[reply]
See phab:T183061. — xaosflux Talk 16:10, 16 December 2017 (UTC)[reply]
The normal process of copying within Wikipedia attributes that copying, within edit history, to the editor who was actually doing the copying. These imports are copying while attributing the copying to someone who didn't do the copying, to a wiki that they didn't edit. To repeat part of what I wrote above, attribution has multiple components: who wrote something, what was written, when it was written, and where it was written. The problem is that imports without tagging misrepresent at these components: who wrote it (since it is really the import script copying over what someone else wrote rather than that person writing it directly) and where it was written. These attributions aren't accurate because they are edits made to a specific wiki that are now being attributed as edits to a different wiki. —Lowellian (reply) 01:03, 15 December 2017 (UTC)[reply]
Regardless, the way it's happening is a confusing and annoying mess. The "welcomes" in unintelligible tongues of men and angels (for all the good they do) are just a symptom.
The problem appears to be, these so-called attributions are undecipherable. If this were not the case, the scripts would not have such problems. And since attribution is required, this is serious. Andrewa (talk) 01:46, 15 December 2017 (UTC)[reply]

Auto-created local accounts though no imported contributions

Here's what I haven't yet understood about this:
So, on wikis where articles previously have been imported, local accounts are created in order to attribute the contributing users. But most or at least some of the newly created accounts that have been reported in this thread have no contributions! Are those accounts just a side effect with no big significance (other than causing annoying welcomes and uncertainty)? Or do they reflect that contributions by that user have been imported, but the details have somehow been lost and won't therefore appear in the contributions list? --Pipetricker (talk) 10:09, 15 December 2017 (UTC)[reply]

This recent comment by jhsoby at phab:T181731 seems to say it's basically a side effect, related to Wikidata. (@Anomie) --Pipetricker (talk) 10:38, 15 December 2017 (UTC)[reply]
This comment is probably more relevant.  BTW, for some reason your ping didn't work. Possibly because the edit edited another line in addition to adding a comment. Anomie 15:15, 16 December 2017 (UTC)[reply]
I don't see how that comment is relevant: I didn't mean to ask about edit counts, but about why for example ilo:Espesial:Contributions/Pipetricker and wuu:Special:用户贡献/PrimeHunter have no contributions listed, when those local accounts were created by the maintenance script. --Pipetricker (talk) 09:59, 17 December 2017 (UTC)[reply]

Information vs text

The issues raised at Wikipedia:Village pump (miscellaneous)/Archive 57#Rogue bot or what on other language wikipedias seem better discussed here. Perhaps I should in hindsight have come here first.

One of the possible problems that occurs to me is really a restatement of the initial problem raised here by Lowellian. Somehow, we've lost sight of the difference between information, which can't be copyrighted but must be sourced by references, and text, which can be copyrighted and so must be attributed.

Accurate translation preserves the information but not the text. So when going from one language wiki to another, the references should be kept, but the attributions should not be.

In English Wikipedia, other language Wikipedias are not acceptable references, so translation into English should not be a problem: The sources are kept, the attributions discarded. Other language Wikipedias may have different policies on this, in which case it may be acceptable to add a reference to English Wikipedia if translating from here. But regardless, the attributions belong only in English Wikipedia.

Am I missing something? Andrewa (talk) 05:55, 16 December 2017 (UTC)[reply]

It seems you would be interested in reading derivative work. Killiondude (talk) 06:13, 16 December 2017 (UTC)[reply]
Good point but not quite so simple. While translation of a literary work certainly preserves the creative element of the original, the translation of an encyclopedia article does not. The creative aspect of the text of an encyclopedia article is purely in the phrasing (we prohibit original research, for example), and is lost in the translation.
It's a bit of a can of worms with regard to lists, but in general, the translation of an encyclopedia article is not automatically a derivative work.
In fact the better the article is written and the better the translation is performed, the less of a problem this becomes. But it's a very good point, and probably where the paranoia for preserving attributions through translation arises.
Preserving attributions is in general a good thing, and often required, and always important when required. But not as simple as it might appear, as we have been finding out. And playing safe is not always the safest strategy! Andrewa (talk) 07:37, 16 December 2017 (UTC)[reply]
Not to impugn on any of your thoughts here, but Wikipedia is a top 10 website. Much time, effort, and legal help has been put into its licensing. Killiondude (talk) 08:15, 16 December 2017 (UTC)[reply]
Not to suggest that it's not! But perhaps it is not perfect, despite its popularity? Perhaps fixing this might help us to stay popular?
Nor to suggest that my legal opinion is superior to that of the legal department. IANAL. Lawyers are important (perhaps unfortunately, if Shakespeare is to be believed). But they should never be allowed to tell an organisation what it wants to do. Their role is to help an organisation to decide how best to do it.
This has all the earmarks of lawyers calling the shots. The main problem with the legal perspective is that they tend to want everything spelled out in terms only another lawyer would understand. And isn't that exactly what we are doing with these useless attributions? Ones that even our own coding cannot interpret correctly?
There must be a better way. And yes, the lawyers have a role in finding it. And so does the community. Andrewa (talk) 16:36, 16 December 2017 (UTC)[reply]

Global object?

"mw" is the alias for "mediaWiki", the global object.

What is a global object? The Transhumanist 12:58, 19 December 2017 (UTC)[reply]

@The Transhumanist: is this a quote from somewhere else, context may help. You may be referring to the meta:Interwiki map. — xaosflux Talk 13:50, 19 December 2017 (UTC)[reply]
@Xaosflux: Context, right. See below... The Transhumanist 18:39, 19 December 2017 (UTC)[reply]
I think you are talking about a Javascript object. Global variables are declared outside any function, and they can be accessed (used) on any function in the program. Local variables are declared inside a function, and can be used only inside that function. Same thing with global and local objects. mw:ResourceLoader/Core modules Disclaimer: I am not a Javascript expert. (((The Quixotic Potato))) (talk) 17:57, 19 December 2017 (UTC)[reply]
@The Quixotic Potato: That link is where I came across the term. At mw:ResourceLoader/Core modules#mediawiki It says:

This is the mediawiki base module. It initialises the mediaWiki global object (with mw as alias). Alias mw is available everywhere and should be used.

But I can't find the definition of "global object" anywhere. As far as I can tell, it means that "mediaWiki" and "mw" refer to the mediaWiki object (initialized by the mediawiki base module) no matter what script they are used in on Wikipedia.
Does that mean scripts on Wikipedia don't need a bodyguard function to protect the mw alias?
Is "global object" a neologism? It would be nice to be able to look up the definition. The Transhumanist 18:25, 19 December 2017 (UTC)[reply]
I'll ping @TheDJ: who knows a hell of a lot more about this stuff than I do. (((The Quixotic Potato))) (talk) 18:29, 19 December 2017 (UTC)[reply]
mediaWiki is a library using the javascript object structure. It is accessible via the variable name mediaWiki (or mw) in javascript's global scope for variables. Global variables can be written and read by everyone, so they are a bit dangerous to use, therefor we try to limit how many we use, so that we make fewer mistakes. We use bodyguards functions to 'hoist' these global variable names and make them functionally scoped variables (we import jQuery and mediaWiki, and instead of using them directly within the bodyguard, we use by renaming them to the function parameters $ and mw). This has multiple advantages. First, by lifting, we basically get to name them however we want and we can be consistent about that throughout the module (pick one and stick to it). Second, we won't be bothered by other libraries accidentally overwriting the global variable after our code in the bodyguard began executing (these can be very hard problems to debug, as they are cause by someone else's code). Third, it makes it easier to keep track of the global variables that you are using, making it harder to unintentionally use a global variable. For jshint/eslint, we define just the globals that we import with the bodyguard, allowing us to easily spot unintentional global variable usage in any other place in the code. —TheDJ (talkcontribs) 19:47, 19 December 2017 (UTC)[reply]
@TheDJ: I don't understand the answer (I was expecting a "yes" or "no" response). Do "mediaWiki" and "mw" have the power of reserved words (in scripts on Wikipedia)? Do scripts on Wikipedia need a bodyguard function to protect the mw alias? I generally include the following, but is it providing any benefit?
// Start off with a bodyguard function to reserve the aliases mw and $
( function ( mw, $ ) {
    // body of program goes here
}( mediaWiki, jQuery ) );
What am I missing?
By the way, I'm still having trouble grasping a particular term. On the page at mw:ResourceLoader/Core modules#mediawiki it says:

This is the mediawiki base module. It initialises the mediaWiki global object (with mw as alias). Alias mw is available everywhere and should be used.

In the quote above, what exactly does "mediaWiki global object" refer to? I look forward to your replies. The Transhumanist 15:31, 30 December 2017 (UTC)[reply]
I assume "global" refers to scope (computer science) (especially global scope) whereas object refers to a Javascript object. In other words, it is an object which any program can access on a specific MediaWiki page. Changing some attribute of the object will cause all other programs or scripts accessing the object to be affected. (And for that reason, I assume that the majority of attributes of the object are read-only.) --Izno (talk) 18:35, 19 December 2017 (UTC)[reply]
@Izno: Thank you. Any comments on TheDJ's Dec 19 answer above? (I don't get it). The Transhumanist 23:26, 23 December 2017 (UTC)[reply]

Simple diff

Normal diff - text and markup
Simple diff - text only, no markup

I'm looking for someone or a team who can program Mediawiki. I'm sick of squinting at diffs trying to find the text changes amongst all the markup. I want a simple diff option: a diff that shows only changes to article text and none of the markup. To make this simple diff to the right, I (1) copied just the article text from the current version and a historical version of an article (2) pasted my copy of the historical version into a new wiki page and saved it (3) replaced that text with my copy of the current version and (4) clicked "cur" beside the older version in that page's history. It's this process I'd like to be automated. Can anybody help?

I submitted this to the WMF "community tech wish list" last month. There were 214 proposals and this proposal shared 21st place with two others. The WMF community tech team has only been funded to address the ten most popular proposals.

