In T407978#11298546, @DLynch wrote:

It'd be fine to change it, but since it's been the way it is for over a decade it'd probably be good to make sure there's actually community consensus to do so.

I'm not sure how to write a reliable test for this, any suggestions are welcome.

It's an oversight, working on it.

I can confirm the same issues exist in Romanian

In T214998#11165163, @TheDJ wrote:

This should make that easier and not harder, so that should be nice.

In T376845#11129252, @Jdlrobson-WMF wrote:

I think it would be helpful if people could share example URLs to charts which are working around this or would benefit from this (particularly where the axis become unreadable on mobile devices). Having more test cases will be helpful for ultimately getting this addressed.

Awsome news, looking forward to the moment we can use Automoderator on rowiki as well.

I'm tagging @MusikAnimal since he is marked as the contact person for this task.

@Xqt , why did you choose to implement this directly in Page instead of ItemPage with a wrapper on top in Page? Is it OK if I move it?

Function already existing in class Page

@Dimitrie569 translation fixes are done at translatewiki.net - a volunteer-based translation service. For this, see: https://translatewiki.net/wiki/Special:SearchTranslations?query=Adaud%C4%83&language=ro .

For Romanian, I suggest reaching out to User:Klaul who lead this research project on NPOV: https://meta.wikimedia.org/wiki/Research:Grounding_NPOV_pillar_in_post-censored_information_ecology

Hi. Responding to T388215#10655076, I can help review the model in Romanian.

In Romanian we have:

• WP:EJV - peacock words, value judgments
• WP:FE - weasel words (peocock are included here)

@Pablo your wikidata query for "All MEPS from Wikidata" misses the cases where no start time is present - around 350 MEPs [1] [2]. This also affects recent legislatures, with only 630 MEPs out of 705 MEPs [3] (actually more, as some don't finish their mandate and the Brexit replacements) present in tab "MEPS (2019-2024)". There are no red links on enwiki [3], so this must be a data issue.

Awesome news, thank you very much!

I took a look at recent log entries and could not find another issue like this. I propose we close the issue since it seems just a glitch in the matrix.

Right, it's been a while since I translated locally, I had forgotten we could do that. :) Created https://ro.wikipedia.org/wiki/MediaWiki:Ipinfo-widget-special-ipinfo-link and closing the issue. Thanks for the help!

Thanks, I updated the translation! Is it possible to import the translations manually or we have to wait for the next automatic import?

It's yet another place where people would check for IP info (it was the first click I made when looking at the initial on-wiki report). If possible, I'd like to confirm that the IPv4/v6 identification works correctly there as well. I don't have an example that it doesn't, just want to make sure (e.g. with a test).

@mszabo will this CR also fix the display at https://ro.wikipedia.org/wiki/Special:IPInfo/~2024-14596 ?

Thanks for the quick responses, I guess we can close the task then! Just one question: is this documented somewhere on-wiki for future users?

A very early empirical look at how PatrocleBot works (using multilingual model) shows no changes in revert numbers.

I understand you don't want to give and take the right when a preference is changed, but maybe then rename the right to "eligible for vieweing IPs"?

All the users meet these requirements. However:

This was caused by T378336

In T369610#10228321, @gerritbot wrote:

Change #1080250 had a related patch set uploaded (by STran; author: STran):

[operations/mediawiki-config@master] Give the `abusefilter-maintainer` group protected vars access

https://gerrit.wikimedia.org/r/1080250

Done, fixed, thank you!

In T356102#10144392, @diego wrote:

We will need a different model for new article evaluation. I'll be happy to work on such model, but until now, we haven't received requests about this.

I see an opportunity to use this endpoint for evaluating new pages. Any thoughts @achou or @diego ?

In T372299#10082137, @Kgraessle wrote:.

Would all communities want this, or should Undo vs Rollback be a configuration?

I think we should keep it configurable.

@Ifeatu_Nnaobi_WMDE here you go, simplified as much as I could: https://ro.wikipedia.org/wiki/Utilizator:Strainu/T279069

@KCVelaga_WMF is it possible to run this notebook on PAWS? If not, how can I reproduce the results?

