Wikipedia:Request a query/Archive 6

Archive 1

←

Archive 4

Archive 5

Archive 6

Archive 7

Editors who would become extended confirmed upon their next edit

For statistical purposes, I'm wondering if I could get a count of editors who would become extended confirmed upon their next edit? That is, editors with 500 edits, account age >30 days, and are not/have never previously been extended confirmed? Best, HouseBlaster (talk • he/they) 02:03, 17 August 2025 (UTC)

122. List at quarry:query/96352. —Cryptic 02:21, 17 August 2025 (UTC)

Would editors with more than 500 edits also qualify? How about those with exactly 499 edits (assuming that no pages they have edited get deleted before their next edit)? Certes (talk) 10:41, 17 August 2025 (UTC)

More than 500 might, if they hadn't edited since before extendedconfirmed existed; I don't know offhand whether qualified users were promoted en masse at the time, and can see arguments either way for whether they should have. Will investigate later. No, also ones who made their 501st edit before their account was 30 days old, but none since. So, yeah, likely lots.

499 edits wouldn't, since autopromotion happens on the 501st edit. Deletion doesn't matter either, since user_editcount is only incremented when you make an edit, and not ever decremented. —Cryptic 12:04, 17 August 2025 (UTC)

So, looking like 27736, unless I missed something else. The first autopromotions happened at 23:17, 5 April 2016. (That seems way late to me - I'd have guessed closer to 2010 - but it turns out to be right.) 903 of these accounts have edited since then (page ten of quarry:query/96352), so 26833 are entirely left over from before the group existed. —Cryptic 12:34, 17 August 2025 (UTC)

Thanks, Cryptic and Certes :) HouseBlaster (talk • he/they) 16:25, 17 August 2025 (UTC)

1RR pages missing editnotice

Could I get a list of pages in Category:Wikipedia pages subject to a one-revert restriction for which the corresponding editnotice (Template:Editnotices/Page/[basename]) does not exist? Feel free to leave out the ones in category talk, template talk, etc. Thanks in advance. Extraordinary Writ (talk) 02:14, 10 September 2025 (UTC)

How broad is "etc."? Do you just want mainspace-talk pages, then? —Cryptic 02:45, 10 September 2025 (UTC)

Never mind, quarry:query/97051 has all of them, sorted by the categorized page's namespace. First 2000 and some are for mainspace pages. —Cryptic 02:59, 10 September 2025 (UTC)

Great, thanks. Extraordinary Writ (talk) 05:35, 10 September 2025 (UTC)

Need a table for breaks down of total sidebar usage by type

Please see Wikipedia:Village pump (proposals)#Remove non-ordinal series templates for context. Could someone please help build a query that counts ordinal side bars, such as historical chronologist or otherwise ordinal, a.k.a. numbered lists versus the rest of topical based side bars, as well as the transclusion counts for each and the respective totals? Thanks in advance. Raladic (talk) 16:37, 10 September 2025 (UTC)

Bearing in mind that we can't see the templates' wikitext (see top of this page), how do you propose to distinguish between ordinal and non-ordinal sidebars? For that matter, it's not immediately obvious to me how to programmatically distinguish them even if we could see the wikitext. (And what the heck is a trans liaison count?) —Cryptic 16:46, 10 September 2025 (UTC)

And what the heck is a trans liaison count? - “transclusions”, autocorrect on mobile, sorry about that. Unfortunately sidebars are currently only grouped by topical categories, so we likely would need to use the Category:Sidebar templates by topic and count all sidebars im the caregories outside of Category:History and events sidebar templates which likely has most chronological sidebars. Wont be an ideal science, but yes, that is part of the problem, other than the obvious hunch that chronologies/numbered sidebars are a small minority. hoping maybe someone else has some idea to refine. — Preceding unsigned comment added by Raladic (talk • contribs) 16:55, 10 September 2025 (UTC)

4268 non-redirect templates in the second tree, 3479 in the first but not the second. —Cryptic 17:12, 10 September 2025 (UTC)

Wait, you wanted total transclusion counts too. quarry:query/97072. —Cryptic 17:32, 10 September 2025 (UTC)

Plus or minus an 'h'

Because of confusion on two pages, I was wondering how easy it would be to write a query that returned all pages that if a single 'h' was removed from the title of an article, it would still be an existing article. For example Sigma Phi Alpha -> Sigma Pi Alpha. Yes, the majority of these would probably be for greek letter organizations like the example, but curious as to both the results and the sql. Naraht (talk) 11:59, 10 September 2025 (UTC)

Hrm. I don't think sql is the right tool for this. It's possible, but awkward, to express replacing only the first h, or replacing only the second, or replacing only the third; and I can't think of a way around including each of those cases separately in the query, up to however many h's are in the title of the page with the most of them. (I guess that's When the pawn hits the conflicts he thinks like a king what he knows throws the blows when he goes to the fight And he'll win the whole thing 'fore he enters the ring And he'll win the whole thing 'fore he enters the ring with 26, at least for lowercase h's only.)

The right way to go about this is to download the dump of titles in mainspace and run it through a more sequential language like perl or python. Which I might be willing to do if you have a more productive reason to ask this than just mere curiosity. —Cryptic 15:21, 10 September 2025 (UTC)

(Yes, mere curiousity), I was thinking that the right way from a database standpoint would be to create a temp table (not sure if it would be instatiated or not) with article_name, article_name stripped of all h (article_name_stripped), and count of number of h in article_name (h_count). Then a self join could be done on this table where t1.article_name_stripped = t2.article_name_stripped and t1.h_count = t2.h_count + 1. that result, could be tested in another way Yes. that could give cases where it wouldn't be true (like two articles WhXhYZ and WXYhZ) but would give a *much* smaller universe to do look at.Naraht (talk) 15:54, 10 September 2025 (UTC)

You'd have to instantiate and index the temp table to have any hope of the query completing, yes. The WhXhYZ/WXYhZ example can be dealt with by looking for a Levenshtein distance of 1. Sql Server, Oracle, and Postgres all have native functions to compute that, but Mysql/Mariadb does not, nor AFAIK any other function to compute edit distance; while I can google up some implementations of it in mysql, we don't have permissions to create user functions on the toolserver replicas, so you'd still have to download the page titles and do it locally. —Cryptic 17:57, 10 September 2025 (UTC)

Yeah, the h count being off by one and the levenshtein distance =1 should be enough. I *think* a perl library could be used for that. Also, since h would never be the first letter, each of the 26* letters could be its own run. (Yes, I know the English language wikipedia allows for articles which start with either modified latin letters or things in other alphabets, however combined they are tiny). Oh well. Thank you for the ideas.Naraht (talk) 20:19, 10 September 2025 (UTC)

 Completed

List of new WikiProject pages

Resolved

I'd like a query that will find pages meeting these criteria:

• Begin with Wikipedia:WikiProject
• Are not subpages
• Have been created recently (e.g., last few months, this calendar year)

WhatamIdoing (talk) 01:20, 9 September 2025 (UTC)

I recently fixed the one at Database Reports that basically does this but only looks at the past 30 days (from recentchanges). Check this out: Wikipedia:Database reports/New WikiProjects. If you need to go back further, I could rewrite it. Let me know. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 02:15, 9 September 2025 (UTC)

Thanks! I looked back through the history to find some others. It looks like in the last year or so, about 10 pages have been created, and probably more than half of those could be deleted at MFD. I'll take the list to WT:COUNCIL. WhatamIdoing (talk) 02:51, 9 September 2025 (UTC)

Looking at the database report, I'd add to my original criteria: "4. Not a redirect (including 'not a soft redirect')" and "5. Not already deleted." WhatamIdoing (talk) 02:59, 9 September 2025 (UTC)

In the previous report, it included a deleted one only because I asked for it to be deleted after the report was generated. Anyway, I think these criteria are mostly doable, although I'm not sure offhand if the database supports checking for soft redirects. I can probably look at this tomorrow. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 03:17, 9 September 2025 (UTC)

WhatamIdoing: Check out this query. It goes back six months, and should keep out both hard and soft redirects (removing soft redirects was easier than I initially thought). Also, there should never be any deleted ones in the list. Let me know if this works for you. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 04:25, 10 September 2025 (UTC)

Yes, that looks good. WhatamIdoing (talk) 23:08, 10 September 2025 (UTC)

(Wouldn't using the page creation log be easier and faster? Plus, it wouldn't limit you to 30 days' age. —Cryptic 03:05, 9 September 2025 (UTC))

Perhaps. I didn't challenge myself to make the existing report better like I did with "WikiProjects by human changes". I literally just fixed the database bug in it. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 03:10, 9 September 2025 (UTC)

Hrm. Neither of these are going to find pages created somewhere else and later moved to Wikipedia:WikiProject Whatever, like Wikipedia:WikiProject Sustainable Development was in July. I found that by starting at page and working back to a creation timestamp, but that wouldn't work either for a page created long ago and only recently moved. And the move log is miserable to work with, though since we only care about the current location, it's at least feasible. —Cryptic 03:23, 9 September 2025 (UTC)

That looks like the only one, which isn't really surprising. I didn't attempt to look more than one move back in the query, but there were few enough to check manually; all the other moves this year were either from other WP:WikiProject* titles or were reverts of moves away from those. —Cryptic 03:43, 9 September 2025 (UTC)

In that case, I could check for "Wikiproject" as well as "WikiProject". Since we show redirects, we would pick up where it originally came from. That may not resolve all cases, but resolving all cases might be harsh, performance-wise. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 03:45, 9 September 2025 (UTC)

The case I expected to find (and didn't initially) was a page created in userspace or draft:, edited a bit, then moved to WP:WikiProject*. Going back a bit further to the start of 2023, that's happened at (at least) Wikiproject Africa environment, Draft:Wikipedia:WikiProject Australian Transport, User:LoomCreek/sandbox/Wikiproject Housing and Tenant Rights, and User:Zazpot/WikiProject Risk. There aren't enough pages currently at Wikipedia:WikiProject% that querying them is time-intensive; the main issue is that it's very difficult to look more than one move back to omit the innocuous moves like WP:WikiProject Ireland->Baldonnel, County Dublin->WP:WikiProject Ireland but not ones like [[(some hypothetical title in userspace)]]->WP:WikiProject Abandoned Articles->WP:WikiProject Abandoned articles. —Cryptic 03:58, 9 September 2025 (UTC)

Based on my current understanding (and I could be mistaken), this is resolved in the database with rev_parent_id=0 after a page move. So if you're looking for when a WikiProject enters the project namespace, you know that it has become a "new" WikiProject. When it's a draft or in user space, it's in development (or just a user's brainstorming), and we don't care about those for our purposes. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 09:16, 10 September 2025 (UTC)

rev_parent_id isn't completely robust; it's not updated when history is merged or split. So, for example, there are three revisions of 2024 Minas Gerais road crash with rev_parent_id=0 (1264606347, 1264609936, 1264650842) and none of Home library service. The way to get around that is to JOIN revision ON rev_id = (SELECT rev_id FROM revision WHERE /*something with a composite index with rev_timestamp, usually rev_page=some page id*/ ORDER BY rev_timestamp ASC LIMIT 1), which looks like it's less efficient than looking at rev_parent_id but is actually much more so, and always finds the earliest visible revision at a given page (or other constraints).

That's admittedly rare, but it does happen. The more serious problem with your query 97050 is that a page that has more than a six-month delay between creation and their move to a WikiProject title will never show up on it. —Cryptic 15:03, 10 September 2025 (UTC)

Since we're talking about WikiProjects, I'm not sure why we would focus on article examples. Given WikiProjects go through these type of changes, are they of concern to us here? If a WikiProject is merged or split, that is occurring normally per a community decision, likely well after they are "new", and a new WikiProject created from a split would show up as a new page.

Per the last point, I had to pick a cut-off point per the request. It can easily be extended to longer periods. Also, we're looking at new pages starting with 'WikiProject'. If there is a move to a WikiProject title during the six months, it should be captured. I don't see the serious problem here, but if you can rework the query to demonstrate the issue, that would show me what I'm missing here. I am teachable. :) Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 19:19, 10 September 2025 (UTC)

Also, I've written a subquery like you show for a different report to find an original creator of an article, like when it was a draft or in their user space (so I can show the true creator), but since we're only looking at new pages in the project space, this case seems awfully rare. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 19:34, 10 September 2025 (UTC)

Wikipedia:WikiProject Risk was in userspace for more than six years before being moved to its current title. It's conceivable that some day, a page created in 2002 could be moved to a WikiProject title. What you're querying for is pages currently at a WP:WikiProject title whose first revision was within the last six months. That's inevitably going to eventually miss newly-blue WikiProject titles no matter what time cutoff you pick. Since there's generally going to be few results, I'd think it would be more preferable to have false positives (by including all recent moves, even if they were quickly reverted like the one from WP:WikiProject Ireland above; or even if merely from one WikiProject title to another, like the one from Wikipedia:WikiProject Abandoned Articles, also above) than false negatives.

I picked the anomalous rev_parent_id examples because those were the first illustrative ones I found going through my own deletion log and looking for "history merge" and "history split". If you want examples more relevant to the question at hand, then Wikipedia:WikiProject Finance & Investment, Wikipedia:WikiProject Germany, Wikipedia:WikiProject Protected areas, and Wikipedia:WikiProject Women Do News all have two revisions each with rev_parent_id = 0; and Wikipedia:WikiProject Rosblofnari has none. —Cryptic 00:08, 11 September 2025 (UTC)

Per the first point, I don't see significant damage from ignoring these rare cases. After all, we'll see these projects pop up in the "WikiProjects by human changes" and "WikiProjects by changes" reports. But if you want to modify the query I wrote that WhatamIdoing requested, feel free.

