Wikipedia talk:Requests for page importation

Latest comment: 5 days ago by Just Step Sideways in topic Request for transwiki-importer - EggRoll97

Backlog of requests from TJB95

edit

@Graham87: what do you think about all the open requests from TJB95? — xaosflux Talk 13:42, 12 August 2021 (UTC)Reply

@Xaosflux: I'm not a fan; see my comment at Wikipedia:Requests for page importation#Import of de:B3 Biennale des bewegten Bildes 2013 to User:TJB95/B3 Biennial of the Moving Image 2013. Draft:2020 B3 Biennial of the Moving Image has been declined. Graham87 15:38, 12 August 2021 (UTC)Reply
@Bogger: any comments here - this is about importing pages for translating, we're usually quite lenient on that process since it isn't direct to an article - but it is a waste if everything will just be deleted. — xaosflux Talk 16:01, 12 August 2021 (UTC)Reply
hmm, I'm on holidays at the moment, but if someone could move the raw text(s) to my sandbox, ("/TodoList" or similar). I'll finish it next week. Bogger (talk) 16:58, 12 August 2021 (UTC)Reply

Your user name

edit

Strikes again... this [1] pointed to me instead of you. It's not malicious -- some editors (or perhaps their offline edit software?) strip your fancy name and turn it into mine.

So... you'll probably miss some of your mail if you keep the two-tone name. Graham (talk) 17:54, 4 September 2023 (UTC)Reply

@Graham: Thanks, but how come you posted here? In this case it was a sockpuppet of a user with lots of editing issues and this mixup probably had nothing to do with my signature (because I didn't sign anything there). Graham87 02:09, 5 September 2023 (UTC)Reply
Well, to be honest, that was my first direct message to another user. Is there a different format that's preferred? Graham (talk) 18:02, 9 September 2023 (UTC)Reply
@Graham: You write messages on people's user talk pages as you did some time back on mine. Graham87 02:41, 10 September 2023 (UTC)Reply
Isn't this your talk page? Graham (talk) 06:05, 10 September 2023 (UTC)Reply
Oh, I see now that it isn't. Sorry about that -- I clearly wasn't paying attention. Graham (talk) 06:07, 10 September 2023 (UTC)Reply

Export request

edit

Is there a similar place to request large transfers from the English Wikipedia to smaller wikis? I want to transfer all the modules with the prefix Location_map/data/ to Mazandarani (mzn) Wikipedia, but there are too many of them. Ταπυροι (گپ) 08:33, 15 August 2024 (UTC)Reply

@محک: Nope. Maybe ask at the technical village pump for assistance. Graham87 (talk) 15:32, 15 August 2024 (UTC)Reply

  You are invited to join the discussion at Wikipedia:Village pump (proposals) § RfC: Enable the mergehistory permission for importers. Chaotic Enby (talk · contribs) 12:30, 20 November 2024 (UTC)Reply

Request for transwiki-importer - EggRoll97

edit

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.


EggRoll97 (t · th · c · del · cross-wiki · SUL · edit counter · pages created (xtools · sigma) · non-automated edits · BLP edits · undos · manual reverts · rollbacks · logs (blocks · rights · moves) · rfar · spi · cci) (assign permissions)(acc · ap · ev · fm · mms · npr · pm · pc · rb · te)

Hello all. I'm here today to request approval from the community for transwiki import rights. This is normally bundled with sysop, though a separate usergroup does exist to allow this for non-sysops. I don't have experience on-wiki to show, unfortunately, since this is generally a sysop permission, and I am not a sysop, though I have messed around with it on a local installation, and have read through Help:Import, the page of the same name on MediaWiki(-wiki?), and the strangely small manual page on MediaWiki about importing.

The short and sweet of how this access differs from importer:

  • Only revisions from pre-specified wikis can be imported.
  • There is no ability to modify the XML file, since an XML file isn't involved in the first place (just a straight copy of the revision history of the origin page).