Any volunteers? --Anthonyhcole (talk · contribs · email) 11:08, 23 December 2017 (UTC)[reply]

You're referring to meta:2017 Community Wishlist Survey/Reading/Simple diff I assume. Have you tried WikEdDiff? It is a gadget that makes reading changes much easier, with or without syntax changes. --Izno (talk) 15:10, 23 December 2017 (UTC)[reply]
I played with it a year ago but was unable to make a diff like this. Maybe I'm doing it wrong or maybe it's evolved in the meantime. Are you able to make a simple diff using it, @Izno:? --Anthonyhcole (talk · contribs · email) 13:44, 24 December 2017 (UTC)[reply]
@Anthonyhcole:Please don't use text from Wikipedia, including in screenshot form, without attribution to its author(s); linking to the original page is considered good enough, but a simple statement of "Wikipedia editors" isn't. עוד מישהו Od Mishehu 07:32, 24 December 2017 (UTC)[reply]
More at WP:WPSHOT. --Redrose64 🌹 (talk) 07:56, 24 December 2017 (UTC)[reply]
Thank you both, Od Mishehu and Redrose64; I hadn't thought about that and will do in future. --Anthonyhcole (talk · contribs · email) 13:44, 24 December 2017 (UTC)[reply]
There is already something like this in the works: visual diff. But it needs some further optimization and it needs some better support for changes in big blocks like tables and templates, before it will become available. —TheDJ (talkcontribs) 11:05, 24 December 2017 (UTC)[reply]
I'm aware of that and discussed it with James Forrester. Presently, it can't do anything remotely like what I need, and my guess is it will be at least two years before it can. I want this now. --Anthonyhcole (talk · contribs · email) 13:44, 24 December 2017 (UTC)[reply]

What if I offered to pay for this?

Does anybody here have a feel for the amount of labor this task would involve, and so the order of cost? Or do you know anyone who might be able to sensibly estimate the cost? (The task is automating steps 1 to 4 in my opening post.) --Anthonyhcole (talk · contribs · email) 13:44, 24 December 2017 (UTC)[reply]

@Anthonyhcole: I've made a userscript in less than 24 hours that basically does what you want: User:Evad37/TextDiff.js. It's a bit rough, quite slow, and probably buggy, but it seems to work on a few diffs that I tested, and refinements should be possible. - Evad37 [talk] 19:10, 24 December 2017 (UTC)[reply]
Interesting. How muuch do you want for it, Evad37, and can you show me how to install it (if that's the right word) and use it? --Anthonyhcole (talk · contribs · email) 01:12, 25 December 2017 (UTC)[reply]
@Anthonyhcole: See User:Evad37/TextDiff for installation/usage instructions. I honestly didn't do this for any sort of monetary reward (I didn't see you're offer to pay until after I had finished it), so you don't have to pay me anything. But if you still want to reward me with something, whatever you decide would be fine – email me and we'll work out the details. - Evad37 [talk] 03:53, 25 December 2017 (UTC)[reply]
Jesus Evad37, you are a machine. The WMF should have hired you years ago. — Insertcleverphrasehere (or here) 04:02, 25 December 2017 (UTC)[reply]
I love it User:Evad37 amazing work. Barnstars on it way :-) Doc James (talk · contribs · email) 06:35, 30 December 2017 (UTC)[reply]
Evad37, I just tested this too, and it is awesome. Presenting it as a 'toggle' is perfect. I think I will use this quite regularly. I add my thanks to the others. -- Begoon 06:53, 30 December 2017 (UTC)[reply]

NOINDEX

How did User talk:The359 get to be in Category:Noindexed pages? User:The359 does not know; see User talk:The359#NOINDEX. —Anomalocaris (talk) 07:16, 27 December 2017 (UTC)[reply]

John of Reading discovered the reason: I pasted the whole of this talk page into Special:ExpandTemplates to track it down. In the January 2017 section, above, there is an {{Unblock-auto reviewed}} template; that template calls {{vandal}} to display some links about the IP address; and that template includes the __NOINDEX__ keyword. And, finally, pages containing that keyword are automagically placed in the category. עוד מישהו Od Mishehu 11:09, 27 December 2017 (UTC)[reply]
עוד מישהו: Thanks for the explanation! —Anomalocaris (talk) 02:50, 30 December 2017 (UTC)[reply]

Bot removal of <center> in Template:US Census population

I have requested approval for a bot (BRFA) to remove <center> where it is present inside |footnote= of {{US Census population}} and to replace it with |align-fn=center, in the mainspace. This would cause approximately 20,000 edits, mostly to articles on US locations.

I am looking to do this as part of the WP:Lint effort as <center> is an obsolete HTML tag. From a visual point of view today, there would likely be no change for the majority of readers. From an non-visual point of view, these tags cause validation warnings. At some point in the future (either in MediaWiki or in your Internet browser), these tags will not appear as they do today (it is likely that the tag will either be silently ignored, or displayed as the literal string <center>centered thing</center>).

This first effort is a low-hanging fruit out of the 16 million obsolete tag errors. This task would not be authorized to remove <center> everywhere. (<center> is sometimes context-sensitive, especially in tables where there are superior replacement options.)

Your feedback is appreciated. --Izno (talk) 17:54, 28 December 2017 (UTC)[reply]

I want to say: Thank you for working on this huge and important project, whose goal is to fix pages now, so that some upcoming technical changes to MediaWiki parsing don't result in readers getting visibly broken pages. I appreciate and admire your dedication, and I hope that other technically minded folks will pitch in to help out. Whatamidoing (WMF) (talk) 18:36, 28 December 2017 (UTC)[reply]
I don't know how the Wikipedia template software works, so I have a question. Instead of making 20K edits, is it possible to instead have the template ignore the <center> code and not emit it back out when it generates the HTML? RudolfRed (talk) 23:43, 28 December 2017 (UTC)[reply]
The template could do this, but that makes both the template itself and the page in which the template is placed more complex (without obvious good reason besides avoiding bot edits--the purpose of the template doesn't require the additional complexity), and additionally it requires expensive processing of the input.
The direct answer aside, these are probably going to come off these pages with some other bot run in some way. There are approximately 80k pages in total of this particular tag, and the vast majority are not inside templates. --Izno (talk) 00:06, 29 December 2017 (UTC)[reply]
One other point: teaching newbies to use better syntax is easier when they do not have bad examples to follow.
I did have an aside that PrimeHunter didn't touch on but which came to mind: instead of adding the |align-fn=center, it might be easier just to remove center and then add the alignment by default in the template. --Izno (talk) 00:24, 29 December 2017 (UTC)[reply]
Seems fine to me, though I would advise a low edit rate (because people are weird about that). I also think you should just make the style for that template centered by default, instead of adding align-fn=center, if in any way possible without being too disruptive. One more thing you should check, is if there is any images or other block content, as <center> enforces both text and block alignment, whereas those are two different things in CSS. —TheDJ (talkcontribs) 09:06, 29 December 2017 (UTC)[reply]
@TheDJ: As it happens, 18.5k are using <center> and 6.5k are using align-fn to center the text (total 25k of the 30k uses), so I've just made the default text alignment centered. I'll plan to remove the center tags only now.
In the limited (anecdotal--set was not random) AWB run I've already done on IznoRepeat, I didn't see any files or other block content. Although my regex-fu isn't strong, this is a search that would capture the file/image case, all of which are false positives (I did not realize insource could capture newlines with the . class--+ too). Every other block content is either unaffected by the difference (lists and tables would have a slight rendering delta due to the block centering lost, but I don't see a big issue in the context of these tables) or invalid in wikitext (figure/video/canvas) given the default styling of browsers for the tag in question. --Izno (talk) 14:10, 29 December 2017 (UTC)[reply]

I totally support the removal of the html center tag. -- Magioladitis (talk) 18:23, 31 December 2017 (UTC)[reply]

I'm not particularly familiar with this issue, so maybe you should ignore this comment as clueless :-) Is there a way that the template could be coded so that it interprets <center as |align-fn=center? If so, what's the point of making all these edits? If you can't, I agree that a bot run should be performed. Nyttend (talk) 23:50, 31 December 2017 (UTC)[reply]

@Nyttend: See response to RudolfRed above, broadly, as well as the first paragraph in re to TheDJ. TL;DR: No, not clueless, your suggestion is possible. However, this is only the low-hanging fruit of an effort to remove these everywhere, and given the quantity, those removals will need to be botted also. --Izno (talk) 06:42, 1 January 2018 (UTC)[reply]
I'd stopped reading halfway through this discussion; I saw the response to RudolfRed but not the part about making the template centered by default. Thank you for the pointer. Nyttend (talk) 12:22, 1 January 2018 (UTC)[reply]

Throttle on the number of edits per minute from a single user?

Does the Mediawiki software have any built-in defence against unauthorised bots? Is there any way to detect high editing rates coming from a single user? Does some sort of throttle exist? If not, is it technically feasible to introduce one? – Uanfala (talk) 17:46, 29 December 2017 (UTC)[reply]

