Wikipedia:Bots/Noticeboard

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search
Bots noticeboard

This is a message board for coordinating and discussing bot-related issues on Wikipedia (also including other programs interacting with the MediaWiki software). Although this page is frequented mainly by bot owners, any user is welcome to leave a message or join the discussion here.

For non-urgent issues or bugs with a bot, a message should be left on the bot operator's talk page. If discussion with the operator does not resolve the issue or the problem is urgent and widespread, the problem can be reported by following the steps outlined in WP:BOTISSUE. This is not the place for requests for bot approvals or requesting that tasks be done by a bot. General questions about the MediaWiki software (such as the use of templates, etc.) should be asked at Wikipedia:Village pump (technical).


BOT owners (retired)[edit]

Where would one discuss a useful BOT (User:HotArticlesBot) whose owner seems to have retired (@Kaldari)? I am not currently a BOT operator, but would be willing to, if the choice is to lose a useful bot... (I did leave a note on his talk, and @User:xaosflux does seem to have picked up some runs...so, not urgent. - Mjquinn_id (talk) 15:54, 8 September 2021 (UTC)[]

@Mjquinn id: You can probably email Kaldari to see if he can provide you the necessities for running it. --Izno (talk) 16:03, 8 September 2021 (UTC)[]
And the source is open/published as indicated on the bot's user page, so you could also run it on a bot account of your own after a WP:BRFA. --Izno (talk) 16:04, 8 September 2021 (UTC)[]

Server switch[edit]

SGrabarczuk (WMF) (talk) 01:10, 11 September 2021 (UTC)[]

Using tools like bots[edit]

Should there be a section about using tools like bots, like Halopedia? There might be some people using grammar/spell checkers like LanguageTool or Grammarly that might want to know about using tools like these like bots, such as Halopedia's CIABot or PorpleBot. EthanGaming7640 (talk) 00:56, 14 September 2021 (UTC)[]

Are you looking for WP:MEATBOT? * Pppery * it has begun... 00:59, 14 September 2021 (UTC)[]

Should this job be a bot task?[edit]

Some guidance please on an AWB job which has grown.

For the last few months I have been working on cleaning up WP:Bare URLs. Most of the work is done by Citation bot, to which I feed big batches of articles containing bare urls (via User:BrownHairedGirl/Articles with bare links).

I then follow behind the bot, cleaning up as many as possible of the pages where the bot has made an edit, but not fixed any bare URLs. I use several tools for this, including WP:reFill, which is outdated and has a few vices, such as using the old cite parameter |deadurl=, which is now unsupported and should be converted to |url-status=. Using |deadurl= generates an error message and places the articles in Category:CS1 errors: unsupported parameter.

Rather than fixing this manually, I reckoned it would be faster and more accurate to use AWB to clean up the articles where my use of reFill had created errors.

That worked, and I soon found that it was easiest to just run it on all the pages in Category:CS1 errors: unsupported parameter, cleaning up similar errors created by other editors as well by me. I run it 5 to 10 times per day.

I soon noticed that many of the pages in that category had other simple errors which could be fixed by a regex, so I began adding those regexes. The initially simple AWB setup is now on version 46, with over 40 replace settings. They include mis-spelt and miscapitalised parameters, non-English language parameters which can be translated, and some minor tasks (H:BR fixes, and canonicalisation of some params) which are implemented only if an error has been fixed.

The more I have developed the task, the more pages it has processed. I checked this afternoon and found that it had done 160 edits in the last 7½ days. Should I submit a BRFA for this? --BrownHairedGirl (talk) • (contribs) 16:44, 14 September 2021 (UTC)[]

PS If it helps, I will post a copy of the settings file. Just ping me if you want it. --BrownHairedGirl (talk) • (contribs) 17:03, 14 September 2021 (UTC)[]
FWIW, I used to do thousands of these fixes with an AutoEd script, checking each one for errors before saving. As long as I was not making cosmetic changes, I just marked the edits as minor and kept going. I don't think 160 edits in a week reaches the bot-needed threshold; I would sometimes do that many in a day. The only reason to set up a bot, IMO, would be if you were bothered by having to do these changes manually. Also, if you are looking for more patterns to replace, drop a note on my talk page and I'll link you to my regexes. – Jonesey95 (talk) 18:28, 14 September 2021 (UTC)[]

CW Error #48[edit]

I've be fixing CW Error #48 with Bandersnatch on automatic mode, and have done ~2000 edits, with no incorrect edits that I could see. Is this okay? ― Qwerfjkltalk 17:24, 17 September 2021 (UTC)[]