Per the second point, I will review the implications for the New WikiProjects report and revise the SQL if necessary. Note this wasn't originally my code, and my effort was focused on making sure it didn't fail due to database changes, rather than making it smarter. But I will look at it. Thank you for your ideas! Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 00:20, 11 September 2025 (UTC)

Flooding AFD

@Czarking0 has an idea about how to measure AFD nominations, specifically focused on how many editors open a disproportionate number of AFD nominations. The context is a discussion about AI/LLM users potentially flooding AFD in the future.

This would require knowing the name of the editor (or some sort of unique identifier that's consistent within the set) who created any given AFD subpage + the creation date for that subpage (e.g., I created Wikipedia:Articles for deletion/Theopathy on July 3rd, so the needed information would be something like "WhatamIdoing – 20250703012900").

But rather than doing Czarking0's original analytical plan, which sounds like a lot of work, could you look at a random sample of weeks in a year, and tell us what the most number of AFD subpages created by any individual editor is? If nobody's normally creating more than 10 or 20 in one week, I think we can safely skip the "hard work" part and propose a limit based on common sense. WhatamIdoing (talk) 01:07, 9 September 2025 (UTC)

I think there is a lot value in common sense stats. Out of professional pride, I will note there are a ton of caveats for estimating maximums which I would have to re-familiarize myself with to do a true statistical study. In terms of sampling, I think we could draw meaningful results from eight random weeks in each of the last five years. That is also a small enough amount of data that I could do it by hand if needed. Czarking0 (talk) 01:18, 9 September 2025 (UTC)

Five years takes us back to August/September of 2020, which was during the Covid lockdowns. Editing patterns were off that year. Could we do 10 random non-overlapping weeks in the last four years? WhatamIdoing (talk) 01:22, 9 September 2025 (UTC)

I think I can save you some steps? quarry:query/97012 has, for each week of the year, everyone who's created more than 10, along with the total afds created that week, starting in 2020. I don't think a random sampling of weeks would have given you a good picture, whether or not you picked one of the weeks where a single user created a full quarter of the afds (there were several). But if you really want the full datasets, let me know. —Cryptic 02:46, 9 September 2025 (UTC)

It looks like there are 1,661 records showing an editor creating 10+ AFDs in a single (calendar) week.

The highest in that dataset is 135 in a single week, which means an average of 20 per day should have no effect on anyone. Only nine editors ever listed more than 70 in a single week, which means that an average of 10 per day should have almost no effect on anyone. (The underlying goal is to establish a "no flooding" rule before anyone gets the idea to turn an LLM-based bot loose on AFD.)

@Czarking0, I wonder if you think that more evaluation would really be helpful, or if we should just go with a simple "max 70 noms per week" proposal. WhatamIdoing (talk) 06:46, 14 September 2025 (UTC)

I think this sort of data gets you to something you could build consensus around regardless of my opinion. It really comes down to what does the community consider flooding? 10 in a week does not seem like flooding to me (I think I have done that). 70 in a week does seem like flooding. I agree that additional analysis is mostly academic but I am unsure what the community would think. Czarking0 (talk) 16:21, 14 September 2025 (UTC)

I've asked people what they think 'flooding' is at Wikipedia:Village pump (idea lab)#Quick straw poll about AFD flooding. (Neither of us have done 10 in the same week during the last five years, but I agree with you that 10 in a week feels very reasonable. I might even accept 10 in a day.) WhatamIdoing (talk) 19:55, 14 September 2025 (UTC)

Query to find articles using both Template:use dmy dates and Template:use mdy dates (and similar)

This is a request for a query to find articles using both Template:use dmy dates and Template:use mdy dates, motivated by this conversation. I know that it is possible to use "hastemplate" searches for this, but I'd like to incorporate it into a database report.

I would also like to include similar conflicts, such as articles that contain both {{Use British English}} and {{Use American English}}, so a general-purpose query that lists pairs of such templates would be helpful, so that I can expand the report as more conflicting pairs of templates are identified.

Please let me know if this request is not specified well enough. – Jonesey95 (talk) 21:01, 15 September 2025 (UTC)

quarry:query/97196. —Cryptic 22:03, 15 September 2025 (UTC)

Super. And thanks for tweaking Wikipedia:Database reports/Multiple "Use" templates. I was making the exact same edit after reading the documentation at {{Database report}}. – Jonesey95 (talk) 22:47, 15 September 2025 (UTC)

Hot articles

I'm trying to come up with a query for the most edited pages with a talk page category (say Category:Horse breeds task force articles) in the last 7 days. This is for use in Module:Database reports/Hot articles which currently supports providing a PageAssessments-recognized project as input but not a category. (Prior art: https://github.com/kaldari/hotarticles/blob/master/runbot.php, but seems a bit obfuscated.) – SD0001 (talk) 15:36, 14 September 2025 (UTC)

The github version jumps through a lot of hoops trying to make things more efficient, but it doesn't matter for the common case and I don't... think... it would help much for cases with many pages in the category and very, very many revisions in the articles. Especially since it only looks at recentchanges anyway, so can only ever see revisions from the last 30 days. (Also, it can go by talk pages linked from a template instead of in a category, but that difference is even smaller now.) quarry:query/97157 for a version that looks at revision, so it can go back further than 30 days; the change for the template version is trivial - categorylinks to templatelinks and cl_* to tl_* - and if it really does get bad performance with ancient articles and you don't need to look past the recentchanges period, that's straightforward too. —Cryptic 16:42, 14 September 2025 (UTC)

Here is a simplified version that goes by the talk page category, but I have been using a more sophisticated approach for my version of Hot Articles that still goes by the PageAssessments project name. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 16:49, 14 September 2025 (UTC)

Hunh. I had thought FETCH FIRST 10 ROWS ONLY was an Oracleism; no idea that Mariadb supported it. —Cryptic 17:13, 14 September 2025 (UTC)

This was something I discovered when hunting down a way to do proper ranking, not wanting to leave out articles that have the same number of edits at the end of the list. I guess it can be called a "10+" result. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 17:24, 14 September 2025 (UTC)

I'd probably have flailed around with something based on DENSE_RANK(). This is a lot more elegant. —Cryptic 17:27, 14 September 2025 (UTC)

(Which oracle?) WhatamIdoing (talk) 19:56, 14 September 2025 (UTC)

Oracle Database by Oracle Corporation. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 20:18, 14 September 2025 (UTC)

Also here is an implementation of my version of Hot Articles which includes the specific query in production (trimmed down a bit from the one in Quarry). Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 17:28, 14 September 2025 (UTC)

Might be worthwhile to go with a recentchanges-based query for the convenient access to rc_bot. You can sort of get at it from revision, but only by looking at the editor's current groups; and that's not guaranteed to be accurate as of when the edit was made. —Cryptic 18:09, 14 September 2025 (UTC)

Thanks all. I set up Module:Database reports/Hot articles by category based on Cryptic's query. I wanted to add it as an option within Module:Database reports/Hot articles but that proved too difficult due to the different number of columns – the by-category listing doesn't contain rating. – SD0001 (talk) 04:40, 24 September 2025 (UTC)

Admins using the block button

Can you tell me how many admins blocked anyone (registered or otherwise) during the last year? We have 827 admins, and I have heard that most of them use their admin tools only occasionally, so I would expect the number who use the block button specifically to be low – but maybe you'll surprise me. The total number (no need for individual names) is enough for me. WhatamIdoing (talk) 20:37, 6 October 2025 (UTC)

@WhatamIdoing: According to this query, 414 of them made at least one block in the past year, out of which 41 made more than 1000 blocks each (~50% and ~5% of all admins, respectively). NguoiDungKhongDinhDanh 21:03, 6 October 2025 (UTC)

quarry:query/97803 has totals for different uses of the block button too. Do you need me to filter accounts that currently aren't admins anymore? I can think of at least one where that's the case. —Cryptic 21:04, 6 October 2025 (UTC)

No, that's good enough. Thanks! WhatamIdoing (talk) 21:35, 6 October 2025 (UTC)

Why is this query so slow?

This query of mine has been running for about 24 hours. Its content is as follow:

SELECT
	CONCAT('Thảo luận Thành viên:', page.page_title) as 'Page name',
    COUNT(revision.rev_id) as 'Revisions'
FROM
	page
JOIN
	revision ON revision.rev_page = page.page_id
JOIN
	actor ON actor.actor_id = revision.rev_actor
WHERE
	page.page_namespace = 3 AND
    page.page_is_new = 0 AND
    actor.actor_name = 'TuanminhBot' AND
    is_ipv4(page_title)
GROUP BY
	page.page_id
HAVING
	COUNT(DISTINCT revision.rev_actor) = 1 AND
    COUNT(revision.rev_id) > 1
ORDER BY
 	COUNT(revision.rev_id) DESC

I expect this to return something between 10,000 and 300,000 rows, give or take. An almost identical query, whose intention is to return only the amount of pages, has also been running for 21 hours. NguoiDungKhongDinhDanh 08:41, 7 October 2025 (UTC)

The actor view was introduced in a 2019 downgrade. It joins the actual actor table, which is hidden, to eight other tables, which slows down many queries dramatically. Using the actor_revision view instead may undo some of the damage. Certes (talk) 10:49, 7 October 2025 (UTC)

@Certes: quarry:query/97845 (actor) took 105 seconds to return 100 rows, compared to quarry:query/97844 (actor_revision)'s 117 seconds. Am I using it wrong? NguoiDungKhongDinhDanh 11:23, 7 October 2025 (UTC)

The benefit in using actor_revision over actor is that, as Certes says, it simplifies the actually-executed query, which in some cases makes the optimizer choose a better query plan. Most of the time, it won't matter, and it probably won't here.

I wrote a big long thing here about using revision_userindex instead of revision - that choice of view does make a difference, since it allows use of indices including rev_actor - but since TuanminhBot has a edit count of some 34.5 million, it's still more selective to look at page first. Query plan (either for revision or revision_userindex) looks like:

query plan

+------+-------------+---------------+-------+------------------------------------------------------------------------------------+---------------------+---------+---------------------------+----------+------------------------------------+
| id   | select_type | table         | type  | possible_keys                                                                      | key                 | key_len | ref                       | rows     | Extra                              |
+------+-------------+---------------+-------+------------------------------------------------------------------------------------+---------------------+---------+---------------------------+----------+------------------------------------+
|    1 | PRIMARY     | actor         | const | PRIMARY,actor_name                                                                 | actor_name          | 257     | const                     |        1 | Using temporary; Using filesort    |
|    1 | PRIMARY     | page          | index | PRIMARY,page_name_title                                                            | PRIMARY             | 4       | NULL                      | 15611429 | Using where                        |
|    1 | PRIMARY     | revision      | ref   | rev_page_actor_timestamp,rev_page_timestamp                                        | rev_page_timestamp  | 4       | viwiki.page.page_id       |        1 | Using where                        |
|   14 | SUBQUERY    | revision      | ref   | rev_actor_timestamp                                                                | rev_actor_timestamp | 8       | const                     | 33267992 | Using where                        |
|   13 | SUBQUERY    | logging       | ref   | log_actor_type_time,log_type_action,log_type_time,log_actor_time,log_actor_deleted | log_actor_type_time | 8       | const                     |  6444298 | Using index condition; Using where |
|   12 | SUBQUERY    | recentchanges | ref   | rc_actor,rc_actor_deleted                                                          | rc_actor_deleted    | 8       | const                     |        1 | Using where; Using index           |
|   11 | SUBQUERY    | filearchive   | ref   | fa_actor_timestamp,fa_actor_deleted                                                | fa_actor_deleted    | 8       | const                     |        1 | Using where; Using index           |
|   10 | SUBQUERY    | oldimage      | ref   | oi_actor_timestamp,oi_actor_deleted                                                | oi_actor_deleted    | 8       | const                     |        1 | Using where; Using index           |
|    9 | SUBQUERY    | image         | ref   | img_actor_timestamp                                                                | img_actor_timestamp | 8       | const                     |        1 | Using index                        |
|    8 | SUBQUERY    | block         | ref   | block_actor_deleted                                                                | block_actor_deleted | 9       | const,const               |        1 | Using index                        |
|    7 | SUBQUERY    | archive       | ref   | ar_actor_timestamp,ar_actor_deleted                                                | ar_actor_deleted    | 8       | const                     |  2576075 | Using where; Using index           |
|    6 | SUBQUERY    | block_target  | ref   | PRIMARY,bt_user                                                                    | bt_user             | 5       | const                     |        1 | Using index                        |
|    6 | SUBQUERY    | block         | ref   | bl_target                                                                          | bl_target           | 4       | viwiki.block_target.bt_id |        1 | Using where                        |
|    5 | SUBQUERY    | user          | const | PRIMARY                                                                            | PRIMARY             | 4       | const                     |        1 | Using index                        |
+------+-------------+---------------+-------+------------------------------------------------------------------------------------+---------------------+---------+---------------------------+----------+------------------------------------+

where you can see that it had to fall back on a full table scan of page with an estimate of 15.6 million rows to look at. (All the entries with select_type "SUBQUERY" are for resolving the actor view; using actor_revision will cut them down to just a single entry, the one currently numbered 14.) Only thing I can think of to improve that is to add another condition on page_title so that it only looks at page titles starting with digits (since is_ipv4(page_title) isn't indexable) - adding AND page.page_title BETWEEN '1' AND ':' to the end of your WHERE clause gets the estimate row count on page down to 7805216, about half.