Our current configuration is 8edit/min for non-logged in and very new users. — xaosflux Talk 17:50, 29 December 2017 (UTC)[reply]
I believe there are different limits for different actions, but admins and bots are exempt from them (just as well because I made >400 edits in two minutes the other day). HJ Mitchell | Penny for your thoughts? 17:53, 29 December 2017 (UTC)[reply]
Correct, for example it is 20 Special:Email/day for normal users. — xaosflux Talk 17:55, 29 December 2017 (UTC)[reply]
Here is someone making 400+ edits in two minutes, see Wikipedia:Village_pump_(technical)#Twinkle_and_bot_edits. (((The Quixotic Potato))) (talk) 17:56, 29 December 2017 (UTC)[reply]
See Wikipedia:Village_pump_(proposals)#Twinkle's_"unlink_backlinks"_feature_and_meatbot_edits for a discussion on that. — xaosflux Talk 17:58, 29 December 2017 (UTC)[reply]
From a technical point of view, we could change the edit rate across the entire project, it would need a well discussed RFC to do it. As @HJ Mitchell: mentioned above, certain accounts (like admins and bots) are exempt. — xaosflux Talk 17:59, 29 December 2017 (UTC)[reply]
@Xaosflux: Are these limits documented anywhere? Perhaps Wikpedia:Rate limits should be turned blue? HJ Mitchell | Penny for your thoughts? 18:35, 29 December 2017 (UTC)[reply]
@HJ Mitchell: see InitialiseSettings (search page for wgRateLimits). For the tech manual on this configuration see mw:Manual:$wgRateLimits. — xaosflux Talk 18:43, 29 December 2017 (UTC)[reply]
  • So where should a proposed threshold be, in edits per minute? It needs to be high enough so that it lets in those who use AWB at a rate where they're able to review individual edits. And it should be low enough so it stops other Twinkle unlink incidents (or people running AWB too fast to see what they're doing). Are there any acceptable editing practices that will become difficult with such a limit? – Uanfala (talk) 20:00, 29 December 2017 (UTC)[reply]
  • Most proposals get at least linked to Wikipedia:Village pump (proposals), you could blue link Wikpedia:Rate limits, document the current and discuss on the talk (with discussion links in from well advertised venues) - it would be worth spending time drafting and discussing to determine (a)what is the actual current problem (b)is there an existing way to deal with this? (c)what are the edge cases?. Twinkle automation, AWB, etc are all things to consider. The throttle is in edits per seconds so it could be 100 per min, 1000 edits per hour, 10 edits per second, etc - I don't think they can be stacked (e.g. 10edits per second AND 50 edits per min). The throttle can apply to a usergroup, and different throttles can apply to different user groups. — xaosflux Talk 20:25, 29 December 2017 (UTC)[reply]
  • This seems like an overreaction. The rate limits are to protect the entire project, not against abusers of a single tool. As the tool notes: "Warning: You take full responsibility for any action you perform". The rate limits are very high for a reason, because during normal use they should NEVER be tripped. I consider the usage referred to as a 'normal' usage of the servers and as such they should not be adapted. The rate limits have so far not been documented per WP:BEANS. I currently don't see the need, it's not like something like this happens every month or so. —TheDJ (talkcontribs) 22:03, 29 December 2017 (UTC)[reply]
    From a less VPT perspective, I agree mostly with TheDJ here; it appears this conversation is driven by a behavioral edge case - and those can be controlled administratively. — xaosflux Talk 00:13, 30 December 2017 (UTC)[reply]
    Well, if there's an easy method to prevent such behavioural "edge cases" from occurring, and if this method does no collateral damage, then it should be the default choice, no? For me, the question is whether there is such collateral damage. That's why I'm wondering if there are any situations where it's acceptable for a user to be editing at bot speed without having a bot tag. – Uanfala (talk) 01:36, 30 December 2017 (UTC)[reply]
    And for the record, I don't think it's only edge cases. True, Twinkle unlink fails get to ANI about once a year. But we don't know how many get solved without the drama, and we know that Twinkle's unlink feature is not the only tool that can be misused. I occasionally witness (and quietly deal with) AWB cock-ups that would have never occurred had the perpetrator been forced to actually see what they were doing (as they were supposed to anyway, and not that they don't know it: in one of those cases it was an admin who did that). – Uanfala (talk) 01:44, 30 December 2017 (UTC)[reply]
  • Strong oppose. At the very least, admins need to be able to quickly revert large-scale vandalism, even if the vandal figures out a way to get away with doing it slowly. There are also definitely circumstances where I have needed to make rote edits more quickly than proposed, to tackle large tasks. bd2412 T 02:15, 30 December 2017 (UTC)[reply]
    • So there are situations in which it's legitimate for a non-admin to be making edits at bot speed without bot approval? Just for the avoidance of doubt, I haven't made any proposal, nor have I named anyone or tried to stop them from pursuing their favourite projects. I'm just trying to find out if there are such projects that will get disrupted. – Uanfala (talk) 02:27, 30 December 2017 (UTC)[reply]
      • Large scale disambiguation fixes following a consensus-based page move would be an appropriate time for a trusted editor—admin or not—to make edits at such a rate. I can easily do forty edits a minute that way. bd2412 T 02:43, 30 December 2017 (UTC)[reply]
        • This is precisely the most common situation in which I've witnessed AWB cock ups. But again, if an editor is capable of making forty edits per minute and sees what they're doing, I see no problems in that: any hypothetical proposed limit will anyway be around a similar number. – Uanfala (talk) 03:02, 30 December 2017 (UTC)[reply]
    • @BD2412: please note: Account creators, Bots, Bureaucrats, Stewards, and Administrators (and a few global groups) are exempt from ratelimits. Any change to that setting would not impact them. — xaosflux Talk 03:06, 30 December 2017 (UTC)[reply]
      • Still leery of throttling if applied to long-term trusted editors. If we're talking about comparatively untested accounts, that's different, but it's not clear to me that any leeway is being suggested. bd2412 T 03:09, 30 December 2017 (UTC)[reply]
        • @Bd2412: I'm not really advocating for this, but technically a "fast editors" type of configuration could be tied to an existing user group (e.g. rollbackers, extended confirmed) or yet another new group could be made to set a different limit. {E.g. (ip/newbies get 8edits/60seconds), ("fast editors" get 50edits/60seconds), (exempted are still exempted)}. — xaosflux Talk 16:51, 30 December 2017 (UTC)[reply]
Reasonable, fast but not limitless, enough to prevent major disruption. Galobtter (pingó mió) 17:18, 30 December 2017 (UTC)[reply]
Of those responsible for the situations that brought me here, more than half happen to hold various sorts of semi-advanced permissions, so I don't think tying the configuration to a particular group will have enough of a benefit. And I don't know if introducing the complication of a dedicated new user group will be commensurate with the scale of the problem that this will be trying to solve. – Uanfala (talk) 17:47, 30 December 2017 (UTC)[reply]
Ehh, 50 edits per minute is still slower, and would've meant it would've taken 8 minutes for him to do that instead of 1. Galobtter (pingó mió) 17:52, 30 December 2017 (UTC)[reply]
Well, 8 minutes is not much different from 1 minute: the community's reaction times are generally slower than that. But then that's an issue with any reasonable rate limit. As far as I understand the settings, if a user exceeds the limit then they get html error 500. This is so generic that I suspect the user is unlikely to figure out why this has happened. Is there any way to throw up a more meaningful message, something along the lines of "The wikipedia servers have detected that you're going too fast. Please have a look at the bot policy and the guidelines for automatic editing."? – Uanfala (talk) 18:08, 30 December 2017 (UTC)[reply]
@Uanfala: I expect it would have caused their process to end when it hit an error, then would have to be manually restarted afterwords - over and over again. It would not prevent them restarting, but perhaps would pause to reconsider. — xaosflux Talk 18:12, 30 December 2017 (UTC)[reply]
@Uanfala: Also, they wont get a 500 error, they will get: MediaWiki:Actionthrottledtext, this unexpected result may make client side scripts (like Twinkle) fail out though. — xaosflux Talk 23:38, 30 December 2017 (UTC)[reply]
It actually returns a 429 (too many requests) along with that message. Bots & user scripts like Twinkle could use this and provide useful feedback to the editor. FACE WITH TEARS OF JOY [u+1F602] 14:43, 31 December 2017 (UTC)[reply]
Well, this is more helpful then. I got the reference to error 500 from mw:Manual:$wgRateLimits. – Uanfala (talk) 17:23, 31 December 2017 (UTC)[reply]
On an earlier question: Reverting vandalism is not an admin-only task. Therefore, that, too, is a situation in which it's legitimate for a non-admin to be making high-speed edits.
Also, if null edits and purging are rate-limited (I don't know if they're counted the same way), then there wouldn't necessarily be anything to "see". WhatamIdoing (talk) 20:16, 2 January 2018 (UTC)[reply]

KML data for Wilts & Berks Canal article

There seems to be a problem with the kml data on the Wilts & Berks Canal article. In the "Map of route" section, there is a GeoGroup template, with options to show all co-ordinates on Google maps and on OSM. Clicking the Google maps option works ok, and shows the 20 or so points on the map. Clicking on the OSM option shows just a single point, and when you click on the description in OSM, it takes you back to the article on Wiltshire, rather than the Wilts & Berks Canal. I have tried replacing the GeoGroup template with a kml template, but that looks and works in the same fashion. I have checked the OSM option on several other canal articles, and they all work fine. It feels like the kml data that is passed to OSM is for the wrong article (ie Wiltshire), and that flushing something somewhere might fix it, but I have no idea if this is the case of not. Cheers. Bob1960evens (talk) 18:11, 30 December 2017 (UTC)[reply]

I think I may have found part of the solution. It appears that the ampersand in the title Wilts & Berks Canal confuses whatever invokes OSM, and it picks up the data from Wilts_, rather than Wilts_&_Berks_Canal. I don't know if anything can be done about that, or whether this should now be reported as a bug, and if so, where? Bob1960evens (talk) 18:16, 30 December 2017 (UTC)[reply]
Yes, it looks like some weird urlencoding thing - the template tries to encode the & as %26, then osm4wiki on the toolserver encodes that again into %2526, and it then seems to treat this as the end of the page title. Seems to be a bug in osm4wiki and I'm not sure how to report that.
There is a workaround, though! If you tell the template to look at Wilts and Berks Canal, which is a redirect to the ampersand title, it follows that redirect okay and picks up all the points - it displays a little bit clunkily but it's definitely functional. {{GeoGroup|article=Wilts and Berks Canal}}. I've made the change on the page. Andrew Gray (talk) 00:05, 31 December 2017 (UTC)[reply]
There are maintainers listed at https://tools.wmflabs.org/admin/tool/osm4wiki. PrimeHunter (talk) 00:18, 31 December 2017 (UTC)[reply]
Thanks for your help. Bob1960evens (talk) 09:57, 1 January 2018 (UTC)[reply]

How did this happen?

A few minutes ago, I decided to take a quick glance at the New Pages Feed. When filtered to articles created "by newcomers," three appeared on the list: Qasymbek Arabin, Allan Katerega, and Marcus Van Wattum. The first article was created on December 28 by an IP. This case is especially remarkable; I thought IPs had been forbidden from creating article for over a decade, ever since the Seigenthaler incident. The latter two articles were created by non-autoconfirmed users on December 5 and 27, respectively. Both dates are after the implementation of the ACTRIAL.

Is there some kind of loophole here? Biblio (talk) 02:02, 31 December 2017 (UTC)[reply]

Drafts moved to mainspace... The 3rd one by a new user, as soon as they became autoconfirmed, maybe a COI, from username - but nevertheless, moved by autoconfirmed user. -- Begoon 02:20, 31 December 2017 (UTC)[reply]
I see. Evidently, some AFC reviewers do not mark pages as patrolled when they move them. Biblio (talk) 02:32, 31 December 2017 (UTC)[reply]

File deletion - small bug

I just deleted file:Sui_Dhaaga_2018.jpg and got the message back:

"Sui Dhaaga 2018.jpg" has been deleted undelete | salt.