I of course understand fully if the community believes it best that the importing rights be left to administrators. If not, and the community is okay with this, I gladly volunteer to help out. EggRoll97 (talk) 15:12, 21 November 2024 (UTC)Reply

Notified AN and VPM. EggRoll97 (talk) 15:14, 21 November 2024 (UTC)Reply
I'll be honest, I know very little about importing, but the question that comes to mind is, "Do we need another person to handle imports?" I.e. is there some backlog of import requests that are waiting to be filled and a dearth of people qualified to do the work? I also note that WP:Importers says the transwiki importers group ... is mostly deprecated. RoySmith (talk) 16:12, 21 November 2024 (UTC)Reply
Need is perhaps a harder one to quantify. There's a few requests every so often, and I think Xaosflux's definition of RFPI as quiet at one point still holds fairly true. There were a couple missed requests in the latest archive, though those are a while ago, at this request and this request a year ago with no response to the requests. I think a bigger thing is that RFPI is essentially maintained by two people (xaosflux and Graham87) meaning the bus factor is...quite low, considering that while any administrator can do transwiki imports, no non-importer admin has responded to a request since nearly 4 years ago (JJMC89, in Archive 6). That was a denial of an importation request. The last actual import by anyone other than Graham87 or Xaosflux is also JJMC89, in Archive 5 4.5 years ago. I think it's safe to say there's not exactly a lot of people doing this work. So as to a huge backlog, probably not. As to a dearth of people qualified and interested, I'd say that's more likely, given the most verbose instructions with regards to importing are a userspace page by Graham. EggRoll97 (talk) 17:24, 21 November 2024 (UTC)Reply
I have no idea what happened with the import requests that got missed, honestly. I do *very occasionally* miss things on my watchlist (though I very much try to avoid that). The fact that my watchlist mistake/s caused a lack of response like that should tell you everything, I guess. Graham87 (talk) 04:28, 22 November 2024 (UTC)Reply
Observations from the second-most prolific importer, by a negative margin of ... 10,994. Most of Graham87's imports have been from archives to preserve the earliest creations and revisions that were not captured in the attribution chain. I'm not sure if there is much left to history-merge there anymore.
On import in general, I think import is a little-used, under-advertised and under-understood function that I personally find valuable when I find a reasonably well-composed article in a non-English WP that can be translated with some degree of fidelity. Importing provides attribution to the non-enwiki contributors, rather than a cut-paste transfer that loses attribution, a thing that would be regarded askance within enwiki. I have pretty much entirely used it for things that I'm working on, rather than by request,and I haven't combed the recent logs. It can of course be abused, wittingly or unwittingly (I am hoping I haven't goofed it up in the past), and I really think the option to import transclusions should be turned off by default, given the almost never-used and unlikely reasons for doing that. But I think that it can be handled safely by trusted users - and I don't think anyone has ever argued about Graham87's abilities and competency in that area. I personally think we should be encouraging editors to use and attribute non-enwiki content rather than leaving such things to AI translation; import is an underused tool for that. Maybe I am placing too much emphasis on attribution, but I think it's a blind spot for the project. Acroterion (talk) 17:46, 21 November 2024 (UTC)Reply
I've probably gotten most of the most interesting history out of the Nostalgia Wikipedia by now, but there are still things to do there, largely because when the UseModWiki edits were originally imported in September 2002, the last edit before the software conversion wasn't imported . I'm nowhere near finished with the August 2001 database dump though, which requires XML importing to be used. I'm much less interested in imports of translations and I basically see that as a duty because I'm the entire reason importing was enabled here in the first place. Graham87 (talk) 04:28, 22 November 2024 (UTC)Reply
FWIW, I had an off-wiki chat with a steward regarding the "mostly deprecated" aspect of this. The gist is they don't have a lot of experience handling requests for this, but "in general, we do not override local consensus without seriously good reason". RoySmith (talk) 18:28, 21 November 2024 (UTC)Reply
I'd say it's mostly deprecated because (a) transwiki is normally just handled by admins (b) even when someone asks for xmlimport unless it is really really needed, we (stewards) usually point them getting transwiki sources set up on their project instead. — xaosflux Talk 18:31, 21 November 2024 (UTC)Reply
Question: what do you intend to import? Would you just be taking RFPI requests, or did you have something in particular in mind? It's hard to have any strong opinions in either direction, honestly. WindTempos they (talkcontribs) 18:23, 21 November 2024 (UTC)Reply
I opened this request with a main intention of taking any requests that come through RFPI. I didn't have any particular uses in mind other than helping out there, though I'm definitely open to anyone with other ideas if they're appropriate for uses of the transwiki import tool. EggRoll97 (talk) 23:20, 21 November 2024 (UTC)Reply
There's a longer manual at mw:Help:Import. * Pppery * it has begun... 18:29, 21 November 2024 (UTC)Reply
