Wikipedia:Bots/Requests for approval/Yobot 27

The following discussion is an archived debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA. The result of the discussion was

Denied.

Yobot 27

Operator: Magioladitis (talk · contribs · SUL · edit count · logs · page moves · block log · rights log · ANI search)

Time filed: 10:38, Saturday, December 3, 2016 (UTC)

Automatic, Supervised, or Manual: Automatic

Programming language(s): AWB / WPCleaner

Source code available: Yes

Function overview: Replace ISBN magiclinks with ISBN template

Links to relevant discussions (where appropriate): Here you are:

Edit period(s): One off

Estimated number of pages affected: Roughly ~~275k+~~ 350,000

Exclusion compliant (Yes/No): YesYes

Already has a bot flag (Yes/No): Yes

Function details: ISBN 123456789-0 → ISBN 123456789-0 Parameter error in {{ISBN}}: checksum

Discussion

@Legoktm, Bgwhite, and MZMcBride: -- Magioladitis (talk) 12:26, 3 December 2016 (UTC)[reply]

Will you only be changing if it is currently actually a magic link? (As opposed to free text replacement). — xaosflux ^Talk 15:27, 3 December 2016 (UTC)[reply]

Xaosflux Yes and no. Free texts are almost 0 in fact. Since we detecte them via WP:CHECKWIKI project and fix them regularly. So, in fact one way or another there should be no free texts and no ISBN templates with errors. Me and many other will fix any wrong ISBNs. -- Magioladitis (talk) 15:51, 3 December 2016 (UTC)[reply]

I'll add a check for maximum length though so I won't break anything. -- Magioladitis (talk) 15:54, 3 December 2016 (UTC)[reply]

@Magioladitis: For example, your example above in nowiki tags - will this be ignored? — xaosflux ^Talk 16:51, 3 December 2016 (UTC)[reply]

@Xaosflux: everything in nowiki tags will be ignored. Less than 10 pages currently though. Check Wikipedia:CHECKWIKI/WPC 069 dump. -- Magioladitis (talk) 19:33, 3 December 2016 (UTC)[reply]

This is a big change. I support it, but I don't think there is en.WP consensus to make this change yet. Also, do you want to do RFC and PMID, or should that be a separate discussion? – Jonesey95 (talk) 16:36, 3 December 2016 (UTC)[reply]
- Magic cleanup could all be one task or per-magic tasks; but they would all fall under the same community standard. — xaosflux ^Talk 16:51, 3 December 2016 (UTC)[reply]
- Jonesey hit the nail on the head.
  1. PMID and RFC should be included. Get everything done in one pass.
  2. WMF doesn't exactly know what the final design will be. Parser function? Template? Interwiki? Combination? Need to wait for that.
  3. If WMF is going to give each project some leeway, then a discussion should be had to get consensus.
Disappearance of some magic links. Change from Tidy to Parsoid. WMF's Linter coming online. Is the next year a golden age or dark age for gnomes? Bgwhite (talk) 01:19, 4 December 2016 (UTC)[reply]
Re #3, definitely leeway on how the migration happens and at what pace, but at some point in the future magic links are definitely going to go. And hopefully a golden age :) Legoktm (talk) 01:59, 4 December 2016 (UTC)[reply]

Thank you for taking this on Magioladitis. :) Legoktm (talk) 01:59, 4 December 2016 (UTC)[reply]

I've been switching ISBN magic links to use a template in a semi-automated fashion in the past few months using a dumb Python script. Both Magioladitis and Bgwhite took notice and are now eager to get in on the fun. ;-) The script allows me to see a diff of each edit and then approve or not with a keystroke. I can post that script somewhere if there's interest.