Note that in the undelete and salt links there is a "File:" missing. Please fix. — RHaworth (talk · contribs) 13:12, 1 January 2018 (UTC)[reply]

I just undeleted and redeleted it, and the message is identical. But it's not just this one; I got the same links with File:Nyttend's bad link alignment.png. Commons is different; I just deleted c:File:Kukoamines 2D Structure.png and got links to c:Special:Undelete/File:Kukoamines 2D Structure.png and c:Special:WhatLinksHere/File:Kukoamines 2D Structure.png. Nyttend (talk) 13:37, 1 January 2018 (UTC)[reply]
It must have been caused by this this edit to MediaWiki:Filedelete-success. I have made a fix.[4] It didn't work a minute after my edit but there is sometimes a delay before edits to the MediaWiki namespace take effect. PrimeHunter (talk) 14:26, 1 January 2018 (UTC)[reply]

Lang template

The {{lang}} template (i.e. not the lang-xx ones such as {{lang-de}}) now seems to automatically make the target text italicised. It didn't use to, but now for example {{user de-1}}, which uses the template, is all in italics, this has happened in the last few days I think, though the last edit to Template:lang was made weeks ago. Does anyone know what happened and why? – filelakeshoe (t / c) 00:52, 2 January 2018 (UTC)[reply]

The template was recently rewritten to use Lua, so most of the action now happens at Module:Lang. I haven't been following the recent developments but the italics question figures quite prominently in several thread on the template's talk page. – Uanfala (talk) 01:07, 2 January 2018 (UTC)[reply]
It looks like adding |italic=no to {{lang}} in the user box would remove the italics. Johnuniq (talk) 02:19, 2 January 2018 (UTC)[reply]
Seems to me that this text should be italicized on en.wp. --Izno (talk) 14:02, 2 January 2018 (UTC)[reply]

Template:Timecell

Template:Timecell doesn't show the correct date in the Chinese calendar, and purging doesn't help. The problem might be caused by Template:Ctime:d, which is supposed to output the day of the month, but currently outputs 46. I am not familiar with the template syntax. Can someone fix the problem? Gulumeemee (talk) 03:46, 2 January 2018 (UTC)[reply]

This seems to be part of a whole set of templates written in 2015 by Orienomesh-w (talk · contribs), who hasn't been active for six months:
Documentation is minimal, and there are no discussion pages. In most cases, the only edits not by Orienomesh-w were by Cydebot (talk · contribs), which was renaming a category. In one case, there was also one inconsequential edit by John of Reading (talk · contribs). But in the case of Template:Ctime:N, there have been five edits by three different IPs which may or may not have been detrimental.
What I suspect is that the template can't handle the period between the two New Years Days. I can't confirm this. --Redrose64 🌹 (talk) 11:36, 2 January 2018 (UTC)[reply]
The template can't handle some dates before the Gregorian calendar's New Year. {{Ctime:f|2017-12-30}} and {{Ctime:d|2017-12-30}} produce 07 and 131. {{Ctime:d|2017-10-30}} outputs 70. Gulumeemee (talk) 09:44, 4 January 2018 (UTC)[reply]

Search across a category and a WikiProject

Hi. How would I be able to search in the category Category:No local image but image on Wikidata for biographies that are tagged for the Olympic Project? Is that possible? Thanks in advance. Lugnuts Fire Walk with Me 14:33, 2 January 2018 (UTC)[reply]

@Lugnuts: Fairly certain WP:Petscan can do that for you. --Izno (talk) 14:38, 2 January 2018 (UTC)[reply]
Of course! I think I cracked it with this. Thanks Izno. Lugnuts Fire Walk with Me 14:50, 2 January 2018 (UTC)[reply]

Column or table formatting problems

In looking for an example of an unrelated bug, a couple of us ran across some pages with formatting problems. Here are two: Hazelwood, Missouri#Climate and Broward County Public Schools#Elementary Schools. Could someone figure out whether this is just an accident, or if there are problems with some of the formatting templates? Whatamidoing (WMF) (talk) 20:25, 2 January 2018 (UTC)[reply]

The public schools page at that section is missing an opening template (i.e. a "col-start" or similar), which usually indicates mistake or vandalism. The same for the former. --Izno (talk) 20:50, 2 January 2018 (UTC)[reply]
In the case of Broward County Public Schools, it was caused by this edit from September 2017, which was probably well-intentioned. In the case of Hazelwood, Missouri, it was caused by this good-faith edit in April 2014, which was ultimately caused by a botched attempt to move the text around in April 2011. I've fixed both the articles. Graham87 06:10, 3 January 2018 (UTC)[reply]

Blocking defaults

Is it possible for me to specify a default expiration time and reason on the Special:Block page? --NeilN talk to me 20:45, 2 January 2018 (UTC)[reply]

Not in the UI that I know of, but I think something equivalent can be done with Javascript. What time/reason did you have in mind? Writ Keeper  20:47, 2 January 2018 (UTC)[reply]
24 hours and [[WP:Blocking policy#Evasion of blocks|Block evasion]]. I normally use this interface when blocking LTAs and don't want to leave a talk page message. --NeilN talk to me 20:53, 2 January 2018 (UTC)[reply]
@NeilN: Give User:Writ Keeper/Scripts/simpleBlockDefaults.js a shot (installed in the usual fashion to the .js page of your choice). Might take a second to load, but it should do the trick, I think. Writ Keeper  21:40, 2 January 2018 (UTC)[reply]
@Writ Keeper: Works like a charm! Thank you! --NeilN talk to me 21:47, 2 January 2018 (UTC)[reply]

There are some copyright warnings like MediaWiki:Editpage-head-copy-warn and MediaWiki:Wikimedia-copyrightwarning making users scroll down more than necessary. After more than 10 years of editing Wikipedia I would like to disable them. Is it possible to find a technical solution so warnings could be canceled under some conditions, maybe in preferences? --Janezdrilc (talk) 22:25, 2 January 2018 (UTC)[reply]

A single scroll doesn't really seem like much since that statement is a legally binding release of your work under a different copyright license than would normally be present. In any case, those warnings are set by WMF legal. Even if there was a gadget for disabling them I don't think, for legal reasons, that it would be made readily available. --Majora (talk) 22:32, 2 January 2018 (UTC)[reply]
I can just imagine someone saying that the own the copyright as no warning was present. Emir of Wikipedia (talk) 22:35, 2 January 2018 (UTC)[reply]
You can hide the messages with this in your CSS:
.editpage-head-copywarn {display: none;}
#editpage-copywarn {display: none;}
PrimeHunter (talk) 22:47, 2 January 2018 (UTC)[reply]
Thanks, PrimeHunter. At least you understand me. --Janezdrilc (talk) 23:15, 2 January 2018 (UTC)[reply]

Section citation formatting bug(?)

When I added a citation in a section here: List of Foreign Archaeological Institutes in Greece, it was formatted as part of the last section of the page, although in the preview it appeared correctly; is this a known bug? 195.251.104.50 (talk) 07:20, 3 January 2018 (UTC)[reply]

It's intended behaviour and not a bug. If there is no instruction like {{Reflist}} for where to display references then they are displayed at the bottom of the page or preview. See Help:Referencing for beginners. An editor has added a References section with {{Reflist}}.[5] Wikipedia does not display references at the end of sections. PrimeHunter (talk) 11:07, 3 January 2018 (UTC)[reply]

Username disabling a move

While doing NPP I noticed that PEARL IDEA had made a blank article, so I decided that I would move it to their user space. Not allowed, apparently. The blacklist disallows moves to titles that have nine or more capital letters. This is obviously some regex that needs refining, to stop innocent users from doing their work. BTW the article ended up at Draft:Tera_Baap_Mera_Baap. !dave 15:25, 3 January 2018 (UTC)[reply]

More than 9 capitals with no lowercase letters between them is disallowed. User:PEARL IDEA/Tera Baap Mera Baap is disallowed because 'T' counts as the 10th capital. You can make a suggestion at MediaWiki talk:Titleblacklist but I think it's a rare issue and the blacklist would be unable to test whether the user exists. Administrators and Template editors can make the move. See Wikipedia:Requested moves#Requesting technical moves. PrimeHunter (talk) 16:32, 3 January 2018 (UTC)[reply]
You could ask for autoconfirmed to be added to the conditions for the entry, if we want to allow autoconfirmed users to make such a move. Jo-Jo Eumerus (talk, contributions) 16:39, 3 January 2018 (UTC)[reply]
Eh. I guess the rationale for blacklisting is that WE DON'T WANT PEOPLE SCREAMING IN ARTICLE TITLES, but that is not really a problem if the screaming is part of the user name. (Well, it could be a problem, but that is not a problem to address via title blacklisting). (@PrimeHunter: I do not think we really care about testing whether the username exists. That is a test to avoid screaming, not a test for an allowed subpage.)
The regexp could probably be tweaked to ignore the username part in the screaming test. If I assume the current regexp is something that looks like (pattern){10,} (where the pattern matches for "capitals without lowercase"), and it does not already call ^ to match the start of the string, we can simply use (^User:[^\/]*\/|^)((pattern){10,}). I tested on https://www.regexpal.com/ for basic cases, this seems to do the job of "ignore the User:Foo/ part if it exists". TigraanClick here to contact me 17:23, 3 January 2018 (UTC)[reply]
Note: I just suggested that change on the aforementioned page. TigraanClick here to contact me 17:29, 3 January 2018 (UTC)[reply]

Is it possible....

...to create a bot, or "auto-generated reminder" as follows: an admin blocks an editor, and...

  1. in their block summary for the user log, the admin adds a date command that on a specific date, triggers a bot reminder...
  2. the reminder simply says "Review block log of (bot inserts respective user's name);
  3. the bot posts the reminder to AN, and on the TP of the blocking admin and blocked user.

Please ping me when responding. Thank you! Atsme📞📧 11:53, 4 January 2018 (UTC)[reply]

Yes, of course. (((The Quixotic Potato))) (talk) 12:50, 4 January 2018 (UTC)[reply]
@Atsme: Forgot to ping you. (((The Quixotic Potato))) (talk) 12:51, 4 January 2018 (UTC)[reply]
Thank you, TQP - is this something that can be done locally without having to get the WMF involved? In other words, if community consensus approves a policy change/modification that will allow admins to add such a reminder to the block log, is it simple enough a project that it can be implemented (via java script, perhaps?) without having to jump through the WMF's hoops of fire to get a programmer to write the script? Atsme📞📧 13:41, 4 January 2018 (UTC)[reply]
This COULD be done with a bot since it could be done by an editor (that is read block logs, store data locally, read data locally, make edits). To be done with a bot it would not require any software changes - it would require someone to create and run such a bot. To actually be useful you'd have to convince admins to actually use these custom block log triggers and where the edits should go. — xaosflux Talk 14:22, 4 January 2018 (UTC)[reply]
  1. Download daily block log.
  2. Search each entry for trigger (e.g. "remindme 17/12/2019") and store it in a little database. Check if the date is in the correct format (dd/mm/yyyy v.s. mm/dd/yyyy)
  3. Check if there are remindme instructions for today in the database and send out the appropriate notifications if that is the case.
This isn't very difficult to program. Ideally the person who writes this software already has a bot or bots that run every day (I wouldn't run it on this computer because it is not always turned on). That person would need a botflag to post outside of their own userspace. The WMF does not have to be involved.
(((The Quixotic Potato))) (talk) 14:32, 4 January 2018 (UTC)[reply]
Note that HTML comments do not appear in the message displayed to the blocked user; ideally, you could use that to append metadata to your blocks. This is already routinely done for Sockpuppets. -- Luk talk 13:51, 5 January 2018 (UTC)[reply]