I looked at your edits, and they look fine to me. Be careful with internal links, which should be trimmed instead of removed entirely. It looks like there are tens of thousands of errors to fix, so you might want to file a BRFA. – Jonesey95 (talk) 22:54, 17 September 2021 (UTC)[]

Semi-automated article creation[edit]

Hello! I am currently working on a project which involves semi-automated mass article creation. At the moment, my tool (CreateTaxonPages) requires some manual setup, and then produces one-line stubs based on data from GBIF and Wikidata (it would also edit/create Wikidata items). I would review the article, potentially add more text, and then publish it. I understand that automated article creation requires a BRFA per the bot policy. However, I am currently proceeding very slowly (mainly to test the tool), and I review each edit before saving it (through my main account). I plan to test it for a while, and then slowly speed up article creation while still manually reviewing each page. If it works well, I plan to create a web tool (WebTaxonArticles) for people to use. I plan to eventually have CreateTaxonPages run automatically through my bot account, and I know I'll need a BRFA for that, but would I need a BRFA before this? Thank you! Tol (talk | contribs) @ 03:01, 20 September 2021 (UTC)[]

I would say yes, a BRFA is required, as that’s bot-like creation of articles. ProcrastinatingReader (talk) 03:05, 20 September 2021 (UTC)[]
That's a very tricky question. The described process, where your tool creates a skeleton article that you then manually review and edit before publishing, would fall under WP:ASSISTED. If you stop giving it adequate human review, even if a human is still pushing the publish button, you may wind up at WP:MEATBOT. I'd recommend that, if you make a tool, it be limited to preapproved people who can be trusted to perform the necessary level of review before hitting publish.
Also of note is that mass creation of taxon articles in particular has long proved controversial, as it turns out to be easy for seemingly-reliable databases to contain errors, outdated information, new taxons that are still in scientific dispute, and/or data that requires intelligent interpretation. See, for example, Wikipedia:Articles for deletion/Anybot's algae articles and various related discussions. IMO any BRFA for creating taxons must be strictly limited in the set of taxons to be created and must require that multiple members of the relevant WikiProject have already pre-reviewed the full list of taxons to be created for accuracy, and should probably also require that the wikiproject discussion was widely advertised to other relevant WikiProjects and possibly WP:VPR as well. Anomie 12:52, 20 September 2021 (UTC)[]
@Tol: I'd strongly urge follow up on part of this prior to even opening a BRFA - that the community will be in general supportive of all these assisted creations; this should have a well attended discussion with a consensus found - how attended and how strong of a consensus should be proportional to how large of a job this will be. — xaosflux Talk 13:23, 20 September 2021 (UTC)[]
@Anomie: I'm not publishing the tool any time soon, though I would probably limit it to trusted users in some way (I was considering just using user groups — perhaps autopatrolled). I'm using GBIF for all data, which is (in my experience) fairly reliable. However, I'm still manually reviewing each article and checking against recent articles if necessary. I'm hoping that manual input and review will mitigate problems such as those found in the AfD discussion you linked. @Xaosflux: Should I request input at the Village Pump before speeding up, or is there another preferred venue? Tol (talk | contribs) @ 20:45, 20 September 2021 (UTC)[]
@Tol: if these are all on the same topic, I'd start with a discussion on the related wikiproject if it has any active members - get some initial feedback, then link in to it from VP. — xaosflux Talk 21:05, 20 September 2021 (UTC)[]
@Xaosflux: I saw this as being fairly wide-ranging. I'm currently doing testing with Phormidium, which would fall under WikiProject Microbiology, but I plan to create all sorts of species. I could try to concentrate in an area for a while after obtaining local consensus, then move on to another area. Tol (talk | contribs) @ 21:17, 20 September 2021 (UTC)[]
Tol, before you get to BRFA, given that many taxon pages on Wikidata were effectively created as a result of Lsjbot's activities, which are known to be, well, bad, 1) what quality guarantee can you give on the point, and 2) if every species is not notable (no really, they aren't), why do you think you should create pages here? Izno (talk) 17:02, 20 September 2021 (UTC)[]
(Ah, I see Anomie/Xaos got it from a current revision. :^) Izno (talk) 17:03, 20 September 2021 (UTC)[]
@Izno: All content is sourced from GBIF. Wikidata is only used to check for existing articles, to match up articles and GBIF IDs, and for populating Template:Taxonbar. Tol (talk | contribs) @ 20:48, 20 September 2021 (UTC)[]
As for notability, we do have Wikipedia:Articles for deletion/Common outcomes#Species. My reasoning is that it's an encyclopedic topic that's generally presumed to be notable, and could be helpful to readers, particularly if additional information is added — though I do believe that a (good) stub is better than nothing. Taxonbars also bring identifiers from Wikidata to a reader-facing article. Tol (talk | contribs) @ 21:05, 20 September 2021 (UTC)[]
@Tol: Wikipedia:Bots/Requests for approval/Qbugbot 2 (a bot to create stubs for insects, spiders, and other arthropods) is a good example to follow (including links to relevant WikiProject discussions, a VPP, and an RfC).   ~ Tom.Reding (talkdgaf)  22:15, 20 September 2021 (UTC)[]
@Tom.Reding: Thanks; I'll take a look at that. Tol (talk | contribs) @ 00:33, 22 September 2021 (UTC)[]