@Bugreporter do you have a link to the docs for that feature? It still sounds like a bug to me to report a different block, but it might be just cosmetic.

Marking Unbreak Now as this will clearly cause some duplicate citations and some confusion to less experienced users while it's happening.

@Superpes15 the discussion has been opened for almost 2 weeks. Can we now move on and implement it? I can do the change if you don't have the time.

@jsn.sherman in ballpark figures in a week it recovered a day, during what I suspect is peak usage. Maybe it makes sense to give a few more resources to the process, at least while recovering from an outage?

Verified working, thanks all!

In T353931#9442206, @Urbanecm_WMF wrote:

I think logging opt-outs publicly negates that purpose.

@Etonkovidova no, that was Martin experimenting from the API. From the homepage, the only available method to change mentors is to opt-out of mentoring, then opt-in again. There should have been some log entries with me opting out, then opting in again.

As described in T353805#9420785 this feature is useful in sockpuppet investigations, when we have accounts hunting for a specific mentor. There are several non-exclusive ways to go forward which would still be acceptable in my opinion:

• aggregate opt-outs followed by an opt-in in a single change (e.g. from mentor A to mentor B instead of mentor A -> none -> mentor B)
• expose in GUI a way to directly change the mentor to another - I understand this is already available as an API call
• limit the visibility under a userright that would be assigned to trusted user groups (e.g. patrollers and admins)

Thank you for your response @Urbanecm_WMF . Indeed, the process I did was "give up on mentorship" and then add a mentor. From GUI this is the only available option to me (see attached picture). How can I choose somebody else directly from GUI?

@Aklapper could you recommend someone who can review this change? Thanks.

2 weeks later, there seems to be no progress on the tasks that depend on that script working so I don't really care much about potayto potahto discussions - we need changes to various namespaces throughout the projects.

Already working on this, will put up a patch for review shortly.

@Ladsgroup are there any remaining steps so that bugs blocked on this one can be processed?

I see the error on my production code as well. Probably a duplicate of https://github.com/magnusmanske/petscan_rs/issues/106

Thank you @diego for analyzing the data. If my understanding is correct, the precision is the metric of the model that will be visible to the community - how many of the reverts are correct, so this is what I'm looking at.

In T336934#9350484, @diego wrote:

Great. Having some manual labels is always valuable.
I have done a quick check and I've seen there are few cases were the RR scores are not higher than 0.93. For example, this one:

@jsn.sherman I noticed that some of the technical decisions taken in this task are limiting the number of revisions to be analyzed:

• the inability of the models to score the first revision in a page,
• using the scores saved locally in MediaWiki, which at least in the case of ores might be missing or be different from the score returned by the API
• skipping revert wars

@diego since you graciously offered to dive deep on this comparison, maybe you will find it useful to also see a manual review on some diffs. I asked the patrollers @ rowiki to manually review a set of changes that received a score >=0,93 from the revertrisk algorithms and the results can be found below (column 2 is "would revert", column 3 is "would not revert"):

• revertrisk-multilingual
• revertrisk-language-agnostic

Thank you for pointing that out @achou. Feel free to close the task, but as a note, the wording in the LA page is a bit ambiguous for me - I thought that the Model section applies to the training step.

In T336934#9328589, @diego wrote:

My main question is: Which column are you considering as ground truth (your labels)?

In T336934#8880657, @Samwalton9-WMF wrote:

There are two version of the model: Language-agnostic and multilingual. The multilingual model is more accurate, but only available in 48 languages, as documented in the model card.

@diego do you think there would be any reason to allow communities to select which model to use, or is the multilingual model simply the better option?

EDIT: We discussed this on a call. The short version is that it depends, particularly on the performance of the models, which is still in flux.

In T195318#7642570, @TheDJ wrote:

If anyone wants to fix this, this logic is in LanguagePicker.js of https://gerrit.wikimedia.org/r/plugins/gitiles/mediawiki/services/kartotherian/+/refs/heads/master/packages/babel/#language-resolution