Which gets me to stop and think again, because 15 million isn't all that high to start with - I was thinking in terms of enwiki numbers, where it's around 58 million. Running SHOW ANALYZE for the amended query shows that it gets through all those rows in page reasonably quickly, then hangs. What I'm guessing is happening (and I'm running right up against the edge of my competence here, and am probably already over it) - is that either it's stuck counting and sorting those results (for the HAVING clause), or that there are a lot - lot - of results, enough to overwhelm output. Quarry can't really deal with more than 100k rows in a result, though that shouldn't affect me at toolserver.

Next thing I'd try is to find a couple pages that you know will be output, for example "User talk:193.42.1.7", and restricting page_title to something like "page_title LIKE '193.42.1.%'" to see if you get results that way. I'm out of time to look at this; I can investigate some more tomorrow if what I've written so far doesn't help. —Cryptic 14:04, 7 October 2025 (UTC)

Also, this query won't do what the title says it does. The WHERE clause is dealt with well before HAVING, so informally, you're getting a bunch of pages, then all their revisions, then removing revisions not made by TuanminhBot, then checking to see if there was only one user among what's left; and that last part will always be true. —Cryptic 14:24, 7 October 2025 (UTC)

@Cryptic: Thanks. I'm {{User SQL-0.5}}, so I unfortunately don't understand everything you said. Here's what I could make of it:

• Add AND page.page_title BETWEEN '1' AND ':' to limit the number of indexed pages.
• Try adding page_title LIKE '193.42.1.%' and see if the query runs faster.
• The actor.actor_name = 'TuanminhBot' condition renders HAVING COUNT(DISTINCT revision.rev_actor) = 1 useless.

I came up with quarry:query/97847 (~64,000 rows in ~50 seconds). That's much faster than before, but the pages returned are not guaranteed to be created by TuanminhBot. How do I factor that into the query? NguoiDungKhongDinhDanh 17:53, 7 October 2025 (UTC)

That's all correct. The way you properly limit it to pages where all the edits were made by TuanminhBot (instead of just all the edits were made by the same user) is to stuff all that into a subquery, then select the pages where the single user was the bot, as in quarry:query/97857. Turns out all but one of those pages are; the only exception was vi:User talk:1.52.1.184.

Since you're getting that many results for 1.52.*, you're probably going to have to run separate queries for each pair of digits (0-255).(0-255).* - if the query had returned 640000 results and instead of 64000, Quarry wouldn't have been able to deal with it. I was hoping you'd be able to get by on page_title LIKE '1.%', another query for page_title LIKE '2.%', and so on. —Cryptic 00:51, 8 October 2025 (UTC)

page.page_title LIKE '1.%' returns ~258,000 rows in ~211 seconds. I suppose that's good enough. Thanks! NguoiDungKhongDinhDanh 06:06, 8 October 2025 (UTC)

Inactive rollbackers

Could someone create/update a query to tell me how many of the 6,978 non-admins who have the MediaWiki rollback user right and haven't edited for 3+ years? (There won't be any admins.) This should be similar to Wikipedia talk:Autopatrolled#RfC: Inactivity requirements for autopatrolled. WhatamIdoing (talk) 21:40, 11 October 2025 (UTC)

The same query as in Wikipedia:Database reports/Inactive users with advanced permissions can do this by editing the constants at the start. I've done a one-time run at quarry:query/97992. —Cryptic 01:11, 12 October 2025 (UTC)

That says we have 2,889 inactive accounts with MediaWiki rollback that haven't edited for three years, which is 41%. Thanks. I'm going to make the same proposal at Wikipedia talk:Rollback as we did at WT:Autopatrolled earlier this year. WhatamIdoing (talk) 02:18, 12 October 2025 (UTC)

Under categorized pages

Hi. Is it possible to generate a list of articles by category count? This would need to exclude hidden cats. I'm specifically interested in biographical articles beyond those already identified as needing additional categories. So I guess something like ... sits within Category:Births by year or has Category:Living people and total number of categories is x (would suggest starting with two). Thanks!! Jevansen (talk) 23:27, 25 October 2025 (UTC)

Yes, it's possible. I already have a per-project query like this here. It shows articles with the least amount of categories (4 or less; number can be changed), and excludes hidden cats and stub cats. If you need me to refine it, let me know. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 23:52, 25 October 2025 (UTC)

Excellent, I'll play around with that. Thanks Stefen. Jevansen (talk) 23:59, 25 October 2025 (UTC)

Non-IP user-talk pages with "Blanked IP talk"

Today I came across a still-present mistaken bot edit from several years ago. It made that same edit twice to that same user-talk in a several-day timeframe, so I am wonderif if there are any other such edits that were made and have still not been corrected. User:BD2412bot has been inactive for a few years, but appears to have been making many thousands of edits in the timeframe of that bad edit so I can't easily scan its contributions-list by eye. Is it possible to generate a list of all user-talk pages of non-IP editors that have {{Blanked IP talk}}? DMacks (talk) 15:31, 2 November 2025 (UTC)

I think quarry:query/98734 should be what you're looking for? I outputted the pages as full URLs but can change it to bare titles if preferable. J11csd (talk) 16:26, 2 November 2025 (UTC)

Messed up the link but fixed now. J11csd (talk) 16:29, 2 November 2025 (UTC)

Thanks! No problem for me to convert among formats, once I had the actual information in some format. DMacks (talk) 22:07, 2 November 2025 (UTC)

Simple (I think) category query

Could someone create a query that does the following:

• Looks for articles + drafts in Category:CS1 maint: url-status
• Checks if the page matches a regex that I give it: ( \{\{(?=\s*cite)(?=.*?url-status\s*=\s*live)(?!.*?archive-url).*?\}\} )
• Sorts the pages by the number of matches it has descending, then limiting the number of rows to 900.
• The output will be one column with the article name, another with the number of matches

This'll be be a page that I can run Citation bot on to fix the common CS1 error. EatingCarBatteries _{(contributions, talk)} 06:05, 4 November 2025 (UTC)

The database replicas do not have access to page content, so queries which require checking wikitext cannot be answered with database queries. (Also, that regex doesn't do what you probably think it does. Try testing against "{{cite}}url-status=live" and "{{cite url-status=live}}archive-url".) —Cryptic 06:54, 4 November 2025 (UTC)

Good to know, thank you

Yeah I'm obviously by no means an expert. That regex is AI-generated as I'm still learning, but I have successfully ran it on dumps in AutoWikiBrowser. I knew there were some misses, but I figured that it got most of the stuff that it needed. EatingCarBatteries _{(contributions, talk)} 07:13, 4 November 2025 (UTC)

How many page creations a day by IPs and non-autoconfirmed accounts in draftspace?

In phab:T409366, we discovered that temporary accounts can't create drafts due to a bug. I'd be interested in figuring out how many drafts this group created on a random sample day from before temporary accounts, to figure out the impact of the bug. For example:

SELECT COUNT(*)
FROM revision
WHERE rev_parent_id = 0 /* first revision of a page */
AND rev_timestamp > 20251001000000
AND rev_timestamp < 20251001235900
AND (
  /* revision author is an IP */
  OR /* revision author is not autoconfirmed */
)

But that doesn't use the right indexes and the revision table has a billion rows, so needs optimization to work. –Novem Linguae (talk) 01:04, 6 November 2025 (UTC)

Unless you want to go back a very long time, the creation log is better than searching revision - it's smaller, has more appropriate indexes, it keeps the entry even if the page gets deleted, and it shows where the page was originally created rather than where it is now. The last is a major win for drafts that got accepted - you wouldn't be able to distinguish them just with revision - but it's going to have some false negatives for e.g. userspace drafts that got moved to draftspace.

The harder part will be verifying that a user wasn't autoconfirmed when they made the draft. It's not difficult to verify that the account was registered less than four days before the creation log entry, but the most straightforward way to check whether they had ten previous edits is going to be very, very slow when it runs across a user who has a couple million edits. —Cryptic 01:24, 6 November 2025 (UTC)

If you accept an alternate-world definition of "autoconfirmed" where you have to have 10 currently non-deleted edits, quarry:query/98850 has all results for your October 1 test date, and quarry:query/98851 has counts for each day in 2025. Those are upper bounds; it's likely very common for such users to have deleted edits, so these both will include creations by users that actually were autoconfirmed at the time. If you need accurate numbers, I'd suggest going through the October 1 hits manually - there shouldn't be any false negatives other than the moved pages I mentioned above. —Cryptic 02:41, 6 November 2025 (UTC)

...or at least, there wouldn't have been if I hadn't accidentally omitted creations by users with fewer than ten currently-live edits, whether from before or after the page creation in question. When corrected, that approximation had 177 hits on October 1, not the 74 I found before.

Current versions of both queries check for deleted edits properly too now, unless I did something else similarly dumb. (It's been a very long day.) I can't say I'm at all proud of how I went about it, either. And it doesn't even make a whole lot of difference; including the deleted edit check only removed five draft creations from October 1, and similar numbers from other days in 2025. —Cryptic 04:34, 6 November 2025 (UTC)

FYI, bug fixed. Thanks @Novem Linguae for it. :) – robertsky (talk) 08:23, 6 November 2025 (UTC)

SQL query for pages with external links to twitter.com or x.com

Could someone please run the following query for me? I am unable to do so myself. Thank you!

SELECT DISTINCT page_title FROM page JOIN externallinks ON el_from = page_id WHERE page_namespace = 0 AND (el_to_domain_index LIKE 'com.twitter.%' OR el_to_domain_index LIKE 'com.x.%'); ~2025-32654-48 (talk) 22:30, 10 November 2025 (UTC)

No results for that; el_to_domain_index always begins with a protocol. quarry:query/99001 has what you meant. (Spoiler: there's lots.) —Cryptic 22:39, 10 November 2025 (UTC)

List of transclusions

I am looking for the list of pages that transclude {{Annual readership}}, or one of its redirects. The count tool reports a tally of 53,098, but the 'What links here?' tool (linked as 'sorted list' at Special:WhatLinksHere/Template:Annual_readership) never returns. (I don't think the problem is a volume issue, or flooding the browser; I tried the 'sorted list' link from a template that has 50 transclusions, and it never returned, either.) Can you provide a query that will produce the sorted list? There are a half a dozen redirects; if that complicates the solution, it's fine to just drop redirects. Adding interested party Andrew Davidson. Thanks! Mathglot (talk) 20:57, 8 November 2025 (UTC)

petscan:40332113. —Cryptic 21:11, 8 November 2025 (UTC)

Thank you! And I even knew about Petscan, but forgot I could use it for that. Thanks again, Mathglot (talk) 22:41, 8 November 2025 (UTC)

Yes, and thanks from me too -- it's a useful addition to my toolkit. Andrew🐉(talk) 23:20, 8 November 2025 (UTC)

Also can get it from the command line with wikiget: wikiget -b "Template:Annual readership" -tt > out.txt .. -- GreenC 17:11, 12 November 2025 (UTC)

Refreshing my article-creation list

Thanks to {{Database report}}, I'd like to automate my new-page archive, which hasn't been updated in all of 15 years. (Yes, count 'em--15.) Problem is, I have little experience with SQL (and what little I've garnered comes as a QGIS user), and I haven't encountered anyone else doing what I'm planning to in userspace. (It's saying a lot when DPL4, featured on select Miraheze wikis like mine, gets the job done with simpler and less intimidating syntax.)

Objective: Select all articles created by this requesting user, from newest to oldest (excluding redirects and deleted material), including page and prose size in bytes (and possibly article class and ref/category count)--thus replicating XTools. Automatic updates every three months. --Slgrandson (How's my egg-throwing coleslaw?) 07:28, 17 November 2025 (UTC)

quarry:query/99145 as a prototype. I've included hidden categories in the category count (solely because that's a little less effort). Prose size isn't available. Neither is reference count, though an approximately-correct relative ordering can be found by counting external links; that could be refined by excluding e.g. links to web.archive.org, but it's still going to count links in external links sections instead of references, the 'expand this' link in stub templates, searches from notability tags, etc., and will still exclude non-externally-linked references like books or Harvard refs. —Cryptic 21:23, 17 November 2025 (UTC)

Modification of query for twitter.com/x.com links with date filtering

Could someone please modify the query at quarry:query/99001? I need to count citations to twitter.com or x.com, but only based on page versions from June 2023. This would likely require joining with the `revision` table and filtering on `rev_timestamp` to be between '20230601000000' and '20230630235959'. I understand that extracting dates from the citation text itself isn't possible through Quarry. Thank you! ~2025-32977-37 (talk) 11:27, 12 November 2025 (UTC)

SQL query request for new citations to twitter.com/x.com in August 2023

Could someone please run the following query for me? I am unable to do so myself. The query is intended to count the number of new citations to 'twitter.com' or 'x.com' for each day of August 2023. Thank you!

-- This query counts the number of new citations to 'twitter.com' or 'x.com'
-- for each day of August 2023.
-- NOTE: It cannot extract the publication date from the citation text. Instead,
-- it uses the timestamp of the earliest revision of the page containing the
-- link as a proxy for the addition date.

SELECT
    first_appearance_day,
    COUNT(*) AS new_citations_count

== SQL query request for new citations to twitter.com/x.com in June 2023 ==

Could someone please run the following query for me? It is based on the query for August 2023 but modified for June 2023. Thank you!

<syntaxhighlight lang=sql>
-- This query counts the number of new citations to 'twitter.com' or 'x.com'
-- for each day of June 2023.
-- NOTE: It cannot extract the publication date from the citation text. Instead,
-- it uses the timestamp of the earliest revision of the page containing the
-- link as a proxy for the addition date.