Anyone know a good archive bot?[edit]

I'm making a news page and does anyone know a good archive bot for this? Thanks, Jeb andDinnerbone (talk) 08:49, 21 September 2021 (UTC)[]

Might want to specify what a "news page" is and what the bot is supposed to archive? Jo-Jo Eumerus (talk) 09:00, 21 September 2021 (UTC)[]
Help:Archiving_a_talk_page#Automated_archiving has some info on this. — xaosflux Talk 10:07, 21 September 2021 (UTC)[]

Bots need to upgrade to Pywikibot 6.6.1[edit]

Dear bot operators, bots running Pywikibot must upgrade to version 6.6.1 otherwise they will break when deprecated API parameters are removed. If you have any questions or need help in upgrading, please reach out using one of the Pywikibot communication channels.

Thanks, Legoktm (talk) 18:02, 22 September 2021 (UTC)[]

Bot owner has retired, should the bot still be running?[edit]

HotArticlesBot is operated by Kaldari, who retired on 23 April 2021. This brings up the question as to whether or not the bot should still be running. Editors are responsible for edits made by their bots, and if they are not about to oversee the bot's editing, they can't be responsible, can they?

As I see it, there are two options here, either allow the bot to run as long as it isn't disrupting things, or another editor creates "HotArticlesBot2" to replace HotArticlesBot, which can then be blocked. Mjroots (talk) 08:04, 29 September 2021 (UTC)[]

If I'm allowed to comment here, I'd like to say it's a useful bot and seemingly pretty harmless. It doesn't edit mainspace - just counts edits. I was the one who flagged that the operator had retired, because I was interested in adding its function to a WikiProject. However, if another editor isn't able to actively run the bot or expand its functions to new applications, then I definitely think "HotArticlesBot2" would be useful. Iskandar323 (talk) 08:06, 29 September 2021 (UTC)[]
Ah, a discussion on this was already started here: Bots/Noticeboard#BOT_owners_(retired) Iskandar323 (talk) 08:17, 29 September 2021 (UTC)[]
I think it's ok to let it run until another bot operator takes over the task or until the bot starts creating problems. I see this bot is being run from toolforge (hotarticles account), so the ideal solution would be to ask Kaldari to add a co-maintainer, so that the same bot account can be used and no new setup or BRFA is needed. – SD0001 (talk) 09:47, 29 September 2021 (UTC)[]
  • This operator is far from globally inactive. If someone wants to make a replacement they are welcome to, and I'm quite sure Kaldari would gracefully shut down the old bot if it becomes useless. — xaosflux Talk 10:04, 29 September 2021 (UTC)[]

Just curious, since I don't see a clear indication on his userpage: has anyone just asked Kaldari "will you be able to maintain the bot even though you're retired?" If the answer is yes, this seems like a non-issue. We need more people who can/will maintain bots, regardless of what else they do on-wiki, and there doesn't seem any benefit to rushing to replace the capable and willing. — Rhododendrites talk \\ 18:59, 29 September 2021 (UTC)[]

@Rhododendrites: do you not see the big "Retired" banner at the top of Kaldari's user page? Mjroots (talk) 02:46, 30 September 2021 (UTC)[]
That's ... an odd question. Of course I see it. It's the premise of this thread. It seems obvious that it's worth asking whether or not an active bot operator whose main account is not active will be able to maintain the bot. Not all retirements are equal, and Kaldari's activity on other wikis is one example of that. Beyond that, retirement banners have a funny way of disappearing in time (not always, of course, but it's one of those "indefinite not infinite" sort of things). If someone said "will his bot still run?" you could respond the same way: don't you see the big "retired" banner? Well, yeah, and yet he didn't stop his bot. — Rhododendrites talk \\ 03:07, 30 September 2021 (UTC)[]
I'm not interested in maintaining the bot, but I would be happy to add someone else as a maintainer on the bot account if someone wants to volunteer. Kaldari (talk) 23:18, 1 October 2021 (UTC)[]
@Kaldari You can add me, assuming no one else is volunteering. – SD0001 (talk) 08:15, 2 October 2021 (UTC)[]
I've added SD0001. Kaldari (talk) 02:00, 9 October 2021 (UTC)[]

User:RMCD bot has not run after 30 minutes of Requested_move posted to Talk[edit]

Talk:Christ_myth_theory#Requested_move_1_October_2021 has not been listed on Wikipedia:Requested moves/Current discussions. The template states that this will occur after 30 minutes of the posting. --2db (talk) 16:05, 1 October 2021 (UTC)[]

The bot is down because its operator was blindsided by the September 30 Let's Encrypt root expiry and is still trying to get up to speed and figure out what the hell he needs to do to get the bot back up. – wbm1058 (talk) 21:26, 1 October 2021 (UTC)[]
See WP:Village pump (technical)#Heads up for possible tool breakage todaywbm1058 (talk) 00:48, 2 October 2021 (UTC)[]
Fixed – See also User talk:RMCD bot#‎Bot not working?wbm1058 (talk) 04:07, 2 October 2021 (UTC)[]

Shared accounts policy[edit]

I had a query on my talk page and figure it's worth just clearing up in policy. WP:SHAREDACCOUNT says Exceptions to this rule can be made for ... bot accounts that are maintained by more than one contributor, provided the existence of such an arrangement is made clear and has consensus. This is in the case of a bot account being ran on a server where another individual has technical access to the credentials (perhaps because they're a sysadmin or some similar reason). I've got some of my own thoughts on this, expressed at my talk, but I would say that we don't really enforce any such policy in practice. Many bot accounts have multiple individuals with access to the underlying server (all Toolforge root users, for example, and co-maintainers). Plus I imagine it's more common than disclosed, and even in disclosed cases I'm not aware of any process to gain "consensus" for such an arrangement, or even rubber-stamping by BAG. Does this policy need to be changed to reflect actual practice? ProcrastinatingReader (talk) 10:28, 8 October 2021 (UTC)[]

@ProcrastinatingReader: well, more than that - we have some bots that literally have shared access, by design, and it isn't about because someone may have access to the underlying server. I suppose policy-wise, if the bot account was used to do something bad and none of the operators would take direct responsibility we would could hold all of the operators responsible. — xaosflux Talk 10:32, 8 October 2021 (UTC)[]
And that is what I think that policy is really talking about, the sysadmin of a hosting company thing isn't really what's going on there. — xaosflux Talk 10:34, 8 October 2021 (UTC)[]
I suppose when I say "sysadmin" in this context, I mean a sysadmin who also has their own Wikipedia account (rather than unidentifiable DigitalOcean sysadmins), so more like Toolforge root users/co-maintainers, or if you were the sysadmin for the server hosting ProcBot (of which I'm the coder & listed operator). It's probably not entirely uncommon to have a non-operator involved in the hosting process (especially if the bot is complex). Of course, this individual will likely not be logging into the account and maybe not even editing the bot code, even though they have technical access to view the login credentials. However this case should be treat moving forwards, I don't think the status quo involves any kind of "consensus" process, so don't think the current policy reflects actual practice. ProcrastinatingReader (talk) 10:42, 8 October 2021 (UTC)[]
Two-part addition: (1) if this isn't referring to underlying access to view credentials at all, I think it should be better clarified what it is referring to (since I really have no idea - maybe push access but I can think of exceptions, like Citation bot, that don't adhere to such a rule), and maybe moved into the bot policy as a more natural place for such a provision. (2) If (1) is true, is there a special case for accounts that have privileged viewing access (eg: admin - regarding deleted material; EFM/EFH - regarding private filters). ProcrastinatingReader (talk) 10:51, 8 October 2021 (UTC)[]
I think the key phrase is "maintained by". To speak specifically to this case, TNT isn't maintaining or running the bot, even though it is technically possible. If something were to go wrong, Tamzin would be the one contacted, and only Tamzin. More generally, I do not think we need to update or clarify anything in the wording as it stands. Primefac (talk) 10:59, 8 October 2021 (UTC)[]
There was the similar Wikipedia:Bots/Requests for approval/DeltaQuadBot 9, though that was voluntary and since SQL was removed from access that case is probably largely academic at this point. ProcrastinatingReader (talk) 11:11, 8 October 2021 (UTC)[]
Whether or not a downstream sysadmin has a wiki account or not shouldn't be a consideration. — xaosflux Talk 13:00, 8 October 2021 (UTC)[]
The and has consensus text should be removed. No "consensus" is needed for a botop to add a co-maintainer. – SD0001 (talk) 11:05, 8 October 2021 (UTC)[]
I would (weakly) argue that consensus is needed, if we were to use DQBot as an example - folks specifically didn't want SQL to have access. There's no downside to leaving it in, as per SILENCE consensus is assumed unless someone complains. Primefac (talk) 13:22, 8 October 2021 (UTC)[]
In that sense any action requires consensus. Consensus-by-default (i.e. WP:SILENCE) may well be a decent pragmatic position, but it's straying far from the definition of 'consensus' and I think is confusing to have in policy. If others don't find the prose confusing then perhaps it's just how I'm parsing it, but I think it's more of a case of 'nobody knows that provision even existed' rather than 'it makes sense to most people'. ProcrastinatingReader (talk) 10:18, 10 October 2021 (UTC)[]
So here is an example: Wikipedia:Bots/Requests for approval/OutreachDashboardBot - there are 2 of us that have explicit access to this account's credentials. Either of us would be expected to deal with actions taken under the bot account. It also runs on a cloud services server - but I wouldn't expect those server admins to answer to the enwiki community about that bot. — xaosflux Talk 12:58, 8 October 2021 (UTC)[]
I would like to see us encourage having more bots with multiple maintainers, it's one of the best ways to ensure sustainability. See also wikitech:User:BryanDavis/Developing_community_norms_for_critical_bots_and_tools#Have_multiple_maintainers. Legoktm (talk) 22:04, 10 October 2021 (UTC)[]