Peculiar title case change

Resolved

So I was trying to link to the article List of French generals of the Revolutionary and Napoleonic Wars (see [6]). So I copied the article title as displayed "List of French Generals of the Revolutionary and Napoleonic Wars" -- but notice that "Generals" is capitalized -- and this resulted in a red link (at least at the time). Looking through the article, I don't see anything that would make the title display differently, unless it is buried within some other non-obvious template. Can anyone else spot what is going on here? olderwiser 13:28, 4 January 2018 (UTC)[reply]

Nevermind. I think I see what was happening -- my Chrome browser "helpfully" autotranslated and capitalized General. olderwiser 13:32, 4 January 2018 (UTC)[reply]

How to edit descriptions of articles that appear on mobile devices

I apologize if this is documented somewhere. Is it possible to edit the descriptions of articles that appear on mobile devices at the bottom of the page? I've seen some that are confusing and could be improved. Is it on wikidata? Is it somewhere else on the page? Or is it not part of Wikipedia at all? Thank you. Lollipop (talk) 16:44, 4 January 2018 (UTC)[reply]

Lollipop It's on wikidata. The wikidata item for VPT is - this and has a description of "Wikimedia technical village pump" Galobtter (pingó mió) 16:53, 4 January 2018 (UTC)[reply]
ok thank you..... I will make use of that by editing on the big screen . Lollipop (talk) 17:18, 4 January 2018 (UTC)[reply]
See also Wikipedia:FAQ/Editing#How do I edit mobile subtitles? PrimeHunter (talk) 19:49, 4 January 2018 (UTC)[reply]
Lollipop you may also be interested in WP:Village_pump_(proposals)#RfC:_Populating_article_descriptions_magic_word. Alsee (talk) 20:24, 8 January 2018 (UTC)[reply]

User contributions counter seems to be off

An extremely minor issue, but one I just happened to notice. If you look at my contribs, click "older 50", and then "newer 50", the two most recent edits are missing. You have to click "newer 50" again, and it then shows only those two edits. I'm writing this from my alt account in order to preserve this. Away Lander (talk) 16:51, 4 January 2018 (UTC)[reply]

Huh, that is pretty strange. Shows same for me. Galobtter (pingó mió) 16:55, 4 January 2018 (UTC)[reply]
@Away Lander and Galobtter: This is probably because you each made multiple edits which were logged at the same time (this should happen with page moves, since all two or four edits are from the same action). Note that the first two edits' revision IDs here are actually out of order (hover over the links). This makes the newer/older buttons malfunction, since the links are generated based on the edit timestamps. Jc86035 (talk) 17:16, 4 January 2018 (UTC)[reply]
Ooooh. I meant when I saw his contributions, I think my contributions are working fine at-least for the first two, but interesting. Galobtter (pingó mió) 17:22, 4 January 2018 (UTC)[reply]
@Jc86035: Interesting. The edits right around the 50 mark were a mass-revert of a sockpuppet account so I'm guessing some of them are "out of order" as you describe above. Away Lander (talk) 19:00, 4 January 2018 (UTC)[reply]
Doesn't happen with limit=100, so I added ?limit=50 to your link to make it work for people with other settings. --Pipetricker (talk) 17:19, 4 January 2018 (UTC)[reply]
@Pipetricker: Thanks, didn't even think of that. Away Lander (talk) 19:00, 4 January 2018 (UTC)[reply]

Yes, something is definitely weird with the edits right around the mass-rollback; as I'm editing more pages, the amount of edits being "missed" and showing up only when you click "older 50" twice, is growing. Home Lander (talk) 19:19, 4 January 2018 (UTC)[reply]

Education Program namespace

Why does Special:PrefixIndex find no pages in the Education Program namespace? See Special:PrefixIndex/Education_Program:. The same happens if you try to find a random page in that namespace (Special:Random/Education_Program). Yet, perplexingly, there are plenty of pages in the Education Program talk namespace, and accompanying course pages; see Special:PrefixIndex/Education_Program_talk:. Thanks, -- Black Falcon (talk) 20:29, 4 January 2018 (UTC)[reply]

@Black Falcon: see phab:T139901, also note this namespace is being planned to be removed in phab:T169676 (eventually). — xaosflux Talk 20:42, 4 January 2018 (UTC)[reply]
Thanks, that's very helpful to know. I'll track the tasks in Phabricator. -- Black Falcon (talk) 20:47, 4 January 2018 (UTC)[reply]

Spurious or invisible notification

Oddly, a red 1 was shown in the notifications/alerts at the top (as usual for new notifications). When checking the notifications, nothing new appears to be present (and old notifications have no X to clear them or their group anymore), but the 1 is now gray and persists. —PaleoNeonate23:19, 4 January 2018 (UTC)[reply]

Adding: "Show notifications from other Wikis" is also on in my preferences (and I would expect to see an event telling me from which Wiki, if that was an inter-Wiki one). —PaleoNeonate23:24, 4 January 2018 (UTC)[reply]

If you go to Special:Notifications is there anything pending? — xaosflux Talk 00:22, 5 January 2018 (UTC)[reply]
That occurs if you have scripting disabled. Johnuniq (talk) 00:46, 5 January 2018 (UTC)[reply]
Special:Notifications shows all previous notifications but not this new one. I remember notifications not showing up anymore with noscript, although with scripts completely disabled they normally do. Thanks, —PaleoNeonate02:29, 5 January 2018 (UTC)[reply]
Could someone please ping me to see if I still can see normal notifications or if they generally broke about two days ago? Thanks, —PaleoNeonate06:50, 5 January 2018 (UTC)[reply]
@PaleoNeonate:. Later I'll try that from another site. If scripting is enabled, you will see the site and will be able to remove the notification. Johnuniq (talk) 07:02, 5 January 2018 (UTC)[reply]
This ping notification worked normally, thanks (and got flagged as seen successfully). I now have yet another pending one that I cannot see, however (so two gray ones after visiting the notifications list). So it might be interwiki notifications that recently broke with scripts disabled... —PaleoNeonate07:38, 5 January 2018 (UTC)[reply]
That's what I'm saying although the change happened months ago. If you really don't want to temporarily enable scripting to dismiss the notifications, you can dismiss the ping I sent you from Commons by visiting any page there, for example c:User talk:PaleoNeonate. Johnuniq (talk) 09:16, 5 January 2018 (UTC)[reply]
Thanks, I confirm that I can also receive revert notifications for en-Wiki. I now know that one notification originates from Commons (so visited it), but have no way to know what other Wiki I should visit to clear the other one, unfortunately (I am not actively editing any of the others). —PaleoNeonate08:19, 6 January 2018 (UTC)[reply]
  • I see a different change that just happened in today. When clicking on any notification, you get the drop down list, but every message is teeny-tiny font. Nothing else seems to have changed, and I see nothing under my Notifications preferences that deals with this. What happened? Also, zooming the browser in or out will enlarge or shrink the page visuals, but has no effect whatsoever on the Notifications drop down visuals. I updated to Firefox 57.0.4 earlier. Is it another one of those funky Firefox issues? — Maile (talk) 20:13, 5 January 2018 (UTC)[reply]
I guess you use the Modern skin at Special:Preferences#mw-prefsection-rendering. I have created phab:T184295: "Notifications have small font in Modern skin". PrimeHunter (talk) 20:57, 5 January 2018 (UTC)[reply]
PrimeHunter Ah ha ... you are correct in that it's limited to Modern skin. I tested the other skins, and the notification drop down list is normal on all the rest. Thanks for the info on T184295. — Maile (talk) 21:06, 5 January 2018 (UTC)[reply]

Categories in AFC script

Whenever I try to accept an AfC draft such as Draft:Murder by the Lake, when I type anything in the "Add categories" box, instead of displaying categories, it just says "No results match Foo". GeoffreyT2000 (talk) 15:57, 5 January 2018 (UTC)[reply]

I guess you refer to "Yet Another AFC Helper Script" at Special:Preferences#mw-prefsection-gadgets. You can post to Wikipedia talk:WikiProject Articles for creation/Helper script or the GitHub link there. PrimeHunter (talk) 16:14, 5 January 2018 (UTC)[reply]

Moving to a page that already exists

If you try to move a page to a name that already exists, there is an error page, but it does not contain a link to a target page. It is sometimes a problem: if I would like to move Russian Empire to Russian empire, then I can not just enter a target name Russian empire into search field as it will be automatically corrected to a name of the existing page Russian Empire, so I should do more actions to open this page. Can we just add a link to target there? Wikisaurus (talk) 23:04, 5 January 2018 (UTC)[reply]

Wikisaurus The reason you can't move the page, is that Fish and karate protected it on November 17, 2017, requiring Admin access to move it. The edit summary says "(Edit warring / content dispute: Stop edit warring over this. Discuss it on the talk page. The next editor to change (or rollback, or undo) the flag being used without discussing it gets blocked)". — Maile (talk) 23:19, 5 January 2018 (UTC)[reply]
Maile66, it was an example. Wikisaurus (talk) 23:20, 5 January 2018 (UTC)[reply]
If you are actually able to move the source page but not overwrite the target name then the message appears to display MediaWiki:Permissionserrorstext-withaction and MediaWiki:Articleexists. Neither has access to the target name as a parameter based on translatewiki:MediaWiki:Permissionserrorstext-withaction/qqq and translatewiki:MediaWiki:Articleexists/qqq. That means we would probably have to request a change to MediaWiki itself at phab: to pass the name as $1 to MediaWiki:Articleexists. PrimeHunter (talk) 23:26, 5 January 2018 (UTC)[reply]
Can you provide a different example? bd2412 T 23:27, 5 January 2018 (UTC)[reply]
Just pick any two of your subpages: Special:PrefixIndex/User:BD2412/. But the message is only shown to autoconfirmed non-admin accounts. Without autoconfirmation you cannot move pages at all, and admin accounts can move over existing pages. PrimeHunter (talk) 03:23, 6 January 2018 (UTC)[reply]
PrimeHunter, thank you! I asked on Phabricator: phab:T184357. Wikisaurus (talk) 22:46, 6 January 2018 (UTC)[reply]

Is there a way that I can get a current count of articles made?

Wikipedia:List of Wikipedians by article count has been dead since last fall. How can I find out my current count? bd2412 T 23:31, 5 January 2018 (UTC)[reply]

See the "articles created" link at the bottom of Special:Contributions/BD2412. Johnuniq (talk) 00:01, 6 January 2018 (UTC)[reply]
Not for me, unfortunately - "User has made too many edits! (Maximum 350,000)". bd2412 T 00:03, 6 January 2018 (UTC)[reply]
Are you trying to include all your "articles" that are just massive numbers of redirects (why isn't that being done by a bot if it is important anyway?)? — xaosflux Talk 00:25, 6 January 2018 (UTC)[reply]
Also, this may be able to be done with quarry, ask for what you want at mw:Talk:Quarry and someone will probably write it for you, then you can run whenever you need. — xaosflux Talk 00:34, 6 January 2018 (UTC)[reply]
I am most interested in my real article count, which I would estimate is somewhere in the range of ~3,120 by now. As for the redirects, tens of thousands of them result from closing RM discussions (I was the most active admin there for a number of years, and still dabble a bit), which, obviously, a bot can't do. I have asked a few times for a bot to do sortname redirects, which are probably my next largest batch, but apparently those are not a priority. bd2412 T 00:44, 6 January 2018 (UTC)[reply]
Gotcha, go the database query route, once it is built you can run it anytime. — xaosflux Talk 01:10, 6 January 2018 (UTC)[reply]
Local page Wikipedia:Request a query. — xaosflux Talk 01:16, 6 January 2018 (UTC)[reply]
I was told to use this:
select count(*)
  from revision
 where rev_user = <<your-user-id>>
   and rev_parent_id = 0
Great. Now where exactly do I use it? bd2412 T 19:07, 6 January 2018 (UTC)[reply]
{{doing}} standby. — xaosflux Talk 19:09, 6 January 2018 (UTC)[reply]
@BD2412: that SQL above looks incomplete. First, it will find all 'pages' you created, not just 'articles', 2 (assuming you want it) it will include redirects. Also FYI, your userid is 196446. — xaosflux Talk 19:19, 6 January 2018 (UTC)[reply]
I'm trying a tweak on it. — xaosflux Talk 19:22, 6 January 2018 (UTC)[reply]
@BD2412:, this one might work, it is currently running - check on it in a bit to see if it completes or if it is too big and times out: quarry:query/23937. — xaosflux Talk 19:32, 6 January 2018 (UTC)[reply]
"Query status: killed" - woe is me, I have edited too much. How about just counting from the date of the last update to Wikipedia:List of Wikipedians by article count (if that can be determined)? bd2412 T 19:56, 6 January 2018 (UTC)[reply]
OK, so can't do it using quarry (and I don't think adding in an additional condition will fix it), someone with an offline database may be able to. @Certes:, maybe? — xaosflux Talk 20:04, 6 January 2018 (UTC)[reply]
Does Xtools provide what you need? If not then I don't have anything offline that would help. (I only keep a list of article titles to search.) Certes (talk) 21:12, 6 January 2018 (UTC)[reply]
@BD2412: does 3688 created articles seem right? — xaosflux Talk 23:28, 6 January 2018 (UTC)[reply]
It sounds high by a few hundred or so, but not impossible. I'll take it, though! bd2412 T 23:35, 6 January 2018 (UTC)[reply]

A request on a generic infobox

Dear all, in Wiki Project Ancient Egypt, we have a problem with the pharaoh infobox (visible on hundreds of articles). We noticed that the [show] button in the infobox, next to "Royal Titulary" is rarely if ever noticed by casual readers (see e.g. Nyuserre Ini). This means many cannot see the pharaoh's five names in hieroglyphs with translations. This has all sorts of negative consequences, from complains about the absence of this information to well-meaning people editing the hieroglyphs in the article. Thus, we contemplate the idea of making the [show] button more conspicuous by replacing it by [click to show] or put it in bold, or both. Unfortunately, Template:Infobox_pharaoh does not allow us to edit the appearance of the [show] button which seems to be defined at a higher level, perhaps in templates called by the infobox template. Could someone help us by showing us how to edit the appearance of the show button for pharaoh infoboxes ?Iry-Hor (talk) 17:57, 6 January 2018 (UTC)[reply]

Someone here should hopefully be able to tell if/how the appearance of the [show] button can be changed. But I'm wondering whether, in addition, the caption of the collapsed part of the infobox could be changed to something that will more immediately signal the presence of hieroglyphs. Say, if the "Royal titulary" bit was followed, in brackets, by the hieroglyphic rendition of the concept. – Uanfala (talk) 18:09, 6 January 2018 (UTC)[reply]
Uanfala This would be difficult because it would be hard to pinpoint hieroglyphs for this concept without entering endless debates in Egyptology and second, the hieroglyphs would distord the infobox and I believe make it quite ugly. However I agree that we could say something like [click for hieroglyphs] instead of just [show].Iry-Hor (talk) 08:44, 7 January 2018 (UTC)[reply]
I think the problem is not really specifically with the look of the [show] button, but that other than that, the "Royal titulary" line appears just like the {{{name}}} and "Pharaoh" headings above it, as a heading on a colored background. A heading representing folded content should have a more obvious graphic indication of this, other than the [show] button. Perhaps put [show] on a line by itself below the heading, with a lighter background color. --Pipetricker (talk) 18:30, 6 January 2018 (UTC)[reply]
Yeah, it's not immediately obvious that it can be expanded - just looks like a heading for the remaining stuff in the infobox. Galobtter (pingó mió) 18:37, 6 January 2018 (UTC)[reply]
Pipetricker Galobtter Uanfala Thanks for your ideas, as you pointed out the problem is that it looks like "Royal Titulary" is the heading of the section below when it is really the heading of the name section which is hidden until the button [show] is clicked. We could also add another heading under it for the next section, I don't know what's best really and on top of that I do not know how to change the show button.Iry-Hor (talk) 08:42, 7 January 2018 (UTC)[reply]
This demo shows the basic minimum. It illustrates that the text "show" is set by the javascript that is associated with the NavFrame class. Therefore, you can't configure it by modifying an infobox template or any of its subtemplates. --Redrose64 🌹 (talk) 14:15, 7 January 2018 (UTC)[reply]
Redrose64 ok so that means the only way to make the show button more conspicuous is that there is no way?Iry-Hor (talk) 14:33, 7 January 2018 (UTC)[reply]
mw:Manual:Collapsible elements#With custom toggle link implies that the toggle link may be customised, and even supplies three HTML elements that it claims will carry out such customisation. But it doesn't say where these should be placed, nor does it provide a working demo. I also can't find where in mw:MediaWiki:Gadget-NavFrame.js this code would hook to. --Redrose64 🌹 (talk) 15:29, 7 January 2018 (UTC)[reply]
Redrose64 Thank you, I will try to find out how this works.Iry-Hor (talk) 18:41, 7 January 2018 (UTC)[reply]
You might replace this:
| headerstyle = background:#decd87;padding:0.1em; {{#if:{{{titulary_notes|{{{notes|}}}}}} | |display:block;margin-bottom:0.3em;}}
| header = {{bigger|[[Ancient Egyptian royal titulary|Royal titulary]]}}<!--(resized here rather than via headerstyle otherwise [show/hide] link also resized)-->
with this:
| headerstyle = font-size:150%;background:#decd87;padding:0.1em; {{#if:{{{titulary_notes|{{{notes|}}}}}} | |display:block;margin-bottom:0.3em;}}<!--(resize [show/hide] link)-->
| header = <span style="font-size:75%;">[[Ancient Egyptian royal titulary|Royal titulary]]</span><!--(now resized header smaller to emphasize [show/hide] link)-->
Sandbox version at right uses |headerstyle= to set the whole header to 150% of normal size then applies a size reduction to the header title leaving the [show/hide] at the larger size. Play around with the values till you find something that you like.
Trappist the monk (talk) 15:49, 7 January 2018 (UTC)[reply]
Trappist the monk Thank you for your precious help! I will implement something like this when the discussion on the subject is closed (see talk page of the infobox). Also is there a way for the button to say something else than [show]?Iry-Hor (talk) 18:40, 7 January 2018 (UTC)[reply]
I don't know how to change the [show/hide] label. That appears to be hard-coded at the top of mw:MediaWiki:Gadget-NavFrame.js.
Another think that you might consider is adding |showhide=left which will mode the [show/hide] label to the left side of the header. Because this is en/wiki and English is read left-to-right, doing that make make the label more obvious to readers without the necessity of size changes.
Trappist the monk (talk) 19:19, 7 January 2018 (UTC)[reply]
@Iry-Hor and Trappist the monk: You can do this by using the newer mw-collapsible and mw-collapsible-content classes instead of NavFrames. {{Hidden/sandbox}} has the newer classes but there are a few things that need to be done (like centring the header text with a div with margin on left and right, since the show/hide button is different, and maybe checking that the slight change in display doesn't break anything?) before the code in {{Hidden begin/sandbox}} is moved to the main {{Hidden begin}} template (which {{Hidden}} uses). The left-side button is also currently a local bit of CSS which isn't in MediaWiki core like the standard mw-collapsible classes. Jc86035 (talk) 05:55, 9 January 2018 (UTC)[reply]
Jc86035 Will this allow me to replace [show] by [click to show] and put it on the left ?Iry-Hor (talk) 06:36, 9 January 2018 (UTC)[reply]

Total edits per month

Is there a way to get a list of total edits in English Wikipedia per month for 2017? -- Magioladitis (talk) 19:31, 6 January 2018 (UTC)[reply]

https://en.wikipedia.org/wiki/Special:ApiSandbox#action=query&format=xml&list=allrevisions I think you'll have to split it up by month yourself. (((The Quixotic Potato))) (talk) 20:11, 6 January 2018 (UTC)[reply]
The Quixotic Potato Thanks! -- Magioladitis (talk) 21:11, 6 January 2018 (UTC)[reply]

The Quixotic Potato how about the total number of edits per month? -- Magioladitis (talk) 21:15, 6 January 2018 (UTC)[reply]

That one is more difficult, I'll have a look. (((The Quixotic Potato))) (talk) 21:20, 6 January 2018 (UTC)[reply]
mw:Talk:Quarry Wikipedia:Request a query @Xaosflux: Would a Quarry query be a good solution? (((The Quixotic Potato))) (talk) 21:25, 6 January 2018 (UTC)[reply]
Or a querry quary. Your choice. --Redrose64 🌹 (talk) 21:28, 6 January 2018 (UTC)[reply]
See the WMF's stats page for English Wikipedia. isaacl (talk) 22:11, 6 January 2018 (UTC)[reply]
@Magioladitis: See also [7] (((The Quixotic Potato))) (talk) 22:16, 6 January 2018 (UTC)[reply]
https://stats.wikimedia.org/v2/#/en.wikipedia.org/contributing/edits (you can change the linechart to a table).
Month Edits
2017-01-01 5118176
2017-02-01 4744859
2017-03-01 5049862
2017-04-01 4687982
2017-05-01 4992655
2017-06-01 4982182
2017-07-01 4790873
2017-08-01 4818284
2017-09-01 4789522
2017-10-01 4838996
2017-11-01 4772932
2017-12-01 4971554

(((The Quixotic Potato))) (talk) 23:36, 6 January 2018 (UTC)[reply]

The Quixotic Potato Wow. Thanks, Magioladitis (talk) 00:39, 7 January 2018 (UTC)[reply]

Note that the message at the top of the wikistats page says that the first task for the new interface is to make the numbers match those produced by the old interface. I'm not sure why, but the numbers produced by WikiStats2 are considerably higher than that of the old page. isaacl (talk) 20:25, 7 January 2018 (UTC)[reply]

Add Reading List to the Desktop

Hi All Apologies if this has already been raised as an issue (I couldn't see anything on here but didn't exactly do a deep dive to search). I'm very non-technical so have no idea of the feasibility of this suggestion. I love the reading list function on the app and was wondering if anyone knew how to add this to the desktop version so that it syncs between your devices?

Many thanks — Preceding unsigned comment added by 80.192.134.161 (talk) 23:04, 6 January 2018 (UTC)[reply]

Reading list? Eh, what that? --Redrose64 🌹 (talk) 23:05, 6 January 2018 (UTC)[reply]
No, but you can see meta:2016_Community_Wishlist_Survey/Categories/Reading#Browser_reading_list and the tracking task, phab:T91902. — xaosflux Talk 23:32, 6 January 2018 (UTC)[reply]

Implementing Wikivoyage banner in a Wikipedia

Hello! I ask here since I think there'll be more feedback: is it possible to import the Wikivoyage main page banner into a Wikipedia? I'm an admin in the Basque Wikipedia and want to propose some changes to the main page, and this feature looks extra cool. -Theklan (talk) 00:03, 7 January 2018 (UTC)[reply]

If you mean the automatically scrolling content at wikivoyage: then the feature is called Carousel. Try copying the Carousel pages and Carousel lines in other pages in a Carousel search of their MediWiki namespace. I don't know whether you need more. Their Main Page code uses wikivoyage:Template:Banner so also copy that if you want to use Carousel in the same way. PrimeHunter (talk) 00:35, 7 January 2018 (UTC)[reply]

Editing DYK nominations on mobile is broken?

For some reason, when I try to edit a DYK nomination on my phone or tablet using the mobile site, the edit page just refuses to load: the loading icon pops up, then it just gets stuck. Editing using the desktop site works though, and editing any other page on mobile seems to work. Anyone else have this problem? I left a message at WT:DYK but didn't get any replies. Narutolovehinata5 tccsdnew 01:41, 7 January 2018 (UTC)[reply]

Example: Template:Did you know nominations/Bill Dubuque. The only pencil icon in mobile says "Edit the lead section of this page", but it goes to https://en.m.wikipedia.org/w/index.php?title=Template:Did_you_know_nominations/Bill_Dubuque&mobileaction=toggle_view_mobile#/editor/1 which keeps trying to load. editor/1 is the first named section. The lead is editor/0 so it should have said https://en.m.wikipedia.org/w/index.php?title=Template:Did_you_know_nominations/Bill_Dubuque&mobileaction=toggle_view_mobile#/editor/0 which works. I guess mobile makes a wrong edit url because it gets confused by the lead code using a template to produce a section heading which is also visible in the source but as part of a template parameter:
{{DYKsubpage
|monthyear=January 2018 
|passed=<!--When closing discussion, enter yes, no, or withdrawn -->
|2=
{{DYK conditions}}
====Bill Dubuque====
...
}}
PrimeHunter (talk) 12:39, 7 January 2018 (UTC)[reply]
Best to use the Reveiw or Comment link on the top110.227.71.46 (talk) 15:45, 7 January 2018 (UTC)[reply]

Help:Customizing toolbars features not working

Anyone else not able to get any of the features described at Help:Customizing toolbars to work?

I have loaded various p-personal and pt-tb tools in User:SMcCandlish/common.js and none of them have any effect at all. I thought perhaps that meta:MoreMenu (available under Special:Preferences##mw-prefsection-gadgets, in "Appearance") was interfering, but turning it off had no effect. Anyone else had this issue and worked it out? At this point I seem left with deleting one item from my common.js and common.css pages, and turning off all optional preferences, one at a time, which would take all day.  — SMcCandlish ¢ >ʌⱷ҅ʌ<  11:49, 7 January 2018 (UTC)[reply]

@SMcCandlish: Wouldn't creating a new account for testing purposes be quicker? I would recommend using a name like SMcCandlish (alt). There are several people here who have an alternative account they use for testing techy stuff. (((The Quixotic Potato))) (talk) 12:08, 7 January 2018 (UTC)[reply]
Your personal js and css pages run when you preview them so many things can be tested if they should have an affect when viewing those pages. Try changing null !== moveSubpagesBox to undefined !== moveSubpagesBox in the old version [8] of User:SMcCandlish/common.js. PrimeHunter (talk) 12:26, 7 January 2018 (UTC)[reply]
I already have a bunch of doppelganger accounts ("SMcLandish", etc.) to prevent impersonation, so I could probably use one of those. Good idea. PS: I wasn't really asking if anyone would go over my js and css pages; people have better things to do. Was just hoping someone had some kind of "oh, yeah, that happens when [well known issue here]" kind of answer from previous experience.  :-)  — SMcCandlish ¢ >ʌⱷ҅ʌ<  12:44, 7 January 2018 (UTC)[reply]
That nullundefined trick did have a positive effect; I'm at least getting the "subpages" menu to appear now, and "auto ed" is now also showing up under "More". I'm not sure where I got that null from, presumably an old version of WP:PAGEMOVER, which currently uses the undefined version.  — SMcCandlish ¢ >ʌⱷ҅ʌ<  12:55, 7 January 2018 (UTC)[reply]

A note about wmf labs AFD stats…

The log takes your earliest timestamp, and attaches the !vote to that date even while acknowledging that you !voted later. This messes up your logs if you delsort but then come back a week later or so and !vote. See December 30 and you'll find a Jan 6 !vote listed in the middle of those Dec 30s. L3X1 Happy2018! (distænt write) 17:03, 7 January 2018 (UTC)[reply]

I'm investigating this and should have a fix out pretty soon. Enterprisey (talk!) 19:05, 7 January 2018 (UTC)[reply]

Image won't render?

In the Blood in stool, there's a section of wikimarkup that says

[[File:Red feces.png|thumb|Hematochezia typically presents with bright red blood mixed  in with the stool.]]

which renders as

File:Red feces.png
Hematochezia typically presents with bright red blood mixed in with the stool.

rather than the expected image. What gives? The markup seems valid to me, and replacing the image what different one renders correctly. What's the problem/fix? Headbomb {t · c · p · b} 01:01, 8 January 2018 (UTC)[reply]

File:Red feces.png is on MediaWiki:Bad image list. PrimeHunter (talk) 01:12, 8 January 2018 (UTC)[reply]
So, how do we bypass this? Headbomb {t · c · p · b} 01:41, 8 January 2018 (UTC)[reply]
You can post a request at MediaWiki talk:Bad image list. PrimeHunter (talk) 01:45, 8 January 2018 (UTC)[reply]
I added it as an exception to disallowing use of this image. Please note, for future reference, that PrimeHunter is correct about the correct place to make these requests. עוד מישהו Od Mishehu 07:53, 8 January 2018 (UTC)[reply]
@PrimeHunter:As an admin, you could have also done this - see here how to do it - since Wikipedia is not a bureaucracy. There is no doubt that allowing this exception is reasonable. עוד מישהו Od Mishehu 08:22, 8 January 2018 (UTC)[reply]

Can anyone safely run a links check on New Jersey Women's Hall of Fame? I ran the "Fix dead links", but I'm still concerned. I found under "External links" section that the New Jersey Women's Hall of Fame website link was a redirect to an Asian language site. I removed it. I randomly tried another link on the page, and it also redirected to an Asian language site. Can anyone safely check the links on this article and remove anything redirecting to Asian websites? — Maile (talk) 14:11, 8 January 2018 (UTC)[reply]

@Maile66: This looks like a case of usurped website. You can set |deadurl=usurped in the CS1/2 citations to remove the offending links. --Izno (talk) 14:34, 8 January 2018 (UTC)[reply]
@Izno: set it where? I don't fully understand. — Maile (talk) 14:40, 8 January 2018 (UTC)[reply]
@Maile66: In each citation template that you think needs to be fixed. --Izno (talk) 14:42, 8 January 2018 (UTC)[reply]
@Izno: I won't actually know unless I open every page link to see which ones are redirecting. I was hoping there would be something more automatic. — Maile (talk) 14:45, 8 January 2018 (UTC)[reply]
@Maile66: No, there's nothing automatic. I think it's fairly safe to say, having opened one or two links, that the entire website has been usurped. --Izno (talk) 14:48, 8 January 2018 (UTC)[reply]

"Wikipedia:New_user_landing_page" should be able to opt out or outright removed

What if I want to view the non-existent page's deleteion log to see why it shouldn't be created? — Preceding unsigned comment added by Ywwuyi (talkcontribs) 14:23, 8 January 2018 (UTC)[reply]

@Ywwuyi: can you clarify? The View History (then the View Logs) controls work on the page Wikipedia:New_user_landing_page. Also, this page exists and has no deletions in the history. — xaosflux Talk 14:46, 8 January 2018 (UTC)[reply]
I go to any non-existent page and will get redirect there. I don't want to view WP:New user landing page, but a page that has yet to be created. Ywwuyi (talk) 14:57, 8 January 2018 (UTC)[reply]
@Ywwuyi: this is by design, only on the Article namespace. See WP:ACTRIAL. The reason why is that very new editors (such as your self) are not allowed to create pages in that namespace directly. Feedback to improvement of the landing page or suggestions for specific edits on it can be submitted at Wikipedia talk:New user landing page. — xaosflux Talk 20:02, 8 January 2018 (UTC)[reply]
The deletion log is hard to find for logged in users without autoconfirmation. If you log out then you get the deletion log at for example Rifal Lastori instead of being redirected to Wikipedia:New user landing page. See Wikipedia talk:Autoconfirmed article creation trial/Archive 5#Deletion log not shown. PrimeHunter (talk) 20:18, 8 January 2018 (UTC)[reply]
(a) The normal way is to: (i) click the redlink; (ii) in the browser's address bar, remove the query string parameters &action=edit&redlink=1; (iii) find where it says ?title= and after this, insert this string: Special:Log&page= and press ↵ Enter
(b) The alternative way is to (i) go to any page; (ii) click the "View history" tab; (iii) click the "View logs for this page" link near the top; (iv) in the browser's address bar, locate the page= parameter, remove whatever follows that and append the name of the page that you're interested in (use underscores instead of spaces). --Redrose64 🌹 (talk) 20:42, 8 January 2018 (UTC)[reply]

Warning on lowercase title

I've added {{lowercase title}} at eldiario.es per WP:NCLOWERCASEFIRST, and the template shows a red warning "Warning: Display title "eldiario.es" overrides earlier display title "<i>Eldiario.es</i>".". The template works fine at pages like iPad or eBay.

What am I doing wrong? Diego (talk) 14:28, 8 January 2018 (UTC)[reply]

P.S. I've found that the error message only appears when templates {{Infobox Newspaper}} and {{lowercase title}} are both included. Diego (talk) 14:35, 8 January 2018 (UTC)[reply]

@Diego Moya: That just means there is a second display title being defined; in this case, in {{infobox newspaper}}. Review the use instructions there to fix the issue. (Recommendation: you may want to use the DISPLAYTITLE magic word in the article after fixing the displaytitle issue in the infobox rather than a passthrough template such as {{lowercase title}} since you should include the italics i.e. {{DISPLAYTITLE:''eldiario.es''}}.) --Izno (talk) 14:39, 8 January 2018 (UTC)[reply]
Thanks, that solved the problem. It required adding |italic title=no to infobox newspaper to prevent it from adding an incorrect uppercase title definition. Diego (talk) 15:37, 8 January 2018 (UTC)[reply]

16:19, 8 January 2018 (UTC)

Can somebody help me with my edit request? I am trying to add a |text= option per {{Legend#Full parameter list}}.--Nevéselbert 18:15, 8 January 2018 (UTC)[reply]

Indication whether Talk page posts are present or not

It annoys me that I have to view an article's Talk page to see whether any posts have been made. Could there be an indication next to the link? Example: 0 if posts not present. Even better might be to keep the link's colour light red (as if the Talk page is totally empty) even if templates are present. In other words, disqualify the templates from changing the colour. Akld guy (talk) 19:32, 8 January 2018 (UTC)[reply]

I don't know if this is technically feasible, but I absolutely second this proposal. – Uanfala (talk) 19:58, 8 January 2018 (UTC)[reply]
1. Realistically, I don't think the page-rendering software is going to read every linked talk page to see if there are any section headings in it, just to avoid annoying a few editors (this does not annoy me). And that would be the only way to do it, the page history does not distinguish between "posts" and other edits.
2. Your "even better" is a non-starter anyway, as it would make it impossible to see if the target page exists. Redlink means no page, and it's unlikely we're going to invent a third color for this purpose—even if that's something within the control of our developers. ―Mandruss  20:02, 8 January 2018 (UTC)[reply]
It would be nice if MediaWiki assigned a class to the "Talk" link for an almost empty talk page so the link could be styled like Help:Link color#Styling all links just for you. But if it was a general MediaWiki feature then you would need a general way to define "almost empty". PrimeHunter (talk) 20:34, 8 January 2018 (UTC)[reply]
Even if possible, that's a lot of feature creep for a minor annoyance to a small minority of editors. How about making pings and other userpage links generate notifications when added after the fact, or at least devising a way to make it clear that no notification was generated? That's a major annoyance to what I suspect is a majority of editors who use notifications. It's actually more than an annoyance, as notifications are an essential part of discussions. This, not single-click-saver features, is the kind of thing where our developers should be spending their limited time. ―Mandruss  20:51, 8 January 2018 (UTC)[reply]
Mandruss, in your personal preferences ("Notifications" tab) you can turn on the option of receiving notifications for each successful (or unsuccessful) ping you make. – Uanfala (talk) 23:06, 8 January 2018 (UTC)[reply]
@Uanfala: I checked the "Failed mention" box and still got no indication that this edit did not generate a notification. It isn't clear to me that that option addresses the problem I'm talking about. In any case, if there were an option for such a solution it should be enabled by default. ―Mandruss  23:32, 8 January 2018 (UTC)[reply]

If we're brainstorming, something like [talk (3)] rather than [talk] to indicate 3 level 2 headers could very likely be scripted. Headbomb {t · c · p · b} 20:54, 8 January 2018 (UTC)[reply]

That's a pretty good idea. I wrote a script that does that: User:Enterprisey/talk-tab-count.js Enterprisey (talk!) 21:55, 8 January 2018 (UTC)[reply]
@Enterprisey: Thanks. I'd like to check it out, and I guess it's time I learned how to enable something like that for my account. How? ―Mandruss  22:07, 8 January 2018 (UTC)[reply]
Oops, I forgot to write documentation and installation instructions - I'll ping you when those are posted. Enterprisey (talk!) 22:10, 8 January 2018 (UTC)[reply]
Mandruss, installation instructions are up at User:Enterprisey/talk-tab-count. Enterprisey (talk!) 22:16, 8 January 2018 (UTC)[reply]
@Enterprisey: Thanks. It shows (14) for Talk:Donald Trump, which has 10 L2s and 5 L3s. ―Mandruss  22:24, 8 January 2018 (UTC)[reply]
Okay, I fixed it so it only counts level 2's. I don't really want to fix it further to display 10 instead of 9, because it would be too expensive to check for section headers inside everything transcluded on the page. Enterprisey (talk!) 22:35, 8 January 2018 (UTC)[reply]
Roger. Guess we're done here then, and everybody's happy. The performance hit seems large enough to be noticeable, but no problem if the user deems the benefit worth it. I'll decline. ―Mandruss  22:42, 8 January 2018 (UTC)[reply]
Whoa! That was quick! Who would have thought this would get solved so easily. Good job, Enterprisey! – Uanfala (talk) 23:04, 8 January 2018 (UTC)[reply]

Can someone add this to the Gadgets please? (((The Quixotic Potato))) (talk) 04:42, 9 January 2018 (UTC)[reply]

Sure. There used to be a dedicated page, but now all we need is a WP:VPT discussion, which I'll put in a new subsection. Enterprisey (talk!) 05:30, 9 January 2018 (UTC)[reply]

Hmmm, my suggestion got hijacked to a totally different one. I'm not impressed. Akld guy (talk) 05:34, 9 January 2018 (UTC)[reply]

Were you asking for a post count in any link to a talk page, not just the one in the tab at the top? That would be technically feasible too, if a bit visually noisy. Enterprisey (talk!) 05:40, 9 January 2018 (UTC)[reply]
@Akld guy: It seems to me that Enterprisey's tool gave you exactly the functionality you asked for: it puts the number of talk page posts that have been made to a talk page at the top of the article, so that way, if a talk page hasn't had any posts, it will read 0. Please give it a try: see User:Enterprisey/talk-tab-count for installation instructions. Mz7 (talk) 06:25, 9 January 2018 (UTC)[reply]

Proposal to make talk-tab-count a gadget

The user script User:Enterprisey/talk-tab-count was written as part of the above discussion to display the post count of a talk page on the "Talk" tab. Should it become a gadget? Enterprisey (talk!) 05:55, 9 January 2018 (UTC)[reply]

In the article, "List of SpongeBob SquarePants episodes", the featured star is unusually huge. In fact, it takes up much of the article. I doubt it is just my browser since Archive.is also captures that same mistake, seen here. Yoshiman6464 ♫🥚 22:06, 8 January 2018 (UTC)[reply]

I saw what you were talking about, but it went away when I reloaded the page. Enterprisey (talk!) 22:18, 8 January 2018 (UTC)[reply]
@Enterprisey: Because I made an edit. (((The Quixotic Potato))) (talk) 22:19, 8 January 2018 (UTC)[reply]
That's interesting; I wonder why two copies of the template were necessary. Enterprisey (talk!) 22:20, 8 January 2018 (UTC)[reply]
I have removed one of the two. (((The Quixotic Potato))) (talk) 22:22, 8 January 2018 (UTC)[reply]
(edit conflict × 2) Purging and a null edit didn't seem to fix it for me but moving the {{featured list}} template up did. Not sure why. – by AdA&D at 22:21, 8 January 2018 (UTC)[reply]
Hm, weird editconflicts. (((The Quixotic Potato))) (talk) 22:25, 8 January 2018 (UTC)[reply]
I guess the problem when {{featured list}} is at the bottom is related to the page being in Category:Pages where template include size is exceeded. Templates at the bottom are not supposed to be transcluded at all in that case. A link to Template:Featured list was displayed as normal for pages in the category, but maybe a messy partial transclusion was made. PrimeHunter (talk) 22:32, 8 January 2018 (UTC)[reply]
I've seen a massive FA star before, and that was also on a page that had maxed out the template limits. It was the "natural" size, as displayed at File:Cscr-featured.svg. --Redrose64 🌹 (talk) 23:32, 8 January 2018 (UTC)[reply]

WikiData moving to new servers - data loss possible if you try to save during the move

From Tech News (posted above a while ago, but not everyone reads it who should):

Wikidata will be moved to its own database servers. This is because it is growing and needs more resources. Because of this you will be able to read but not edit Wikidata and the German Wikipedia between 06:00 and 06:30 UTC on 9 January. You might lose edits if you try to save during this time. This includes editing the language links on other wikis. [23]

 — SMcCandlish ¢ >ʌⱷ҅ʌ<  22:48, 8 January 2018 (UTC)[reply]

Reset edition calling cite module for cite_book

There is a convoluted math-source template, "Template:Introduction to Algorithms" which attempts to alter authors based on "edition=" or "1=" but cannot reset "edition=3" as "edition=3rd" so I generated "version=3rd edition" which works, but cannot reset "edition=" as blank to hide when invoke of the Lua script cite module. Used in page "Merge sort". Are there any other solutions, beyond simply call {{cite book}} rather than invoke module? No hurry on this. Thanks. -Wikid77 (talk) 02:00, 9 January 2018 (UTC)[reply]

I've reverted to {{cite book}} for now. I think Module:Citation/CS1/Wrapper could use a blacklist parameter to prevent the module from using specified parameters, |edition= in this case. Perhaps |blacklist=edition,example2,example3 would prevent the module from passing |edition=, |example2=, and |example3= to the wrapped template. — JJMC89(T·C) 06:17, 9 January 2018 (UTC)[reply]
Yes, until this works properly reversion is appropriate. I encountered similar issues trying to get the module invocation to work for {{mathworld}} (in that case not deployed because it was more obviously problematic already for the sandbox/test cases); there didn't seem to be a way to get it to ignore the template-specific parameters that should not be passed to the invocation. —David Eppstein (talk) 06:29, 9 January 2018 (UTC)[reply]