I did actually link to that one as part of the three pages I've read through. It's a good read, definitely, along with the local version as well, with the key difference being the local version focuses a lot more on xmlimport, and the MediaWiki version has both. EggRoll97 (talk) 23:25, 21 November 2024 (UTC)Reply
(I was thinking of XML importer. The following is still true to a much lesser extent with transwiki importer, but not enough to be that concerning.) It is worth everyone understanding that importer is one of the most dangerous roles in terms of potential for misuse or abuse. It is the only non-admin permission that allows people to modify the database so quickly and so extensively that, in the event of an error or account compromise, it is conceivable that a sysadmin rollback of all recent changes to the wiki would be necessary to fix things. Now, EggRoll97 is a trusted user who holds multiple advanced rights that also have significant misuse/abuse potential, and this is not in any way to say that I distrust them! And in fairness, that database-rollback situation is something any admin can trigger too, although with a bit more effort than an importer, and I'm uh, not going to get into exactly how. So this isn't an oppose, but I think it's important we're all on the same page about the scope of the right being requested. -- Tamzin[cetacean needed] (they|xe) 22:18, 21 November 2024 (UTC)Reply
Note, this request is for transwiki importer, not xml importer. Yes, misuse of transwiki can cause problems, but not to the same extent as xml importing. And yes, we have ~851 admins that can do transwiki right now. I would expect that most of them have little to no understanding of the tool, but also expect that they know better than to launch tools that they don't know how to use. — xaosflux Talk 22:36, 21 November 2024 (UTC)Reply
Which I guess raises the question, "How does one learn how to use import?" EggRoll has at least read the manual and done some experimentation with a local install of the software. That's more experience than I have with import. I'm not meaning to minimize the exposure here, but at the same time, having only two people on the project who know how to do something isn't a good thing either.
I suspect most of those 851 admins don't even know that import exists, and that may explain the lack of misuse of the tool more than self-restraint. That we hand it out to every admin leads me to believe we've just made a mistake on how we bundle permissions. I'd rather we take the bit away from admins by default and give it selectively to trusted people who have expressed a specific interest in it and invested the time to learn how it works, perhaps in conjunction with some one-on-one tutoring from somebody who is already an expert. RoySmith (talk) 23:11, 21 November 2024 (UTC)Reply
By default admins get a very broad set of capabilities, which is one reason a major review point in admin candidates is trustworthiness - including trusting that they know their own technical limits. — xaosflux Talk 23:34, 21 November 2024 (UTC)Reply
Currently, a lot of the "how does one learn" is by doing, either on a local installation, another wiki, or by some other method. I imagine the historical context behind the right being given to all admins is just because all "new, shiny" rights are generally dumped into the sysop toolset by default unless the community unbundles it into a different group. I don't think it's necessarily a problem that all admins have transwiki import by default, though, given it does apparently help at least one admin on a regular basis. EggRoll97 (talk) 23:57, 21 November 2024 (UTC)Reply
it is conceivable that a sysadmin rollback of all recent changes to the wiki would be necessary to fix things -> huh? Even with XML import it should be possible (albeit a pain) for admins to clean up everything IMO. Are you sure you aren't talking about a world where gerrit:967541 hadn't been merged, where it was truly possible for every admin to cause irreversible carnage. * Pppery * it has begun... 22:53, 21 November 2024 (UTC)Reply
I'm referring to the abuse potential of importing a massive number of revisions with noratelimit. At least, I ran my comment by someone smarter than me and they said I was right to characterize things that way. But maybe not, idk. 🤷 But yes, this is significantly less of an issue without XML import involved, I'll grant that. -- Tamzin[cetacean needed] (they|xe) 23:59, 21 November 2024 (UTC)Reply
I will add for reference that I already do possess the technical noratelimit right by virtue of being an account creator, though the take on it being a possible problem isn't a bad take. I don't think it's a significant risk, so long as everyone receiving this right goes through some form of community scrutiny, but I could certainly be wrong about it. EggRoll97 (talk) 00:01, 22 November 2024 (UTC)Reply
My experience has been that even with imports of a few diffs, there is a cost in server effort. It's not the same as flipping a flag when deleting. There is a fairly high failure rate; I assume that other operations are given priority and it times out. Acroterion (talk) 01:16, 22 November 2024 (UTC)Reply
I'm not necessarily sure about the potential for misuse with transwiki importing. I suppose some BEANSy tricks are conceivable, but I would think XML importing would probably be the dangerous one. XML import, by comparison, can (and I don't think stating this one is really spilling any beans, since it's a usecase stated on the import help page) modify usernames and revision history to an incorrect version, which would be difficult to track. I'd say that's really only dangerous because of how heavily the site relies on diffs to detect disruption, but that kind of misuse isn't really possible with transwiki importing. Now, that doesn't mean transwiki import should just be handed out to every autoconfirmed editor or anything to that degree, and those granted this should definitely be trusted, given it is primarily a very technical tool, but I think like Pppery has mentioned, the software has come away from irreversible actions (for the most part, looking at you, protected edit filters, though even those are being technically fixed). EggRoll97 (talk) 23:53, 21 November 2024 (UTC)Reply
Until January 2024 if the revision history contained two revisions with the same timestamp it was impossible by any means to separate them. The common way of doing this was old-style history merges (by delete/undelete rather than MergeHistory), but of course importing could cause it too. I'm fine spilling the beans on that because it was fixed. * Pppery * it has begun... 23:56, 21 November 2024 (UTC)Reply
Yeah, I seem to remember hearing something or other about that, though I don't think I looked too closely into it at the time. EggRoll97 (talk) 00:00, 22 November 2024 (UTC)Reply
When was the last time anybody used xml import? As far as I see, it's a vestigial ability that I suppose might be used in a very particular instance, but I'm not sure what that would be, in preference to simply seeking the addition of a source that isn't already in the transwiki list. Acroterion (talk) 00:33, 22 November 2024 (UTC)Reply
It appears to be used fairly regularly for database dump imports and some Nostalgia Wikipedia imports (see the log) and was last used about two days ago. The last one from RFPI is the beginning of last month to import from the Japanese Wikipedia, which isn't on the interwiki source list. I'll note that adding to the interwiki source list requires a community discussion, a Phabricator ticket, and a patch to the software, so for infrequent or one-off imports from non-added wikis it's likely less of a hassle to just use importupload. EggRoll97 (talk) 04:07, 22 November 2024 (UTC)Reply
I was going to add a pithy comment linking to the log of my last XML import but got edit-conflicted and the above reply is much better than anything I was going to write here. Graham87 (talk) 04:28, 22 November 2024 (UTC)Reply
Support. Transwiki import is not that big of a deal. It's similar to Special:MergeHistory except that instead of moving revisions one page to another, you copy revisions from another wiki to this one. There's a small list of wikis whitelisted for imports on enwiki: meta, nost, de, es, fr, it, pl, outreachwiki, test2wiki, commons. Except for test2wiki, all of these are trusted WMF projects and I don't see issues importing content from them. Just take care to always keep the "Include all templates and transcluded pages" option unchecked – it's probably never useful for enwiki (perhaps we should look into getting the option disabled). – SD0001 (talk) 07:15, 23 November 2024 (UTC)Reply
I don't think a non-admin transwiki importer is a tenable combination to have in enwiki's environment. I just processed a request here, which required a use of Special:MergeHistory and then a deletion in addition to the import, both of which need admin rights. I think this situation, through no fault of EggRoll97's, is probably more common than not, and will lead to law of the instrument problems. * Pppery * it has begun... 02:31, 30 November 2024 (UTC)Reply
Pppery, there are definitely some that need admin rights for at least part of the process (for reference, hist-merging is now not admin-only, as importuploaders can do so), though just looking through Graham's last few transwiki importations, this, this, this, and this, among others, including revision imports, required no admin actions at all outside of the initial transwiki importation. I don't think the law of the instrument necessarily applies to the extent you state though. The transwiki import right is fairly narrow, and while I could go about deletion in a roundabout way to achieve about the same result, it would be a significant betrayal of the community's trust, and I would expect an admin to block me quite quickly for that kind of abuse. EggRoll97 (talk) 06:17, 30 November 2024 (UTC)Reply

A proposal

edit

Let me suggest a sort of traineeship, modeled on how new SPI clerks are trained. Looking at WP:RFPI right now, I see six requests filed over the last month or so, four of which were done and two of which were declined. I'm assuming that's typical. So how about EggRoll watches this page for the next few months and every time a request comes in, they comment on it, explaining how it does or does not meet our policy and recommending whether it be approved or not. Maybe even practice the process by performing the imports into their sandbox wiki. Then one of our experienced importers can say whether or not they agree with EggRoll's analysis. At the end of a few months, we should all have a better feel for how well EggRoll understands the policies and practices around importing and be in a better place to make an informed decision. RoySmith (talk) 23:45, 21 November 2024 (UTC)Reply

I don't think that'd be necessary and would gum up the process, waiting for EggRoll before doing anything. Especially because of time zone issues. Most of the requests are routine (though you do occasionally get unusual ones like the Bengali Wikipedia file-namespace request that's currently up). Maybe I've been too harsh on some of the German Wikipedia self-promoters or newbies over the years and I don't want that to be the final word (and this is where new blood would be good). For what it's worth, I for one would support giving EggRoll the import right now without a training period. Graham87 (talk) 04:28, 22 November 2024 (UTC)Reply
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.