I think switching to the {{ISBN}} template is safe and doesn't need to wait for anything. From the few thousand pages now using this template, I've seen no adverse effects and don't expect to. Switching to a template brings ISBN syntax in line with pretty much every other type of similar link on the wiki, with the same tracking mechanisms, edit interface, etc. We really should've done this a long time ago, but now is the next best time to get this over with, in my opinion. --MZMcBride (talk) 05:19, 5 December 2016 (UTC)[reply]

There is interest, MZMcBride :) Somewhere on Github would be fine --Edgars2007 (talk/contribs) 15:51, 26 December 2016 (UTC)[reply]

Hi Edgars2007. Sure, here it is: Special:Permalink/756836310. It's not a great script, but it works well enough, I suppose. --MZMcBride (talk) 03:26, 27 December 2016 (UTC)[reply]

MZMcBride, ahh, a simple one-liner :) Thank you! --Edgars2007 (talk/contribs) 16:59, 28 December 2016 (UTC)[reply]

Approved for trial (75 edits). Please provide a link to the relevant contributions and/or diffs when the trial is complete. — xaosflux ^Talk 05:02, 16 December 2016 (UTC)[reply]

Given the large scope of this, I would propose turning off all other changes in AWB, and making only this change. That is also what BenderBot seems to be doing with the http -> https changes it makes. There is no reason why any other changes need to be made at the same time, and Yobot has a poor track record when making general fixes - in fact, it is currently blocked because the bot was making incorrect edits unrelated to the bot task which was supposed to be running. — Carl (CBM · talk) 23:52, 17 December 2016 (UTC)[reply]

CBM In this case since it is a Find and Replace issue there will be problems at all. I have skip checks for this one. Problems occur because I do ot have skip checks for all of Yobot's tasks. -- Magioladitis (talk) 08:57, 20 December 2016 (UTC)[reply]

Because of the scope, I would still prefer to see no general fixes or other changes. Just do the proposed change and get it over with. There are several discussions going on elsewhere about the possibility of removing gen fixes from all Yobot tasks. No reason to put them on another. — Carl (CBM · talk) 11:57, 20 December 2016 (UTC)[reply]

I want to register a concern that I can't find where consensus has been established to do this. The ISBN and PMID magic links are used by editors who don't use citation templates (myself included), and they are very quick and easy to use. Should there not be a wider discussion before removing them from enwiki? SarahSV ^(talk) 17:28, 22 December 2016 (UTC)[reply]

It seems they are going to be deactivated across all Wikipedia languages. So with no change there would just be a plain text ISBN or PMID. Whether that would be better or worse than the proposed change is a question, but it seems the status quo isn't an option. — Carl (CBM · talk) 20:12, 22 December 2016 (UTC)[reply]

SarahSV See User_talk:MZMcBride#ISBN_magic too. -- Magioladitis (talk) 20:16, 22 December 2016 (UTC)[reply]

I find myself agreeing with MZMcBride: Let's get this over with. On CBM's point, if it would be easier/less controversial less risky to run it with everything else turned off, then let's do this as a solo task. That might also give us more room in an edit summary to explain that we're updating to remove reliance on a deprecated feature that will be removed. WhatamIdoing (talk) 20:21, 22 December 2016 (UTC)[reply]

Edit summaries can use a very small piped link (e.g. blahblah|Task 27) that links to a full description on the bot page. It really doesn't need that much room. — xaosflux ^Talk 20:29, 22 December 2016 (UTC)[reply]

Magioladitis, thanks. I can't follow what happened at phabricator:T145604. Someone proposed it, and then what? Thank you for pointing out that it ought to be discussed. I do remember you adding the magic numbers for PMID not long ago, so to have to remove them again seems odd. They're very useful. SarahSV ^(talk) 20:56, 22 December 2016 (UTC)[reply]

Technical RFCs have a more extensive and diverse process than the ones we're used to. There is always a Phab task (sometimes whole trees of tasks), a page on mediawiki.org if the idea isn't shot down early, a real-time discussion on IRC, usually discussions on e-mail lists (frequently multiple discussions) as well as on wiki, and sometimes in-person meetings (e.g., at the hackathon associated with Wikimania). RFCs are not only listed centrally but supervised by a committee to make sure that decisions (if any) get implemented, and they are advertised through multiple channels. This one, for example, has been advertised in Tech News and mailing lists. It is not unusual for a technical RFC to last for many months or even years.