SELECT
    first_appearance_day,
    COUNT(*) AS new_citations_count
FROM (
    -- Subquery to find the first revision timestamp for each unique external link
    SELECT
        el.el_id,
        DATE_FORMAT(MIN(rev.rev_timestamp), '%Y-%m-%d') AS first_appearance_day
    FROM
        externallinks AS el
    JOIN
        revision AS rev ON el.el_from = rev.rev_page
    WHERE
        -- Filter for twitter.com and x.com links using the indexed domain column
        (el.el_to_domain_index LIKE 'http://com.twitter.%' OR el.el_to_domain_index LIKE 'https://com.twitter.%' OR el.el_to_domain_index LIKE 'http://com.x.%' OR el.el_to_domain_index LIKE 'https://com.x.%')
    GROUP BY
        el.el_id
) AS link_creations
WHERE
    -- Filter the results to include only links first appearing in June 2023
    first_appearance_day BETWEEN '2023-06-01' AND '2023-06-30'
GROUP BY
    first_appearance_day
ORDER BY
    first_appearance_day ASC;

FROM (

   -- Subquery to find the first revision timestamp for each unique external link
   SELECT
       el.el_id,
       DATE_FORMAT(MIN(rev.rev_timestamp), '%Y-%m-%d') AS first_appearance_day
   FROM
       externallinks AS el
   JOIN
       revision AS rev ON el.el_from = rev.rev_page
   WHERE
       -- Filter for twitter.com and x.com links using the indexed domain column
       (el.el_to_domain_index LIKE 'http://com.twitter.%' OR el.el_to_domain_index LIKE 'https://com.twitter.%' OR el.el_to_domain_index LIKE 'http://com.x.%' OR el.el_to_domain_index LIKE 'https://com.x.%')
   GROUP BY
       el.el_id

) AS link_creations WHERE

   -- Filter the results to include only links first appearing in August 2023
   first_appearance_day BETWEEN '2023-08-01' AND '2023-08-31'

GROUP BY

   first_appearance_day

ORDER BY

   first_appearance_day ASC;

</syntaxhighlight> ~2025-33047-06 (talk) 16:04, 12 November 2025 (UTC)

Not sure what you are asking. Are you unable to use Quarry and run it yourself? -- GreenC 17:07, 12 November 2025 (UTC)

This isn't going to do what you want. If it completed (it wouldn't), it would show, of the pages that currently have an external link to twitter or x, the ones that were created in August 2023. There's no guarantee that the links were present since creation, and no reason to think that no links were newly-added to existing pages in that month.

To find when the links were actually added, you'd have to use the text table, and compare the wikitext between past revisions. That's not practical, even if you only look at pages that currently have a matching external link; and the public replicas don't have text anyway. Best bet is to download a consecutive pair of database dumps from around then, extract a list of pages containing external links from each, and compare them. —Cryptic 18:25, 12 November 2025 (UTC)

Nonetheless, quarry:query/99051 has what your query above would've resulted in, if it worked (unlikely) and if it didn't double-count pages for each link they have matching one of those formats. —Cryptic 19:27, 12 November 2025 (UTC)

SQL query to count total URL citations to twitter.com and x.com

Could someone please run the following query for me? I am unable to do so myself. This query counts the total number of URL citations to twitter.com and x.com. Thank you!

SELECT COUNT(*) FROM page JOIN externallinks ON el_from = page_id WHERE page_namespace = 0

 AND (el_to_domain_index LIKE 'http://com.twitter.%'
      OR el_to_domain_index LIKE 'https://com.twitter.%'
      OR el_to_domain_index LIKE 'http://com.x.%'
      OR el_to_domain_index LIKE 'https://com.x.%'); ~2025-34523-00 (talk) 15:13, 18 November 2025 (UTC)

224820. (Be aware that if a single page has more than one identical link to a matching url, it'll only be counted once.)

Creating a non-temporary account here logs you into quarry: as well, so you won't have to wait on someone else running these for you. —Cryptic 20:57, 18 November 2025 (UTC)

SQL query for new citations to twitter.com/x.com on August 3rd, 2023

Could someone please run the following query for me? I am unable to do so myself. This query is intended to count the number of new citations to 'twitter.com' or 'x.com' on August 3rd, 2023. Thank you!

SELECT COUNT(*) FROM externallinks AS el JOIN revision AS r ON el.el_rev_id = r.rev_id JOIN page AS p ON r.rev_page = p.page_id WHERE p.page_namespace = 0 AND r.rev_timestamp BETWEEN '20230803000000' AND '20230803235959' AND (el.el_to LIKE 'http://twitter.com/%' OR el.el_to LIKE 'https://twitter.com/%' OR el.el_to LIKE 'http://x.com/%' OR el.el_to LIKE 'https://x.com/%');

~2025-34472-43 (talk) 11:43, 19 November 2025 (UTC)

How are you coming up with these queries? —Cryptic 18:19, 19 November 2025 (UTC)

Query request: Daily count of revisions adding twitter.com/x.com links in August 2023

Hello, I am trying to find the number of *revisions* on English Wikipedia (enwiki) that added an external link to 'twitter.com' or 'x.com' for each day in August 2023. I have been unsuccessful in finding an existing query on Quarry that meets these specific requirements. I also attempted to create a new query myself, but I was unable to log in. Based on previous discussions on this page, I understand that directly querying the wikitext to determine when a link was added is not feasible with the public database replicas. Could someone please assist me in obtaining this data? Perhaps there is an alternative method, such as analyzing database dumps. Any help would be greatly appreciated. Thank you! ~2025-34883-99 (talk) 16:04, 19 November 2025 (UTC)

This is the same problem as recently asked at Wikipedia:Village pump (technical)/Archive 225#Find articles that have recently had a specific word added?. —Cryptic 18:30, 19 November 2025 (UTC)

Querying "Number of page watchers who visited in the last 30 days" on a list of pages

I don't think this is doable, but I figured I would ask anyway, just in case. In the page information for each page, there is a value shown for "Number of page watchers who visited in the last 30 days". Can this looked up by page in a database query somehow? I believe I can look up the number of watchers, but not this value, unless I'm missing something. Stefen 𝕋ower_{'s got the power!!1!} ^{Gab • Gruntwerk} 07:21, 24 November 2025 (UTC)

You can't do either. They're queries against the watchlist table, which isn't in the public replicas. includes/Watchlist/WatchedItemStore.php, look for countWatchers() and countVisitingWatchers(). —Cryptic 15:33, 24 November 2025 (UTC)

Single-article users

This query would be for articles created by accounts with no other edits. In other words, the only edits were to a single article, which was created by said account, and there would be a list of such pages. RanDom 404 (talk) 17:24, 25 November 2025 (UTC)

This turns out to be somewhat difficult and very slow due to the way the indexes are set up. So before I spend a couple hours on it, some clarification please:

• By "single article", you mean a page currently in the main namespace, correct? There's going to be a lot of users who've created a single user page, draft, or talk page, and edited no other page.
• By "no other edits", do you mean no edits to other articles, or no edits to any other page at all?
• How about deleted edits? If a user has fifty deleted edits to Deleted Subject A, but all their non-deleted edits are to Existing Subject B which they originally created, should they still be listed? (They probably will even if you answer no; the query's going to be slow enough already, and it'd be prohibitively difficult to correctly match deleted edits to Draft:Deleted Subject C whose later recreation by this user was moved to Existing Mainspace Title D.)

—Cryptic 04:38, 27 November 2025 (UTC)

1. Yes, just in the mainspace

2. No other edits to any page at all, regardless of namespace. I imagine this could narrow results.

I don't know how many cases of the last one there are, and since you say it is required to run it that way, the query can be ran counting articles where the user had deleted edits. "Articles created by users with no other edits" I guess (except to that article). RanDom 404 (talk) 16:06, 27 November 2025 (UTC)

Split up between quarry:query/99361, 99362, 99363, and 99365. Allowing edits to other non-mainspace pages gave about half again as many results in my tests with articles starting with 'X'. —Cryptic 17:55, 27 November 2025 (UTC)

And there are a total of 139112 results when adding them together. By the way, would an additional query be possible? It would be the same, but with the additional condition: in Category:All articles lacking sources. There wouldn't be as many results to run, since there are only about 40,000 articles in the category. RanDom 404 (talk) 18:00, 27 November 2025 (UTC)

quarry:query/99370. —Cryptic 18:53, 27 November 2025 (UTC)

These should take pretty quick so might as well ask here. What about one for Category:All orphaned articles, and one for articles in both Category:All orphaned articles and Category:All articles lacking sources? RanDom 404 (talk) 17:26, 28 November 2025 (UTC)

quarry:query/99444 and 99445 respectively. —Cryptic

Redlinks

If there's any way to get a list of every redlink on the site (ideally sortable by how many times the same redlink pops up), I'd be eternally grateful. I've been trying to make Wikipedia:Requested articles into something more useful for the average newbie (actually checking to see if there's a good chance the aforementioned subjects are notable or at the very least not spam) and I think having this information could compliment that goal very well. Clovermoss🍀 (talk) 04:40, 2 December 2025 (UTC)

@Clovermoss: This sounds like Wikipedia:Most-wanted articles. — DVRTed (Talk) 04:58, 2 December 2025 (UTC)

Not quite. I want every red link. Clovermoss🍀 (talk) 05:10, 2 December 2025 (UTC)

By "every", do you mean not just from mainspace to mainspace, or do you mean without WP:MWA's limit of being linked from at least 500 different pages? If the former, Special:Wantedpages does that; if the latter, there are a lot of them - about eight and a half million different redlinks - most with only a couple pages linking to each. —Cryptic 06:40, 2 December 2025 (UTC)

The latter. Eight and a half million is way more than I was expecting, wow. Is it even possible to do a query that large? Organizing this is going to take a lot longer than I thought. Clovermoss🍀 (talk) 06:47, 2 December 2025 (UTC)

It is, but not to display it on Quarry. I could email it to you, I suppose. It's much more manageable if you cut off the very bottom - there's about 6 million redlinks linked from only one article, another million from only 2, and 850 thousand from between 3 and 9. —Cryptic 07:02, 2 December 2025 (UTC)

Okay, email me all the redlinks featured in more than two articles. I can worry about the rest a decade from now 😂. Clovermoss🍀 (talk) 07:05, 2 December 2025 (UTC)

You've got mail. Grumble. Or I suppose I can paste it into a bunch of sandboxes. —Cryptic 08:14, 2 December 2025 (UTC)

I've replied. Thank you for doing this. I really appreciate it. Clovermoss🍀 (talk) 08:25, 2 December 2025 (UTC)

And re-sent. Remember, you inflicted this on yourself! —Cryptic 08:28, 2 December 2025 (UTC)

Request for log entries containing "Legume" in 2022

Hello, I am trying to answer a research question that requires finding a specific public log entry. Could someone please run a query to find all public log entries (from any log type) that meet the following criteria?

• **Date Range:** From January 1, 2022, to November 30, 2022.
• **Keyword:** The log summary, comment, or description must contain the keyword "Legume".

The target of the log entry is not necessarily the "Legume" page itself. I need the full details of any matching log entry, including the date, time, user, action, target, and the full summary text.

Thank you for your help! ~2025-38095-93 (talk) 00:09, 3 December 2025 (UTC)

Here's 5. It might be case sensitive, so if you need something besides "Legume", such as "legume", let me know. I'm also not sure if the relationship between comment_id and log_comment_id is 1:1 or not, so that could be a possible source of bugs / missed log entries. –Novem Linguae (talk) 04:14, 3 December 2025 (UTC)

comment rows are reusable - hence comment_hash - and you'll see that especially with a few like comment_id 10 (the empty string), or for where e.g. in a move, the same comment is reused in the dummy edits to revision on the moved-from and moved-to pages. In practice, though, I'm amazed that there's enough storage saved by reuse to justify the storage used by the hash column and index. Either way, it wouldn't matter for that query. —Cryptic 11:00, 3 December 2025 (UTC)

Blocks w/ "copyright" in block log

Is it possible to generate a query for blocks / recent (most recent 100?) blocks that mention the word "copyright" in the block log?

(In case it's relevant, I would like to see any copyright blocks that don't get reported at the noticeboards for cleanup purposes and do the fancy thing some people do and put it in {{Database report}}... so like if somebody wants to say why that's an awful idea please feel free.) GreenLipstickLesbian💌🧸 10:20, 28 November 2025 (UTC)

Here are the most recent 100: quarry:query/99310. The IP addresses unfortunately aren't in a very readable form but I've included them for completeness where they exist. You can change the number fetched by tweaking line 7. J11csd (talk) 12:36, 28 November 2025 (UTC)

The human-readable ip is in bt_address. —Cryptic 12:47, 28 November 2025 (UTC)

Oh, so it is! I glanced right over it :-S

Query has been updated to use this field and also to do case-insensitivity (somewhat more) correctly. J11csd (talk) 12:57, 28 November 2025 (UTC)

Some more thoughts:

• The explicit collation is unnecessary; converting to utf8 is enough.
• Probably want to find reasons matching %copyvio% too.
• It's not clear that the block table is the right tool for this; it'll omit expired ones. logging with log_type='block' and log_action!='unblock' will include those. Whether they should be included or not isn't clear. (On the other hand, extracting expiry, whether the block is partial, etc. from log_params is fraught; on yet a third hand, none of that's included in the query anyway.)
• This can probably be made smarter by omitting usernames/ips that already have a matching Wikipedia:Contributor copyright investigations/ subpage linked from Wikipedia:Contributor copyright investigations/Bottom. That'll miss the numbered ones like /20111108 (I take it those are for usernames that match real names? The ones I glanced at were.), but nothing to be done for those.

—Cryptic 13:03, 28 November 2025 (UTC)

Rather embarrassingly, I just realized I forgot to respond to this : thank you @J11csd and @Cryptic, this is exactly what I was dreaming of! Best early xmas present ever!

And re: Cryptic's last point- yes, you guessed correctly, numbered CCIs are primarily for real or real-sounding names, to avoid accidentally harming the real person if somebody Googles them. The number is actually the date the CCI was opened, which, rather unfortunately, is not always the same day as the block. GreenLipstickLesbian💌🧸 11:47, 4 December 2025 (UTC)

Twinned WP–Help page titles

Further information: Wikipedia:Namespace/Help vs. Wikipedia

Resolved

I am looking for a list of page titles which exist in both the Help and Wikipedia namespaces, where neither is a redirect. This arises out of this recent merge proposal regarding Wikipedia:Translation and Help:Translation, with the discussion turning partly on what would be appropriate to each page if they were not merged. I later realized there are other such pairs, such as Wikipedia:Substitution and Help:Substitution where the same question could be asked, and I wondered if there aren't many such WP-Help pairs, where it would be profitable to have a higher level discussion at VPI or somewhere about such twinned pages. Hence, this request.

As for SELECTion columns, page title (without namespace) in col 1, I would say, and then some paired columns with adjacent 'Help' and 'WP' stats, perhaps something like this:

Title, Age-H, Age-W, Watchers-H, Watchers-WP, PgViews30days-H, PgVw30-WP, UniqueUsers-H, UqUsers-W, TotEdits-H, TotEdits-W

or whatever seems useful and not a pain in the neck to produce.

For the WHERE, I thought of exclusions for redirects, but probably there could be others that might be helpful, maybe non-disambig page (Template:Disambiguation), non-stub (Template:Stub), non-essay (Template:Essay), non-set index article (Template:Set index) whatever seems good here.

The tricky part, it seems to me, is how to sort them. It should be something that elicits those pairs that are 'most in need of community attention' at the top, admittedly a very wishy-washy description. Some proxies that come to mind for that are avg page views, avg page watchers, avg total number of unique editors, avg age; and do we take the max instead of avg, or do we just pick stats based on one NS? Maybe try a few sorts, and see how they compare. Anything that elicits 'Translation' and 'Substitution' towards the top are probably good; then again, I don't know what the competition is. Thanks! Mathglot (talk) 20:59, 12 December 2025 (UTC)

Note to self: for VPI, html-ize, add cols for merge/Rfcs, remarks, etc. Mathglot (talk) 21:14, 12 December 2025 (UTC)

The watchlist table isn't in the public replicas for privacy reasons, so we don't have access to watchers at all - the only way to get them is to go to Page Information onwiki for each of them, one by one. And page views isn't in the main database at all; the least painful way is to paste the results onto a page onwiki, feed that into https://pageviews.wmcloud.org/massviews/, and manually paste the results of that back onto the page. But the rest I can do. There's only a few. quarry:query/99985. Do you also want ones where one or the other (or both) is a redirect, but points somewhere besides the corresponding page in the other namespace, and not both them pointing at the same third page. There's 319 of those. —Cryptic 02:16, 13 December 2025 (UTC)

Oh, this is great, thanks! And I am relieved that there aren't a lot more (although this is enough to provide great data for a VPI discussion, and keep us busy afterward). Thank you for the tip about massviews; with output limited to a few dozen, that sounds worth doing once I wikitable-ize it. Your refinement suggestion regarding redirects sounds like a good idea. In that case, do you envision the two pagename columns becoming four, or what would happen? I'm just thinking ahead to the sortable wikitableized version, and making sure we have a column we can sort on to push all the redirects to top or bottom. If you need a test item to peruse, I know that there is a Help:Table, and that WP:TableWP:Table^[noredir] is a redirect to a set index page. If there's 319, can we have that as a separate quarry id, so I can link the one you already have, and that one, too? Mathglot (talk) 02:30, 13 December 2025 (UTC)

Oh, I was suspicious of the rows with only 1 editor, and saw that Wikipedia:Content model for example is a soft redirect. Is it easy enough to exclude those? Otherwise, the '1 editor' is a good enough flag to just ignore them, or I can do it when I reprocess the file. Mathglot (talk) 02:35, 13 December 2025 (UTC)

Actually, never mind that. VPI readers may want to know about those. Mathglot (talk) 02:37, 13 December 2025 (UTC)

quarry:query/99986 for the mismatched redirects. Some of them are simply a product of the previous query (WP:Visual editor and Help:Visual editor point to WP:VisualEditor and Help:VisualEditor respectively); some are distinct cases like Help:National varieties of English being a sort-of-a-disambig while Wikipedia:National varieties of English redirects to Wikipedia:Manual of Style#National varieties of English.

Soft redirects aren't redirects so far as the software's concerned, just pages with a template on them; I could detect them by looking for transclusions of {{soft redirect}}, but wouldn't be able to tell what they soft-redirected to, like I can with normal redirs. —Cryptic 02:44, 13 December 2025 (UTC)

Super useful; thanks for your efforts. Both of these should stimulate some interesting discussion. I've tagged this discussion as 'resolved' at the top. Much appreciated! Mathglot (talk) 03:05, 13 December 2025 (UTC)

For those who are curious/interested in following this further, see Wikipedia:Namespace/Help vs. Wikipedia. Mathglot (talk) 09:15, 13 December 2025 (UTC)

Percentage and raw numbers of article edits made by members of each user group

Hi, I'm posting here after being directed this way by the folks at the village pump. I'm interested in quantifying the amount and fractions of article edits (excluding Talk, policy pages, etc.) performed by each user group: unregistered (anonymous), new accounts, confirmed and autoconfirmed, and extended confirmed.

For the purposes of the query, I would like to exclude edits by bots and those made using scripts. The ideal timeframe for data would be a whole year (preferably calendar year 2025 to date), and historical data for past years if possible. A total tally of edits (as far as there is data available) would also be appreciated.

I believe that this would be a great addition to the pages Wikipedia: Wikipedians and Wikipedia:Who writes Wikipedia?, as well as being a test of my working hypothesis that the vast majority of article edits are performed by a minority more experienced users, as per the Pareto principle.

Thanks in advance for your attention to this. Please let me know what your thoughts are on the feasibility of this request and feel free to suggest alternative or additional criteria or to request any assistance or information that may be needed to procure this data. CVDX (talk) 22:02, 29 December 2025 (UTC)

Not at all feasible. "New accounts" isn't defined; autoconfirmed isn't an actual usergroup, it's computed anew each time it needs to be checked, and that's not scalable to the 60-something million edits that happen each year; even when the group is real, there's no way to tell what groups a user was in at the time they made an edit, and that's significant even for edits in the past year when you're talking about extended confirmed and lower; there's no practical way to exclude bot edits from accounts that were formerly bots, and that becomes very significant for historical data; and excluding script edits is complex, and only possible when a script is either well-established-enough that its edits are tagged, or when it consistently marks its edit summaries (a much smaller minority than you'd think). —Cryptic 22:37, 29 December 2025 (UTC)

Things I can do that come fairly close: distinguish between registered users and ips and temporary accounts; exclude accounts that are currently bot-flagged; bin edits made by registered and temporary accounts (but not ips) by the user's current total edit count, with arbitrary cutoffs - I'd suggest 1 edit, 10, 500, and a couple thousand, to start with; bin edits by registered accounts by time elapsed between registration and the edit, or by absolute registration time if you prefer (which would answer questions like "what percentage of edits are made by accounts that registered before 2010?"). —Cryptic 23:43, 31 December 2025 (UTC)

Pages in a category, but not linked on a corresponding page

Specifically, I'm looking for pages that are in Category:Game Boy Advance games but aren't linked on the List of Game Boy Advance games.

I gave it a try, but my SQL skills are lacking and I got stuck on the inside of the NOT EXISTS part: quarry:query/100504. --Retro (talk | contribs) 19:09, 31 December 2025 (UTC)

You were pretty close. quarry:query/100518. —Cryptic 23:38, 31 December 2025 (UTC)

Appreciate the help! Retro (talk | contribs) 00:33, 1 January 2026 (UTC)

Pages eligible for U6 and transclude Template:Infobox ship begin

Hi! Would it be possible to get a list of pages unambiguously eligible for WP:U6 and also transclude TM:Infobox ship begin? (That is, the user has made no edits outside userspace, the user subpage is more than six months old, it is not a .js/.css subpage, it is not in Category:User namespace book pages, and transcludes TM:Infobox ship begin.)

Boring details of why this is helpful: because these are slated for deletion, there is no point in doing the conversion to {{infobox ship}} per a TFD outcome; better to skip the line and U6 them now. Best, HouseBlaster (talk • he/they) 22:45, 13 January 2026 (UTC)

Only sixteen. quarry:query/101029. —Cryptic 00:07, 14 January 2026 (UTC)

Thanks, Cryptic :) HouseBlaster (talk • he/they) 00:20, 14 January 2026 (UTC)

Statistics of pages which have been speedily deleted and declined as drafts

Hello everyone, recently I have proposed the idea to have a disclaimer on the sign up page when creating an account, you can read the idea and discussion at Wikipedia:Village pump (proposals)#The Wikipedia sign up page disclaimer idea. I am here to ask about statistics of the last month (although I think a couple months would be more favourable) of the amount of pages which have been deleted or denied as applications due to being contrary to Wikipedia, or people trying to write about themselves, family member etc. Specifically deleted under criteria G1, G11, G15, A1, A3, A7 (especially), A9, and A11. These criteria are the ones I think are most important to the overall proposal, and any statistics (like monthly amounts, either showing a rise, steady pace or fall) would help greatly. Thank you all. Mwen Sé Kéyòl Translator-a (talk) 10:11, 17 January 2026 (UTC)

Speedy stats going back to the start of 2023 at quarry:query/101108. AFC declines aren't logged AFAIK and aren't visible once the page is deleted, so there's no direct way to get historical data; you could try looking at Wayback Machine archives of categories like Category:Rejected AfC submissions. —Cryptic 10:54, 17 January 2026 (UTC)

Thanks, I’ll look at those stats you’ve given me. I’ll then compile them and compare. Mwen Sé Kéyòl Translator-a (talk) 11:10, 17 January 2026 (UTC)

Recreated Pages

Hey, would it be possible for somebody to help me with a query that lists all drafts that have been recreated from G5 deletions in the past 45 days or so? I don't know if this is possible but I'd greatly appreciate the help if it is. Thanks! --Seawolf35 ^T--C 02:34, 30 January 2026 (UTC)

Recreated in the last 45 days, or deleted G5 in the last 45 days? —Cryptic 02:44, 30 January 2026 (UTC)

And do you want pages that are currently redlinks, or just currently-existing ones? —Cryptic 02:45, 30 January 2026 (UTC)

@Cryptic Just currently existing ones that have been recreated from G5 deletions. This is for sock hunting as it is painful to go through each and every deletion log manually from the page feed. --Seawolf35 ^T--C 02:48, 30 January 2026 (UTC)

Hrm. Do you want pages that are currently in the draft namespace including pages created elsewhere and then moved to draft (for which I'd go by their earliest currently-not-deleted revision); or pages that were created in the draft namespace and might have been moved elsewhere since then? Only going to be looking at G5 deletions on draft titles either way. —Cryptic 03:00, 30 January 2026 (UTC)

Cryptic: Could you do both? If not then pages created in the draft mainspace. --Seawolf35 ^T--C 03:03, 30 January 2026 (UTC)

I think I misunderstood. Pages currently in the draft namespace. --Seawolf35 ^T--C 03:05, 30 January 2026 (UTC)

quarry:query/101579. —Cryptic 03:12, 30 January 2026 (UTC)

Cryptic Thank you very much! I don't know how people understand all this SQL stuff but I'm glad some people share their knowledge. --Seawolf35 ^T--C 03:15, 30 January 2026 (UTC)

quarry:query/101580 for the other method, looking at the creation log. It finds less since it doesn't see that, for example, Draft:Caleb Alloway started at User:Intermezzo118/sandbox. —Cryptic 03:17, 30 January 2026 (UTC)

About cleaning mass message lists of inactive users

Hey there, I would love it if someone could help me with a query that lists all of the users in a mass message list who have no edits or logged actions in 12 months. Ideally, it might also display the date of the last edit so the list can be sorted by that value. Thanks in advance, and let me know if you have any questions! —TechnoSquirrel69 (sigh) 04:19, 19 January 2026 (UTC)

I can do it for lists that are in Category:MassMessage delivery lists, I suppose. MassMessage can use any page or category as its list of recipients, though, and plenty of the ones I see scrolling past on my watchlist aren't in that category (and don't have their content model set to Mass Message Delivery List either; I think that automatically puts them in the cat?). quarry:query/101151. And you're going to want to get wider consensus than "I personally think this might be a good idea" before you start removing names, too. —Cryptic 05:30, 19 January 2026 (UTC)

Thanks, this is very helpful! I was only planning to clean out one list, not go on a spree with every list in existence. —TechnoSquirrel69 (sigh) 05:54, 19 January 2026 (UTC)

Which? —Cryptic 06:12, 19 January 2026 (UTC)

This one, which is related to an event I just organized (and future events I'm planning). —TechnoSquirrel69 (sigh) 15:47, 19 January 2026 (UTC)

quarry:query/101175 specifically for that page. —Cryptic 16:05, 19 January 2026 (UTC)

How about doing things in reverse, and asking for any mass message list that has not been used in 18 months, and then targeting them for deletion. (I'm baffled to see MM lists specific do 2013-14 still on the books.) I'm aware of several people from other projects who subscribe to certain enwiki-only mass messages because they want to be aware of what's happening here; many actions here have an impact on other projects down the road. Risker (talk) 07:56, 19 January 2026 (UTC)

quarry:query/101159 has all 544 by their last logged use. About 460 have gone unused for more than a year and a half. —Cryptic 09:20, 19 January 2026 (UTC)

@TechnoSquirrel69, the Wikipedia:Feedback request service has a bot that removes inactive users. It could probably be adapted to do the same for other lists. WhatamIdoing (talk) 21:08, 4 February 2026 (UTC)

I appreciate you for letting me know! Taking a closer look, it appears to be user-configurable by list, which is super nice. I know the current operators, so I'll talk to them about it and put it on the lists I'm using if appropriate. —TechnoSquirrel69 (sigh) 04:28, 5 February 2026 (UTC)

You're welcome. I hope it proves useful to you. WhatamIdoing (talk) 04:34, 5 February 2026 (UTC)

AfC submissions by user class

Recently, I opened a discussion in WP:VPIL regarding restrictions on who can make drafts. I believe some statistics on how many drafts are submitted by TAs and unconfirmed accounts may help me. If it is possible, could anyone find the amount of drafts currently awaiting review by user class? CabinetCavers----DEPOSIT OPINION, [valued customer] 15:28, 11 March 2026 (UTC)

RecentChanges shows about 200 still-extant Draft: pages created by TAs during the last week. This will not include any that have been deleted or that have been moved to the mainspace. The same filter for (still?) non-autoconfirmed editors finds about 800 article creations, and around 450 for editors who have made 10–500 edits, and 500 for experienced editors. Adding it up, we're looking at about 2,000 new pages in the Draft: space (not counting those already deleted or moved to the mainspace) per week, of which about 10% are from TAs and 40% from newcomers. That should give you a general idea of what a proper query would be likely to find (e.g., couple thousand articles per week, TAs are a small but non-trivial minority, etc.).

One of the things that would be interesting to know is whether that has changed since the introduction of TAs. For example, is March 2026 different from March 2025? WhatamIdoing (talk) 17:46, 11 March 2026 (UTC)

The creation log will show pages even if they've been deleted, and will show where they were created as opposed to where they are now. The onwiki interface doesn't have any way to filter by namespace or creator, though.

But neither of those is relevant for drafts-awaiting-review, which is different from just "in the draft namespace" - it can include userspace drafts, and of course there's lots of unsubmitted and already-reviewed submissions in Draft:. Finding out who actually pushed the submit button is nontrivial, but who created the draft is easy. quarry:query/103016. Like the section above, this uses the accounts' current edit count, not the count when the draft was created; it does compare the accounts' registration times to the drafts' creation times correctly. —Cryptic 19:49, 11 March 2026 (UTC)

So: of the current drafts awaiting AFC attention, 7% were created by IPs/TAs, 27% were from (current) newbies, 51% were from (now) autoconfirmed editors, and 15% are from (now) extended confirmed. WhatamIdoing (talk) 20:11, 11 March 2026 (UTC)

Articles created by new users

Hi, would it be possible to obtain over a sample period of 1 Jan 2026 to 28 Feb 2026:

• The number of articles created in mainspace (excluding redirects)
• The  % of articles created in mainspace (excluding redirects) by users who had less than 100 edits
• The % of articles created in mainspace (excluding redirects) by users who had less than 100 edits
• The % of articles created in mainspace (excluding redirects) during the sample period t of article creations by editors with less than 100 edits that are tagged as a mobile edit on their first revision

Thanks. Kudpung กุดผึ้ง (talk) 14:54, 9 March 2026 (UTC)

Finding out how many edits a user had at some time in the past is hard. In bulk, it's effectively impossible. I can look for new articles created by users who now have under 100 edits, but that's a very different set even over a two month period. And I suspect this is going to be dwarfed by the number of new pages created in draft anyway. —Cryptic 00:17, 11 March 2026 (UTC)

Would it help to reduce the sample period to 30 days, from the last 30 days for example? For this exercise I'm not interested in drafts, only new pages that are created directly in mainspace or moved there from a user draft or sandbox. Kudpung กุดผึ้ง (talk) 11:12, 11 March 2026 (UTC)

Still isn't going to be what you asked for, but will be a bit closer, at least.

You're asking for pages currently in mainspace, created in the last 30 days, now? That's different, and somewhat harder, then created directly in mainspace (and maybe deleted or somewhere else now). Original request - or at least how I read it - was basically searching Special:Log/create; this is the equivalent of looking at each recently-edited mainspace page's history, so wouldn't be able to find anything that wouldn't. —Cryptic 19:54, 11 March 2026 (UTC)

I didn't think it would make it harder, quite to the contrary in fact, but I do not understand the technology involved in such datamining. To reinforce an upcoming presentation I still need some basic stats. How about simply: The % of articles created in mainspace (excluding redirects) in the last 30 days by editors who still have less than 100 edits Kudpung กุดผึ้ง (talk) 00:42, 12 March 2026 (UTC)

116381 creations; of those, 40379 nonredirects; of those, 3066 (7.59%) nonredirects by users with fewer than 100 edits and 505 (1.25%) nonredirects by mobile edits by users with fewer than 100 edits. Some other combinations of those criteria in the query. —Cryptic 03:59, 12 March 2026 (UTC)

Requesting a list of all in articlespace that are the same except for Capitalization (with no redirects)

This query would return something like

• ORANGE BEAN has Orange Bean, Orange bean and OranGe BeAn where the articles are identical except for capitalized/uncapitalized letters. Exclude all redirects.Naraht (talk) 21:33, 22 March 2026 (UTC)

@Naraht Do you happen to know of an example of such a page?Polygnotus (talk) 21:35, 22 March 2026 (UTC)

There's some examples at WP:DIFFCAPS. —Cryptic 23:12, 22 March 2026 (UTC)

@Sohom Datta Do you happen to know if this is even possible? Its utf8mb3_general_ci (ci meaning case insensitive), right? Polygnotus (talk) 21:40, 22 March 2026 (UTC)

PolygnotusNo, I don't I thought about it in terms of whether a redirect by Alphabetization would *always* be allowed. I'm pretty sure it is possible for user names, since that is something that isn't listed as a listing of usernames.

Sohom Datta As far as I've seen, the only limitation is the first letter needs to be capitalized. Certainly allowed in subpages. I can create User:Naraht/AAA and User:Naraht/AaA separately.Naraht (talk) 21:57, 22 March 2026 (UTC)

No, page titles are stored as uncollated binary data. (And TIL that "utf8" on MariaDB means utf8mb3 by default, not -mb4, which breaks for titles like 𝔹.) —Cryptic 07:57, 23 March 2026 (UTC)

This is wholly impractical by query. It's almost doable by downloading the list of mainspace titles from the monthly database dump and doing a case-insensitive sort on it, but excluding redirects from that is harder. I'll see if I can put something together. It won't be soon. —Cryptic 22:43, 22 March 2026 (UTC)

quarry:query/103518. Data as of about 13:57 5 March 2026. There were 6421 pairs, 43 sets of three, and just one with four (Kink/KinK/KiNK/KINK). Sure hope you had a better reason to ask than just curiosity. —Cryptic 07:40, 23 March 2026 (UTC)

RFC dataset

The bot that manages RFC templates now has a public database, s51043__rfcbot_p. This has enabled queries such as quarry:query/100675. (People interpreting the results should remember that not all entries represent a separate RFC.) WhatamIdoing (talk) 00:31, 30 March 2026 (UTC)

Prevalence of infoboxes in articles

Goal: I want to add a factual statement to MOS:INFOBOXUSE that says something like "As of 2026, n% of non-redirect, non-disambiguation, non-list articles contain an infobox."

This requires knowing:

• the number of pages in the mainspace that aren't redirects (easy), dab pages (easy enough), or lists (difficult?)
• the number of articles that have an infobox.

Module:Infobox is transcluded into nearly 4.4 million mainspace pages. Module:Autotaxobox is transcluded into 0.6 million. I don't know how many other independent/base templates exist.

Anything within about 1% or so is good enough. How much of this do you think you could answer? WhatamIdoing (talk) 17:31, 24 March 2026 (UTC)

The hard (and non-technical) part is finding all/most of the other independent/base templates. Given a list, I can certainly give you a total excluding redirects, disambigs, and "List[s] of..."-titled pages, and not double-counting pages using more than one infobox type. —Cryptic 17:45, 24 March 2026 (UTC)

I've asked at Wikipedia:Village pump (technical)#Finding all the infoboxes. Hopefully, someone will have tried to find them all before, and can easily share the information. Alternatively, they may be able to reassure me that these are the only two that are significant. WhatamIdoing (talk) 17:49, 24 March 2026 (UTC)

Is MediaWiki talk:Common.css/to do#Infobox what you need, or do you think it would be better to work from these search results? WhatamIdoing (talk) 22:54, 24 March 2026 (UTC)

Neither. The search has both false positives and negatives and isn't easily transcribable; Izno's notes to himself aren't legible (or, since it includes search results, transcribable; and maybe not comprehensive either); and my own scheme of walking the Category:Infobox templates tree won't work because, ha ha ha, people categorizing things have no discipline at all. As usual. It's overwhelmed with things like Infobox templates > Infobox templates by country > Place infobox templates by country > South Korea place infobox templates > South Korea subdivision templates > South Korea city templates > Seoul templates > Category:Seoul Metropolitan Subway templates > Template:Incheon Subway and Infobox templates > Science and nature infobox templates > Chemistry infobox templates > Chembox templates > Chembox documentation > Wikipedia chemical data validation > Chembox and Drugbox articles with a broken CheMoBot template > Template:Stdinchicite. Not to mention some 53000 articles. Sorting any of these methods out would take many hours of work that I just don't have. —Cryptic 13:32, 1 April 2026 (UTC)

Okay. What do you need? Is a list like:

• Module:Infobox
• Module:Autotaxobox
• Template:Old-fashioned non-module infobox only used on three pages

useful to you? WhatamIdoing (talk) 17:14, 1 April 2026 (UTC)

Yes. Probably don't need the three-pages one, since you said above you're not after exact numbers. Also ok if they overlap, that is, it won't affect anything if you accidentally include a template that's implemented by some module or other template that's also on the list. —Cryptic 17:18, 1 April 2026 (UTC)

Okay, here are the ones I want you to include:

• Module:Infobox
• Module:Infobox mapframe
• Module:InfoboxImage
• Module:Infobox date field metadata
• Module:Autotaxobox
• Module:Taxobox ranks
• Module:Autovirusbox
• Module:Biota infobox
• Module:Infobox3cols
• Module:Infobox ship
• Module:Infobox gene
• Module:Infobox military conflict
• Template:Military navigation
• Template:Infobox service record
• Template:Chembox
• Template:Starbox begin
• Module:Infobox body of water tracking
• Module:Infobox cabinet members
• Module:Infobox video game awards
• Module:Infobox rune
• Template:Infobox game score
• Module:Routemap
• Module:Solar eclipse
• Template:Reactionbox
• Template:CompetitionRecordTableTop
• Template:MedalTableTop
• Template:MedalTop

I have compiled this list in the belief that:

1. If an article contains multiple infoboxes (e.g., Smallpox contains {{Infobox medical condition}} at the top plus {{Virusbox}} in the ==Causes== section, then it will still only be counted as one article.
2. If one of these templates actually does contain a different one in the list (and, e.g., I just didn't notice), that this will not result in overcounting. (For example, Module:Infobox body of water tracking is probably used only inside a template that also uses Template:Infobox, but I've included it because I'm not sure, and some of them are meta templates used only in infoboxes.)
3. If it contains the line {{#invoke:Template wrapper|wrap|_template = Infobox that means that it will be detected as using Template:Infobox.

If this list is too long or my beliefs are wrong, please squawk. WhatamIdoing (talk) 20:58, 1 April 2026 (UTC)

Your first two beliefs are correct. I'm not familiar with Module:Template wrapper; do the templates it wraps show up as transclusions? That is, in your example, if you put that on a given page and looked at Special:Whatlinkshere/Template:Infobox, would that page show up? —Cryptic 14:40, 2 April 2026 (UTC)

Yes, they do. * Pppery * _{it has begun...} 16:28, 2 April 2026 (UTC)

quarry:query/103977. Analyzes non-redirects in mainspace only. 5232582 of 6660937 non-dab, non-list pages, at 78.5%; plus 12237 of 126707 (9.7%) of non-dab lists, 319 of 375178 of non-list dabs (0.085%), and none of the 348 dab lists. I'm likely misclassifying many, many lists as non-lists in those numbers, and deliberately not including set index articles as either dabs or lists. —Cryptic 03:28, 3 April 2026 (UTC)

Is the following a true statement?

"As of April 2026, approximately 78% of Wikipedia articles (excluding redirects, lists, and disambiguation pages) contain an infobox." WhatamIdoing (talk) 05:59, 3 April 2026 (UTC)

I suppose that undercounting the lists means I should have rounded the other direction (79% instead of 78%, or even 80%, if you think the undercounting of lists is particularly severe). WhatamIdoing (talk) 06:00, 3 April 2026 (UTC)

That's fair. I'm more confident with "77% of non-redirect, nondisambigs". —Cryptic 15:14, 3 April 2026 (UTC)

I'll go with that. Thanks. WhatamIdoing (talk) 16:53, 3 April 2026 (UTC)

All edits made on wiki in a single day

I need a query which will return all edits for a single day (e.g. from 20260405000000 to 20260406000000) with the following columns:

1. Revision id
2. Page title
3. Editor's username
4. Editor's user rights
5. Revision tags

sapphaline (talk) 17:47, 6 April 2026 (UTC)

@Sapphaline Do you mean something like https://quarry.wmcloud.org/query/104105 ? Note that I intentionally chose a wiki with little activity. Polygnotus (talk) 17:56, 6 April 2026 (UTC)

Yes! THX! sapphaline (talk) 17:57, 6 April 2026 (UTC)

How can I add revision timestamp to the columns, by the way? sapphaline (talk) 18:00, 6 April 2026 (UTC)

@Sapphaline https://quarry.wmcloud.org/query/104107 Polygnotus (talk) 18:02, 6 April 2026 (UTC)

You'll want the namespace, too. Example, Wikipedia:Example, Module talk:Example, etc would all show up the same in both those queries. —Cryptic 19:06, 6 April 2026 (UTC)

Then you'd need something like https://quarry.wmcloud.org/query/104113 You can also simply add p.page_namespace but then you'd get the number of the namespace, not the name. Polygnotus (talk) 19:08, 6 April 2026 (UTC)

Articles containing images from a specific uploader

The latest incarnation of a [commons + enwiki] sock has uploaded dozens of files (commons) and placed these files in articles (enwiki) sometimes replacing better images that had existed in them or in ways that violate WP:V here. I'm trying to clean up that mess. Simply mass-revert is not sufficient because many of the affected articles have later edits. And I can't just nuke the images on commons because some are within commons policy and others if they get nuked leave gaps in our articles do to the replacement problem. My thought is to generate a list of enwiki articles that contain any image uploaded by the given account-name. Then if they sock again I can rerun the query with the new account. This approach also protects against their using different of their socks to make the edits vs the uploads. Is this a reasonable type of DB query? DMacks (talk) 23:25, 6 April 2026 (UTC)

@DMacks So something like https://quarry.wmcloud.org/query/104119 ? You need to enter the username where it says "Polygnotus". Polygnotus (talk) 23:29, 6 April 2026 (UTC)

Super. Thanks! DMacks (talk) 00:26, 7 April 2026 (UTC)

If your goal is to get that query down to returning nothing and then nuke the user's creations on Commons, there's an edge case to be aware of; the query looks for the most recent uploader, while Special:Nuke looks for the first.

(And won't it be just as much of a problem if the images are used on wikis other than en, too?) —Cryptic 01:15, 7 April 2026 (UTC)

@DMacks If you only want the first uploader you can use https://quarry.wmcloud.org/query/104123 Polygnotus (talk) 01:38, 7 April 2026 (UTC)

Number of non page-move mainspace redirect pages created by a specific user

I have a request, but it's a doozy. At WP:AN#Iljhgtn I have been trying (and mostly failing) to create a query to accurately count how many redirects the sockpuppet created. So if someone with better knowledge of Quarry would be able to produce a query that does the following, it would be much appreciated:

Given a specific user, return the number of redirects they have created, subject to the following conditions:

1. Only mainspace (namespace 0).
2. Include both live and deleted counts and a total.
3. The page must have been a redirect at the time of the page creation. quarry:query/104250, my query, suffers from false positives by including pages created by the user that were only later made into redirects (e.g. the page was WP:BLARed or merged into another). I discovered this by running quarry:query/104288 (check the first entry). Simply put, the specific user must be the one who redirected the page.
4. The redirect must not have been created from a page move. (This is something I have no idea how to check for.)

SuperPianoMan9167 (talk) 22:38, 11 April 2026 (UTC)

@Cryptic What do you think of https://quarry.wmcloud.org/query/104292 ? Polygnotus (talk) 22:50, 11 April 2026 (UTC)

@Polygnotus Thank you so much! It appears my 24,000 estimate was accurate. SuperPianoMan9167 (talk) 23:26, 11 April 2026 (UTC)

@SuperPianoMan9167 Yeah 24035 is very close. Maybe Cryptic knows a more elegant way to do this. Polygnotus (talk) 23:29, 11 April 2026 (UTC)

First, you'd want to use revision_userindex and archive_userindex in that query; rev_actor and ar_actor will be most selective, and the plain revision and archive views don't index them.

But if you don't mind it missing creations before 2018-06-27, then the best way here is to use the creation log: logging_userindex...WHERE log_type='create' AND log_action='create', joining change_tag on ct_log_id=log_id. —Cryptic 01:07, 12 April 2026 (UTC)

@Cryptic Thanks. That is a much easier approach but it does miss a few: https://quarry.wmcloud.org/query/104295 Polygnotus (talk) 01:13, 12 April 2026 (UTC)

It probably doesn't matter for redirects, and maybe everyone but me already knows this, but for interest, I noticed something about 'rev_parent_id = 0' that confused me for a while the other day. I have a bit of code that implicitly assumed that a page would only have one revision where rev_parent_id = 0, but during testing I came across a page where there were 2, List of schools in Isfahan for blocked page creator Baratiiman, I guess because the user didn't follow the normal draft process and there was a history merge. Sean.hoyland (talk) 05:02, 12 April 2026 (UTC)

Yes, both history merges and splits, as well as other forms of selective undeletion, break the assumption that rev_parent_id=0 === first revision. It happens rarely enough that it can usually be ignored; I only worry about it for queries that are expected to have few or no results. —Cryptic 05:55, 12 April 2026 (UTC)

It's likely the discrepancy between the two queries is because your first one looks for pages that are currently in mainspace (or were there when they were deleted), and the second looks for where they were when they were created. —Cryptic 16:02, 12 April 2026 (UTC)

Usernames containing only non-spacing marks and combining diacritical marks

For context, I need to know how many usernames will be affected if this global title blacklisting request is accepted. The list of non-spacing marks is here and the list of combining diacritical marks is here. I tried querying for ^[\p{Mn}\u{0300}-\u{036F}]*$ on gUser search, but it said MySQL supports neither \u nor \x with more than 2 hex digits. I also tried CONCAT('^[\\p{Mn}', UNHEX('0300'), '-', UNHEX('036f'), ']*$') on Quarry, but UNHEX() also can't handle more than 2 digits at once. NguoiDungKhongDinhDanh 21:28, 17 April 2026 (UTC)

You're looking for \x{...}. quarry:query/104514. (The explicit conversion to utf8mb4 is important.) I haven't verified your regex. Reference. —Cryptic 23:24, 19 April 2026 (UTC)

Thanks. So Quarry queries can use \x{}, unlike gUser. NguoiDungKhongDinhDanh 23:26, 19 April 2026 (UTC)

I just tried gUser - it shows the failed query when it dies. Its problem is it lacks the conversion to utf8, so it's doing a bytewise regex - and multibyte characters are indeed invalid in that context. Might be worth contacting its maintainer; CONVERT() won't give any more of a performance penalty than the regex match already will. —Cryptic 23:33, 19 April 2026 (UTC)

BTW, the reason your try with UNHEX() failed is because it gives a literal byte sequence; you've got to encode that into utf8. Unicode codepoints U+0300 and U+036f in utf8 aren't 03 00 and 03 6f, they're cc 30 and cd af. —Cryptic 23:48, 19 April 2026 (UTC)

Erm, cc 80, not cc 30. A case in point of why you shouldn't try to do this manually, even as a workaround. and for reference. —Cryptic 00:22, 20 April 2026 (UTC)

Thanks again. Those are some very useful information. NguoiDungKhongDinhDanh 23:51, 19 April 2026 (UTC)

Yearly total number of citations

Hey, wouldn't it be nice to see the yearly total number of references of all english Wikipedia articles? For WP:Statistics. In order to get an impression that Wikipedia has become more reliable.

Alternatively yearly total number of good / very good articles.

Or yearly share of paragraphs with at least one citation. WikiPate (talk) 22:36, 19 April 2026 (UTC)

Without a repository of citations (that I know of), I'm not sure how we can query this. Good idea, though. the Stefen 𝕋ower 22:48, 19 April 2026 (UTC)

(edit conflict)References aren't queryable. See the first section here. Closest we can come is looking at the external links, and only in the current versions of pages. (Besides, past a certain point, number of references isn't directly correlated with reliability. See Wikipedia:Citation overkill.)

Historical number of Good Articles isn't queryable either, but what you can do is look at old versions of Wikipedia:Good articles, or Wayback Machine archives of Category:Good articles. —Cryptic 22:54, 19 April 2026 (UTC)

Meta:Shared Citation is the project to address this. ~ 🦝 Shushugah (he/him • talk) 01:11, 20 April 2026 (UTC)

Thank you for the replies.

Yes I know reliablility is not 100% correlated to the number of total references, still citation overkill happens rarely on Wikipedia due to my experience.

The user WhatamIdoing just wrote that there are graphs regarding good and featured articles in Wikipedia:Good article statistics WikiPate (talk) 08:01, 20 April 2026 (UTC)

Optimisation request - disambig pages with link count

I've been looking to try and find disambiguation pages that are no longer needed - they only link out to one or zero other articles. Updating an old quarry query from @Dragons flight I was able to produce this, which returns the number of internal links from each disambiguation page and takes about ten minutes to run. (It turns out there are about 250 pages that are no longer needed, and also that the record-holder is over a thousand outbound links...). I also tried this version to filter it to just ns0 (ie article) links, but that seems to drag on indefinitely.

Is there a more efficient way to do these? Andrew Gray (talk) 12:36, 26 April 2026 (UTC)

There are no indices on page titles anywhere in the schema except when combined with namespace. Your query doesn't just look for pages categorized into Category:All disambiguation pages; it tries to look for pages categorized into 'All disambiguation pages' in every namespace. The optimizer doesn't - and can't - know that A) there's only a limited number of valid namespaces for any page, and B) that no page will ever be categorized into a page not in the Category: namespace. So what it does is look at all categorizations of all pages, and then filters it down afterwards. Add AND catlink.lt_namespace = 14 to your WHERE clause and the first query completes in about a minute. (It'll take longer to show up on the Quarry interface because it has to store and present those 376k results, but it'll still be well under ten minutes.) —Cryptic 13:03, 26 April 2026 (UTC)

Aha, thought these results looked familiar. I also did this query at quarry:query/97491, in a way that's not affected by the cl_to schema change - it looks for the __DISAMBIG__ magic word, typically as part of a template. —Cryptic 13:09, 26 April 2026 (UTC)

Amazing, thank you so much! Andrew Gray (talk) 17:16, 26 April 2026 (UTC)

I've put a report together at Wikipedia:WikiProject Disambiguation/One-link pages and will chip away at it. Andrew Gray (talk) 18:29, 26 April 2026 (UTC)

There'd be a bunch more if not for navboxes, which I see you've also been chipping away at. (I happened to notice New wave reggae today, for instance, which disambiguates two mainspace pages and has links to 259 more in its navboxes.) quarry:query/104722 shows the obvious ones. I don't work with templates much these days, so I don't know how common it is to make navbox templates without using either Module:Navbox or Module:Sidebar. —Cryptic 18:50, 26 April 2026 (UTC)

Yes, the navboxes are a bit confusing - the regiments and string quartets pages all seem to have an intentional navbox of disambiguation pages, which I'm a little confused by. Meanwhile the sidebar in Turing machine (disambiguation) is a load of entries all of which could reasonably be in text on the page. But at least we can deal with the simplest ones! Andrew Gray (talk) 20:39, 26 April 2026 (UTC)

It's not normal for dabs to have navboxes. I've looked at dabs beginning A–G in the report. I fixed a few by adding missing entries or linking unlinked entries. However, most should either redirect to their sole entry or have that entry usurp the dab's title. Thank you for finding these problematic pages. Certes (talk) 22:17, 26 April 2026 (UTC)

Thanks for looking into these! Fifty down now. I agree most of what are left are probably usurpation cases, and those should be straightforward to go through when I have some time to concentrate on it. Andrew Gray (talk) 12:19, 28 April 2026 (UTC)

Thank you. Most need usurpation but there's a significant number like Adams Crossroads and Amei where the linked title is already good and we should just edit the dab to become a redirect (bypassing double redirects for cases like Aetna, Arkansas). I can fix these later if you didn't get there first. Certes (talk) 13:13, 28 April 2026 (UTC)

@Andrew Gray: Thanks for working back from Z. I've done A–L but skipped a couple where I'd like a second opinion please. KNCC could be treated like KORM but we might add Cinema of Kuwait#Kuwait National Cinema Company if the name is current and the text not too promotional to keep. I'm finding no relevant search results for the abbreviation MTGF so the dab might be best deleted. What do you (or anyone else) think? Certes (talk) 11:29, 29 April 2026 (UTC)

KNCC - the article on the cinema company was at Cinescape and was deleted for being spam, but arguably a notable topic despite that so I'd say go ahead there. They still seem to trade under the KNCC name and the other one is just the branding.

MTGF - agree, feels like this would only be worth redirecting if he was known by the abbreviation rather than the name Andrew Gray (talk) 17:36, 29 April 2026 (UTC)

I think that's the non-usurpation cases done. The two newly found zero-link cases seem useless and can be deleted per G14. Certes (talk) 16:35, 30 April 2026 (UTC)

Excellent, that'll give me something to work on over the weekend :) Andrew Gray (talk) 21:38, 30 April 2026 (UTC)

Thanks. There are still a few judgement calls, mostly to do with WP:INITS. For example, I've no idea whether Edwin Peacock or Edwin D. Peacock is his usual form of address, but the article title is probably not critical as long as one redirects to the other. Certes (talk) 22:12, 30 April 2026 (UTC)

Featured infoboxes

Check my math?

This is a follow-on to Wikipedia:Request a query/Archive 6#Prevalence of infoboxes in articles.

1. What percentage of Wikipedia:Featured articles contain an infobox?
2. What percentage of Wikipedia:Featured lists contain an infobox?

From Special:Search using insource:infobox, I get 78% and 32%, but this might be an undercount.

WhatamIdoing (talk) 19:40, 11 May 2026 (UTC)

I get 88% and 35%. —Cryptic 21:03, 11 May 2026 (UTC)

Thanks! WhatamIdoing (talk) 02:54, 12 May 2026 (UTC)

Sample of 10,000 pages without redirections and deleted pages

Hey, when plugging quarry query 80241 or pagepile 60602 into https://pageviews.wmcloud.org/massviews, one gets a list of 10,000 pages. It is possible to get that list without redirections and deleted pages? WikiPate (talk) 15:21, 11 May 2026 (UTC)

quarry:query/105198. —Cryptic 15:32, 11 May 2026 (UTC)

Thank you! Is it also possible to include and sort by the number of page views (for example for the past 30 days) like in https://pageviews.wmcloud.org/massviews? WikiPate (talk) 15:50, 11 May 2026 (UTC)

No. That data isn't available to Quarry, and isn't in a publicly-queryable database at all so far as I'm aware. massviews has an interface to pull titles from a query on Quarry, but I've only used a couple times and none of them recently, and it's currently giving me a "An unknown error occurred when querying Quarry API" error. —Cryptic 17:23, 11 May 2026 (UTC)

And while I was successfully able to turn that into a pagepile, massviews claims "No data found" when I try to use that as a source. —Cryptic 17:27, 11 May 2026 (UTC)

It works with this quarry:

SELECT p.page_title

FROM page p

INNER JOIN revision r ON p.page_latest = r.rev_id

LEFT JOIN page_props pp ON p.page_id = pp.pp_page AND pp.pp_propname = 'disambiguation'

WHERE p.page_namespace = 0

AND p.page_is_redirect = 0

AND p.page_random >= 0.83921

AND pp.pp_page IS NULL

ORDER BY p.page_random ASC

LIMIT 10000;

In the sample are no redirects, disambiguation pages or deleted articles anymore. WikiPate (talk) 15:04, 19 May 2026 (UTC)

okay. quarry 80241 works on my side. After processing for about 30min it gave me a list of 10,000 articles (incl. redirects and deleted pages) with the respective page views of the last 30 days. I saved the query as an csv. The top viewed pages in that list were Apple Network Server (263,559), Google Maps (210,346), David Benioff (154,098), Nick Jonas (116,031) and Kaitlan Collins (105,413).

70% of the articles in the list have max. 1 view per day on average. But there are deleted pages and redirects included. And in the English Wikipedia are more redirects than articles (11.8 million vs. 7.1 million). So the 70% don't say much. WikiPate (talk) 17:54, 11 May 2026 (UTC)

Drawing up a list of users for m:Requests for new languages

Is it possible to draw up a list of user pages on any language Wikipedia that are in a category for language proficiency for a specific language code? Ideally it'd exclude inactive editors, such as those that haven't edited in the past 3 months. With help from editors involved in the proposal to make the notice, I'd like to notify active editors across the wikiverse about an incubated Wikipedia in a language that they speak, in the hope that they may want to help set up and launch that new-language wiki.

For example, there is currently an eligible proposal for a Xiang Wikipedia (ISO 639-3 code hsn); I'd like a list of semi-active editors whose userpages are in Category:User hsn (or any of its sub-categories) on any Wikipedia, so we can then notify about the proposal. I hope that if a query was made for eg. the Xiang categories/templates, I'd be able to copy the query's code and just replace the ISO code, or Wikidata items for its categories, to do this for other proposals? Kowal2701 (talk, contribs) 17:14, 4 June 2026 (UTC)

I requested a script as well but this may be easier Kowal2701 (talk, contribs) 17:21, 4 June 2026 (UTC)

It is not. Different wikis' databases are split among different database servers, and the English Wikipedia in particular is on a server by itself. While there's some support in MariaDB for queries to access data on a different server, it's decidedly nontrivial and we peons don't have the permissions to do it anyway. —Cryptic 17:28, 5 June 2026 (UTC)

And, especially for user-language-categories that are only on a handful of wikis like Category:User hsn, probably the easiest way is to run Petscan queries for each language. Setting those up is straightforward - a few minutes per wiki looked at if you're familiar with Petscan already, maybe an hour lead time total if you're aren't; plus a half hour at most to manually combine the results - while writing a script to do it is a multiple-hour job. —Cryptic 17:40, 5 June 2026 (UTC)

Thank you, tbh I'll just do things manually, I was hoping it'd be relatively simple to technically do Kowal2701 (talk, contribs) 18:33, 5 June 2026 (UTC)

Kowal2701, I was also going to recommend Petscan. If you look at the link in the intro paragraph at Wikipedia:Translators available#Translators available, there is a model Petscan link there that you can use as a starting point. As is, that one is intended to find translators into English, but you can modify it to work in the other direction. However as already pointed out, it only searches the en-wiki categories, so you would have to try an adapted version on other language Wikipedias. Mathglot (talk) 18:51, 5 June 2026 (UTC)

Articles having the most Talk page archives

Draft: https://quarry.wmcloud.org/query/106069

Looking for the articles having the most Talk page archives, as a first step I started off with this Cirrus query which returns the names of Talk pages with over 100 archives (with integer suffixes 100–1000; excluding date-named archives). There are five such: Talk:Donald Trump, Talk:Jesus, Talk:United States, Talk:Israel, and Talk:Race and intelligence. The query results contains those five, but there are also many unnecessary results, because it returns *all* the archives over 100, not just the highest numbered one. (There are also some undesirable results due to redirects that I don't think Cirrus can exclude.)

Next, I vibe-coded Quarry:query/106069 via Chat GPT with numerous twists and turns and some blind alleys. But in order to get something that worked, even if only partially, I had to retain a restriction I don't really want, namely, excluding date-named archives. So for example, if there are articles with lots of archives with Month-YYYY suffixes, they won't be included because the year suffix is > 999. So, two questions:

1. How good/bad is the vibe-coded result? That is, let's say I *did* want to exclude date-named archives, is the resulting SQL decent, or is it an inefficient mess?
2. Is there a way to get an answer to my real question, namely, 'articles having the most Talk page archives', regardless how those archives are named?

Thanks! Mathglot (talk) 21:37, 6 June 2026 (UTC)

1. It's inefficient, but not hugely so, but the main problem is that it doesn't do anything like what you want, only something that looks similar: it just looks for the the page with the highest number and assumes that's the answer. The correct number could be much higher (in case of unusually-named archives, for example Special:Prefixindex/Talk:Sustainability/ or Special:Prefixindex/Talk:Muhammad/) or lower (someone happens to create Talk:Whatever/Archive 948 and nobody notices). Vibe coding in general isn't helpful unless you either know what you're doing or don't have access to someone who does.
2. quarry:query/106075. This still assumes '/Archive ' or '/Archives/' 'rchive' appears somewhere in the page title following a / - there's almost 8500 of those that don't match /Archive[space] or /Archives/. Know any examples of e.g. date-named archives that don't contain either of those? —Cryptic 22:35, 6 June 2026 (UTC)

This is great, and that third column with unusual archives is very helpful, and user-friendly-forward; many thanks! Regarding the question, not sure how we are defining date-named archives, but do any of these fit those contraints? I note that Talk:United Kingdom, for example, has a few oddballs. Mathglot (talk) 00:07, 7 June 2026 (UTC)

The current version of the query includes all subpages containing "rchive" anywhere after a slash, so it'll see almost everything in Special:Prefixindex/Talk:United Kingdom/: all the nonredirects except for /Basic Topics, /FAQ, /GA1 through /GA4, /Terminology, and /to do. (*/archivebox should probably have been filtered out, too.) Everything in your search is counted, as well.

I wrote the initial version of that comment when the query required either "/Archive " or "/Archives/", assuming that there'd be no more than few hundred others; when I actually looked for them, though, I found an additional 8481 pages matching /.*\/.*rchive.*/ but neither of the first two patterns, which is why I updated the query.

Mostly interested in archives that don't contain "rchive". The closest examples I'm aware of are things like WP:Deletion review/Log/2020 January, and none are in main-talkspace. —Cryptic 00:20, 7 June 2026 (UTC)

This is very useful. I have linked it from Template:Archives/doc#When to use (in expl. note d) and from Help:Archiving a talk page#Statistics. See here for links and mentions. Mathglot (talk) 00:00, 8 June 2026 (UTC)

Unimproved Iranian "village" articles

Hi, I'm looking for a search for all of the EN WP mainspace Iranian "village" articles created by USER:Carlossuarez46 with populations lower than 10, 25, 50, 75, and 100 (separate lists for each) with no substantial non-gnoming additions by any other editor. Each article includes the phrase "At the 2006 census, its population was xx, in yy families", for example "At the 2006 census, its population was 44, in 7 families". If a size-cap is needed 2100 bytes should be on the safe side. No page including a photo should be included. Many thanks for any assistance you might be able to offer with this. FOARP (talk) 18:17, 10 June 2026 (UTC)

Queries can't see wikitext, so there's very little of this we can do here. I can find pages started by CS46, I can find pages with no other editor (or no other non-bot editor), I can find pages currently at or under a certain size, I can find pages without images (or without images other than those on a specific list, like, say, those included by maintenance or stub tags likely to be on his articles). I can't sort them by population unless - unlikely - they're already categorized separately by that; I can't check for "non-gnoming" edits even if you could come up with a programmatic definition of that; I certainly can't check for phrases in the pages, other than to confirm that they're not currently redirects.

That leaves you with, well, nothing. There's zero mainspace pages started by this user that have been edited only him and (maybe) bots, that aren't redirects and have no images at all, even without filtering by size. Remove the no-image requirement too and there's 551. This is, as you're aware, an absurd undercount; before I filtered out redirects, there were some 83696 results, and before I restricted to mainspace, there were hundreds of "Populated places in..." categories. —Cryptic 19:07, 10 June 2026 (UTC)

Thanks, looks like I'm going to have to think of another way of handling this. FOARP (talk) 20:44, 10 June 2026 (UTC)

It's better, at 23694, if you define "non-gnoming edit" as "not flagged minor" and expand the no-bots rule to include any username containing "bot" in it (so as to see formerly-flagged bots). Still at quarry:query/106088. —Cryptic 20:46, 10 June 2026 (UTC)

I'll try to see what I can do with that, cheers! FOARP (talk) 21:09, 10 June 2026 (UTC)

You could try PetScan. First download the output from the Quarry query and strip it down to just a list of page names, one per line. Then, in PetScan's "Other sources" tab, paste that list into "Manual list" and add a search for something like "At the 2006 census, its population was" insource:/At the 2006 census, its population was \d{1,2}, in \d{1,2} families/ (In theory, the bit in quotes doesn't affect the search results, but in practice it will probably time out without it.) I got 7161 results. Certes (talk) 22:01, 10 June 2026 (UTC)

Vaguely relevant digression: we also have thousands of articles about Russian places with tiny populations. Ugol, Kharovsky District, Vologda Oblast is a typical example. Many, such as Vlasyevo, have unqualified titles despite not being obvious primary topics. Certes (talk) 22:20, 10 June 2026 (UTC)

Number of registered editors last year

I want to add the 2025 numbers to Wikipedia talk:Wikipedians#Numbers of editors each year. The creation of Wikipedia:Temporary accounts last year means that the old query doesn't return comparable numbers. Can you please give me a new query that does? WhatamIdoing (talk) 03:10, 14 June 2026 (UTC)

I don't know whether you have noticed this, I have only just seen it, but the user view has a tinyint (0 or 1) 'user_is_temp' column with a not null constraint. Sean.hoyland (talk) 06:32, 14 June 2026 (UTC)

Yeah. quarry:query/106283 and quarry:query/106284, eventually. The shiny new ar_actor_deleted index on archive that seems to always take precedence over ar_actor_timestamp might make the latter take ridiculously longer than before, though. —Cryptic 06:46, 14 June 2026 (UTC)

The first says "OperationalError('table resultsets already exists')". The second seems to have worked, and is the total number of registered editors (732,040 in 2025, which is down from 775,435 in 2024, down from 812,635 in 2023). WhatamIdoing (talk) 01:04, 15 June 2026 (UTC)

Eugh. It completed successfully last night after about an hour and a half. I still happen to have the results in command history, since they don't add up to exactly the same as the other query and it took me a while to come up with a plausible reason why.

cohort	COUNT(*)
1	579358
10	124483
100	21898
1000	5687
10000	645
100000	41
1000000	3

(Yes, the same numbers in the 100k and 1m cohorts as in 2025, which is an odd coincidence.) The discrepancy - total 732115 here - is that, as an optimization, the other query only considers users with a user_editcount >= 1, and that field isn't guaranteed to be accurate. Edits imported from other wikis, for example, would create revisions attributed to a user without increasing their editcount here.

I remember seeing problems before where Quarry would complete the query and display results for a while, then start showing errors when it was accessed later. Can't say for sure it was the same error. —Cryptic 01:36, 15 June 2026 (UTC)

Thanks for all this work. I've updated the table and left a note about the discrepancy (75 accounts out of almost 750,000 isn't an important difference, but it's better to document it). The general theme is that we're seeing fewer low-volume editors but keeping the high-volume ones. WhatamIdoing (talk) 23:36, 16 June 2026 (UTC)