user:AnomieBOT and "rescuing" bad population updates[edit]

Here is a user who is doing half-baked updates to many US city articles: [2]. They've done this to dozens of articles.

Usually, this users' updates change the population in one table in the article. They don't remove the old population estimates (probably from 2019). They don't update other stats, like population density numbers, don't update other mentions of the population in the article, and don't update text like "as of the 2010 Census".

Problem is, AnomieBOT comes along and "rescues" the deleted reference definition that caused a visible referencing error. After that, the article has no visible referencing errors -- but is all screwed-up because of the internal inconsistencies and the half-done statistics update.

To make a correct fix, a human has to undo the robot's fix and actually remove the old estimate numbers. Then, edit the article to clean up the density problems, fix the "as of" dates, and verify a few more things before the article is good again.

Trouble is, once AuomieBOT has touched the article, it doesn't have any visible errors anymore and doesn't appear in a category like and instead has to be manually discovered. I think that user:AnomieBOT is doing more harm than good by burying problematic edits while claiming to "rescue" them.

Can AnomieBOT be stopped? Can we have it undo these changes so that the articles are added to the broken references category? -- Mikeblas (talk) 14:41, 8 October 2021 (UTC)[]

I'm not seeing the issue here. For example, looking at Brookhaven, Pennsylvania, a user orphans a reference, which the bot then rescues. You come along and remove the content entirely before replacing it with entirely different content.
First off, you shouldn't be using orphaned reference categories to be making content changes. Second off, the data present in the infobox at the time when the bot fixed the orphaned reference is consistent with the provided data. It's not like the bot put a reference for the wrong content. Third, you should be bringing such matters to the bot operator before going above their heads and bringing it up here.
In summary, this is not an issue that BAG or Anomie needs to do anything about. Primefac (talk) 14:49, 8 October 2021 (UTC)[]
I'll ask you to assume good faith and make an effort to see the issue here. When your first step is to dismiss a concern raised by someone else, then you're not assuming good faith. It comes off as gatekeeping, and not conducive to constructive conversation or collaboration.
The orphaned reference is caused by an editor who (not surprisingly) doesn't understand the side-effects of the template they're editing. Since they remove the reference for the estimate but leave the estimate number, and that's what causes the reference invocation becomes orphaned. Thus, it's not I who removed content -- it was original edit , which removed "estimate=" and "estyear=" parameters from the population history box.
I can utilize the categories in any way I'd like -- thanks, though. The reference that the bot fixes is inconsistent with the first user's intent, which was to remove the population estimate and update it with a decennial census population count.
I've repeatedly brought issues to this bot's owner, and they're also immediately notified whenever I revert the bot's edits. I don't recall that mechanism ever being effective, and don't recall ever receiving a response.
Are you speaking on behalf of the BAG when you explain that this wide-spread and counter-productive editing pattern is acceptable? -- Mikeblas (talk) 15:03, 8 October 2021 (UTC)[]
I don't know why when I explain exactly why a bot edit is not an actual issue with the bot, people assume I'm assuming bad faith; I'm not, I'm explaining why it's not a bot issue. I try to avoid superlatives and making people feel better about their concerns because quite honestly I don't like typing out more than I have to (i.e. short-and-to-the-point is my preference).
The bot does not comprehend content changes. It comprehends that someone has orphaned a ref, which is "bad", and fixing it is "good". If the editor who removed the named reference didn't do a good enough job cleaning up after themselves, the bot helps out. This happens with orphaned ref fixer, dated template parameters, template substitution, etc (plus other automated bots like SineBot). If the content was also supposed to be removed, but the user did not remove that content, the bot has zero, absolutely ZERO, way of knowing that. All it sees is an orphaned reference that needs fixing.
So again, we're not going to disapprove a bot that works (technically) correctly 100% of the time, but is "an annoyance" when the user causing the bot to be triggered does something silly. That's called a GIGO problem, and (at least in this instance) cannot be fixed on the bot's end. Primefac (talk) 15:23, 8 October 2021 (UTC)[]
Step one in this process is to cut a hole in the box should have been to speak with Cameron8782 directly, to explain how their edits are suboptimal and to ask them to improve their editing process. If the bot is getting involved because a human editor is making mistakes, the best way to fix the problem is for the human editor to stop making those mistakes. – Jonesey95 (talk) 15:46, 8 October 2021 (UTC)[]
Good point, forgot that step. Primefac (talk) 15:55, 8 October 2021 (UTC)[]

Is it possible to block a bot from creating pages?[edit]

See User_talk:Bot1058#Talk:Basarabka_and_Moldovanka. The bot has been creating redirects in the talk namespace. Most are benign but just utterly useless blank pages, while several others are actually pages that have been properly deleted per WP:G8. The bot's task listing shows no tasks that would indicate that it should be creating pages. I'm loathe to pblock the bot from the talk: namespace as it actually does useful tasks there, but there's no reason why this should be creating pages, especially ones that have been G8'd. Hog Farm Talk 23:43, 9 October 2021 (UTC)[]

I've left the bot operator a note pointing them at this discussion Face-smile.svg this is likely something that they can fix ~TNT (she/her • talk) 23:48, 9 October 2021 (UTC)[]
The bot has made over 58,000 edits in talk namespace, only 237 of which were page creations, and 204 of those page creations were redirects. Of the 33 page creations that were not redirects, just 15 were blank pages. The lack of an edit summary indicates that the bot thought it was making null edits. Unfortunately null edits are necessary because of limitations of the MediaWiki software. I can't confirm what happened as I only keep the log of the most recent bot run on my machine, but likely these deleted pages were still populating a category that the bot patrols. I think I can patch my code to work around this by double-checking to make sure the pages I null-edit actually have content on them before editing, but as I've been taken to this noticeboard and threatened with a block I've taken tasks 3, 4 and 5 down as a precaution until I get around to fixing this. – wbm1058 (talk) 02:42, 10 October 2021 (UTC)[]
Apologies for coming across harshly, I didn't mean to. I was just a bit aggravated when I got a ping that a bot that isn't intended to create pages went and recreated something that I'd just deleted. I think the bot does good work, and would hate to see if down because I got testy. If all else fails, would it be possible for the bot to log pages it creates and then have someone go through and monitor to make sure they aren't restoring deleted pages? Hog Farm Talk 02:55, 10 October 2021 (UTC)[]
@Wbm1058 You can use nocreate=1 flag in your edit API call. That would prevent creation of the page if it doesn't already exist. – SD0001 (talk) 07:13, 10 October 2021 (UTC)[]
Thanks for the tip. I do recall running into this issue before and dealing with it relatively quietly. I think I was advised that another option was to use "purge" mode rather than "edit" mode. Noting that it's crept up for some reason. Of the 15 blank pages, just one happened in 2018, one in 2019, two in 2020 and eleven so far in 2021. Obviously I can't keep this on my back-burner of low-priority issues any longer, especially since the fix should be relatively easy. – wbm1058 (talk) 12:58, 10 October 2021 (UTC)[]
Heh, I just noticed that my inherited bot framework (previous authors are all retired or dead) has a function purgeCache in it! Which I've never used until now. I'm gonna run some trials to check out how well it works for making null edits. That would be easier than adding a new argument nocreate to function edit. – wbm1058 (talk) 16:36, 10 October 2021 (UTC)[]
Looking at the Purge API, I don't see any reason why I wouldn't always want to use forcerecursivelinkupdate – unless that's a really expensive operation? wbm1058 (talk) 19:47, 10 October 2021 (UTC)[]
IIRC, it puts all the transclusions onto the job queue for updating, much like a null edit. forcelinkupdate updates the links table entries for the current page (e.g. category membership), but doesn't queue up to reparse all pages transcluding the purged page. The operation without either of those parameters just updates the cached HTML without updating links tables. Anomie 00:26, 11 October 2021 (UTC)[]
Thanks. I'm still working on this. Testing just got a "mustbeposted" error: "The \"purge\" module requires a POST request." Assume this was something that was broken by previous API changes that nobody noticed previously because the framework's purgeCache function wasn't being used by anything. – wbm1058 (talk) 19:39, 11 October 2021 (UTC)[]
Yes, that would be the change that made it so there was a 'do you actually want to purge a page' interstitial when you purge a page, some several years ago now. Izno (talk) 21:30, 11 October 2021 (UTC)[]
Fixed – the new bot version uses the updated botclasses.phpwbm1058 (talk) 20:42, 12 October 2021 (UTC)[]

Worst case solution: I have for some time blocked COIBot from creating pages in mainspace (which it should not do but did) until I figured out why it did that sometimes by creating an edit filter for it. That both resolves the problem ánd logs that the problem occurred while you debug/test/patch. --Dirk Beetstra T C 12:03, 10 October 2021 (UTC)[]

Agree, this is a bad, non-scalable solution - out of control bots should just be blocked, not use up valuable abusefilter conditions. — xaosflux Talk 19:28, 10 October 2021 (UTC)[]

Updating the guidelines at this noticeboard?[edit]

We have now had two posts in fairly quick succession where the original editor has not contacted the bot operator prior or even left them a note that there was a BOTN discussion open about their bot. Both of the issues likely would have been resolved at that stage and thus not needed to waste the time of those at this noticeboard who have to see a notification about (essentially) a non-issue.

Now, I am not going to fault either of these editors for doing this, because the entirety of our banner guideline regarding the issue is If you want to report an issue or bug with a specific bot, follow the steps outlined in WP:BOTISSUE first. While I do not think we need to go to the extremes of AN or ANI where the bot operator must be notified, I think we need to make it more clear that the bot operator should be contacted first, and we should really only get a post here if the botop is unresponsive or it's an issue that needs more immediate attention.

If I'm in the minority here, that's fine, I just see little point to "raising alarm bells" for issues that (so far) can be easily fixed somewhere else. Primefac (talk) 11:30, 10 October 2021 (UTC)[]

Given the volume at this noticeboard, I don't think it's a problem in the sense of the board being overwhelmed, but it does tend to offend botops (almost every time I've seen a botop be 'dragged' here, or an issue being reported here in the first instance, the botop has [understandably] taken it negatively). So I think your idea, to clarify the guidelines here, is a good one. ProcrastinatingReader (talk) 11:43, 10 October 2021 (UTC)[]
Indeed, it's not a volume issue, it's comments like ...as I've been taken to this noticeboard and threatened with a block... coming from the botops that has me concerned. Primefac (talk) 11:45, 10 October 2021 (UTC)[]
Yes, I don't want to make too big of a deal about this but you know the reputation this board has after certain bot-ops have been taken here repeatedly and failed to "get the message" even after they were desysopped. I too get just a bit aggravated when I see relatively new admins who sailed through their "ordeals" doing this – it's easier to give a pass to less-experienced editors. At the same time the community is having another village chat over how "toxic" the RfA process is. So sorry if I raised the drama bar a tad too much. – wbm1058 (talk) 12:36, 10 October 2021 (UTC)[]
I'm not trying to make a huge deal about either post specifically, but two in the same week (combined with your comment) made me realise our guidance isn't that great; that's what I'm looking to update. Primefac (talk) 12:42, 10 October 2021 (UTC)[]
I left the noticeboard notification on the bot's talk page, not the operator's, as 1) I assumed that since that page existed and did not redirect to the operator's talk page, that it was likely where bot-specific communications should go and 2) when I've had bot questions in the past for other bots and there was a separate user talk page for the bot, I've gotten replies after posting it on the bot talk page. I guess what confuses me with this whole situation is that this doesn't count as notifying the bot op? As to the block statement, I simply stated that I was going to p-block the bot from the namespace in which it was malfunctioning, which is what admins are told to do, especially since the bot has been doing these things for almost 4 years. Is there a better thing to do in that case, or should you just let the bot continue to recreate deleted pages? Hog Farm Talk 13:55, 10 October 2021 (UTC)[]
My fault for not checking the bot's page, as I assumed TNT's post on your thread meant that no message was left at all.* We can't necessarily require bot operators to always be responsive within 48 hours, but after that timeframe it's not unreasonable to make a post with more eyes on it.
That being said, I still think we need to make that "talk to the operator or post on the bot's talk page" message a little more obvious (instead of being placed on a secondary page with big blocks of text to wade through), with potentially a "please include diffs of botop notification" request, just to avoid any future misunderstandings. Primefac (talk) 14:01, 10 October 2021 (UTC) *I'm really off my game today, I thought the bot post was made 8 Oct not 9 Oct. I think it's time for me to log off and catch my flight now... Primefac (talk) 14:31, 10 October 2021 (UTC)[]
Notices were placed five minutes apart. My biggest issue with virtually simultaneous postings like this is that they place an extra burden on me of needing to respond in two places. Notices placed on bot's talk are preferred and if you feel it's particularly urgent you can ping the bot operator when you post to their bot's talk. Otherwise I may not see it until I check my email and I'm not constantly monitoring my email because it mostly gets spam (which is why I don't do email on my phone). – wbm1058 (talk) 14:21, 10 October 2021 (UTC)[]
Yes, it was wrong for me to take it here so quickly. Hog Farm Talk 15:00, 10 October 2021 (UTC)[]
I think this would be a good idea. It's largely just summarizing the BOTISSUE text which suggests reaching out on talk first, so I don't think it's too controversial a change. I made an attempt at an updated wording, Special:Diff/1049258395, so feedback on that would be welcome. Wug·a·po·des 20:07, 10 October 2021 (UTC)[]

General query[edit]

Last week, two bots that I check on, AdminStatsBot and BernsteinBot, both went off schedule. I checked to see if there was a lag which there wasn't. BernsteinBot started updating but irregularly and it didn't return the regular schedule it had previously maintained with no problems. I posted a note to each bot operator on their user talk pages but neither have been active here in over a month.

But I'm posting here just to see if there was some change or update that would cause both bots to stop updating. Other bots I work with such as AnomeBotIII and SDZeroBot didn't have problems so I'm not sure what is up. Thanks for any information you can provide. Liz Read! Talk! 04:47, 12 October 2021 (UTC)[]

FWIW, YiFeiBot stopped editing a little over four days ago as well (just after midnight UTC on the morning of 8 October). I have posted a note to the bot's talk page, but I thought I would mention it here as well. – Jonesey95 (talk) 05:15, 12 October 2021 (UTC)[]
Yes, it seems like something changed on October 7/8. Liz Read! Talk! 05:49, 12 October 2021 (UTC)[]
@Jonesey95 & User:Liz - I imagine this has been caused by the removal of some deprecated API parameters for obtaining tokens (see phab:T280806). There's nothing really we can do, the bot code will need updating. firefly ( t · c ) 06:11, 12 October 2021 (UTC)[]
Yes, this created problems in gadgets like LintHint too [3]. ಮಲ್ನಾಡಾಚ್ ಕೊಂಕ್ಣೊ (talk) 07:37, 12 October 2021 (UTC)[]
Yes this. According to WP:BAM, Lowercase sigmabot III, ClueBot III and Mathbot are also down since 7 October. – SD0001 (talk) 09:20, 12 October 2021 (UTC)[]
At least one bot was fixed by updating pywikibot, which doesn't sound too tricky to this non-botop. – Jonesey95 (talk) 13:59, 12 October 2021 (UTC)[]
Yes, as indicated at mediawikiwiki:MediaWiki 1.37/Deprecation of legacy API token parameters#Pywikibot, updating it to 6.6.1 or higher fixes the issue in PWB. --Izno (talk) 14:21, 12 October 2021 (UTC)[]
And you know, #Bots need to upgrade to Pywikibot 6.6.1 right above on this page :) Legoktm (talk) 17:16, 12 October 2021 (UTC)[]
BernsteinBot should mostly be back, each report is kind of handled individually so if specific reports are not updating I can take a look. And I've sent a pull request for Lowercase sigmabot III. Legoktm (talk) 17:18, 12 October 2021 (UTC)[]
Thanks everyone for all of the info on this. I rely on BernsteinBot and the bot was doing some updating, just not always the report I rely on. As for AdminStatsBot, I thought it was responsible not only for AdminStats page but for the templates many admins have on their User pages but the template is updating so I guess I was wrong about that. Liz Read! Talk! 21:40, 12 October 2021 (UTC)[]
I looked into AdminStatsBot, it's using an old version of wikitools, which should be straightforward to update, except the code has no license associated with it (I filed T293173 for it on the Toolforge side). Template:Adminstats is updated by Cyberbot I, which seems to be working fine. Legoktm (talk) 05:17, 13 October 2021 (UTC)[]