In terms of "this ought to have been discussed", it has already been discussed, in public, for months now. I looked into this a while ago, and I very strongly doubt that any kind of "discussion" involving Wikipedia editors would have changed the decision. The main people involved are long-time Wikipedians, and they're aware that this change will likely precipitate small changes to more than half a million pages across many thousands of wikis (including wikis not run by the WMF). But the technical benefits are really quite important, so it's going to happen despite these costs.

In terms of our own internal process, this kind of thing falls into WP:CONEXCEPT: The MediaWiki dev community (which is mostly volunteer devs), and not the editors at this one wiki, is responsible for deciding what the core software does. WhatamIdoing (talk) 22:19, 22 December 2016 (UTC)[reply]

I agree that this bot, if approved, should have an edit summary that mentions "Task 27" and links to this BRFA. It should also not perform AWB's general fixes. – Jonesey95 (talk) 18:34, 23 December 2016 (UTC)[reply]

I agree that this should be done without other general fixes. -- Magioladitis (talk) 08:32, 27 December 2016 (UTC)[reply]

xaosflux I'll postpone the bot trial for some days since there are still ongoing discussions. -- Magioladitis (talk) 11:30, 27 December 2016 (UTC)[reply]

That's best. — xaosflux ^Talk 11:54, 27 December 2016 (UTC)[reply]

Given the discussion / confusion at [1], I think it would be best to have a clear RFC on a village pump before any bot implements this task. — Carl (CBM · talk) 14:18, 22 January 2017 (UTC)[reply]

With Yobot being blocked for a month, with no end to that in sight, this should probably be closed as stale until it can be taken up. Kick the issue of consensus down the road until it's brought up again, etc. ~ Rob₁₃^Talk 08:10, 24 January 2017 (UTC)[reply]

BU Rob13 No worries. I can do te test via my main account. The speed issue is not the issue here. I am working on the regex. -- Magioladitis (talk) 08:16, 24 January 2017 (UTC)[reply]

BU Rob13 recall that his task will be done with no general fixes. While ofcourse while editing from my account I can perform general fixes too. -- Magioladitis (talk) 08:18, 24 January 2017 (UTC)[reply]

Moreover, we are not on hurry pending larger consensus. -- Magioladitis (talk) 08:26, 24 January 2017 (UTC)[reply]

My comment was purely procedural. As a matter of managing the BRFA backlog, a blocked bot can't do a trial, so doesn't make much sense to keep this here whether or not the task has merit. I haven't reviewed the task yet, so I couldn't say. ~ Rob₁₃^Talk 08:27, 24 January 2017 (UTC)[reply]

Tasks have been made around for longer. Don't worry about the backlog. If there was to make a bot run I could ask for a specific unblock only to make a test run. It was happened before. -- Magioladitis (talk) 08:40, 24 January 2017 (UTC)[reply]

I am not aware of any ongoing discussions. Maybe time to start testing so that we are done with the test phase? -- Magioladitis (talk) 21:24, 1 February 2017 (UTC)[reply]

@Magioladitis: Getting back to reviewing this - can you point to a local community discussion where a decision has been made on how these should be dealt with? — xaosflux ^Talk 03:46, 3 February 2017 (UTC)[reply]

@Xaosflux: what are the community's options? ISBN magic links will stop working. Only way to maintain the same functionality is to replace them with templates. -- Magioladitis (talk) 08:39, 3 February 2017 (UTC)[reply]

@Magioladitis: The discussion you linked to at the top has 2 additional proposals that differ from your implementation. There seems to be initial agreement that something should be done for these, however I have not seen an emerged consensus as to to "what" that should be. This will need an RfC, feel free to coordinate one and then reapply after support has been determined.

Denied. Task denied due to lack of established community consensus for a job of this size. — xaosflux ^Talk 15:06, 3 February 2017 (UTC)[reply]

The above discussion is preserved as an archive of the debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA.