Talk:Community Tech/Article Alerts for more languages

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search

Tentative roadmap[edit]

I made a tentative roadmap at en:Wikipedia:Article_alerts/Roadmap. Feel free to consult it and tweak it. It could also be hosted on meta if that's preferable for collaboration. Headbomb (talk) 13:04, 14 December 2017 (UTC)

Current bot source code[edit]

The bot is hosted on Bitbucket that I and Headbomb can access. I can't make it public, because it's not secure (literally has the password in plain text). But I can invite whoever is working on this. Or I can just zip and send the code. —  HELLKNOWZ   ▎TALK   ▎enWiki 23:32, 6 January 2018 (UTC)

Thank you! This may be one of the first projects we tackle, we'll certainly reach out as soon as we start investigate in phab:T184304. If you want to zip it and email it over now, send it to wmf:User:Kaldari or me. — Trevor Bolliger, WMF Product Manager 🗨 17:15, 8 January 2018 (UTC)
Sent an e-mail to you. —  HELLKNOWZ   ▎TALK   ▎enWiki 17:45, 11 January 2018 (UTC)
🤘 — Trevor Bolliger, WMF Product Manager 🗨 18:09, 11 January 2018 (UTC)

"We'll be leaving the functionality and technology as-is on English Wikipedia"[edit]

Please provide the option for English Wikipedia to use this if they want. There are a few reasons for this

  • The English Wikipedia would make an absolutely fantastic testbed in terms of functionality and corner cases. We can compare the output of the new bot to the old bot to see if it misses articles it should picked, either because of configuration problems, or because of legit bugs in the bot. We also have a large userbase, with many people highly interested in Article alerts, so we would have lots of eyes on this.
  • It may be possible to recreate most of the functionality of the current bot, and the benefits would scale to other languages.
  • The bus factor greatly endangers the long term viability of Article Alerts on English Wikipedia. If a catastrophic fail happens, or the main coder dies, we have no backup.
  • Some workflows are not implement in Article Alerts, we could have a secondary system to pick up those
  • We very likely could setup pseudoworkflows (Cleanup Watch to keep track of pages entering/exiting cleanup categories).
  • It's possible that some projects prefer the new bot to the old bot, or use both bots. While the current system has a lot of advanced features (e.g. vote counting in discussions), it's quite possible that broader coverage (new workflows/added workflows) is more desired than specialized coverage.

Headbomb (talk) 13:22, 2 May 2018 (UTC)

Community Tech unable to work on this project[edit]

Headbomb, Hellknowz and everyone else watching this project -- I want to bring this status update to your attention. This was a hard decision for the team and for me personally because I really think this project had potential to help hundreds of contributors. I'm sorry again for the team not being able to take this project on. :( -- NKohli (WMF) (talk) 23:23, 7 November 2018 (UTC)

@NKohli (WMF): What's the issue? This does not seem like an unsurmountable hurdle in the least? Especially if Phab:T184304#3994104 is followed. Build a framework, let the individual wikis implement the details. It will likely not work for everything, but it will work for a lot of things. Headbomb (talk) 23:32, 7 November 2018 (UTC)
I agree with you. I think that solution can be implemented and can work. I did a fair bit of brainstorming on the ticket too. The hurdle right now is time and engineering capacity on the team. We are currently in the middle of three other projects and we only have four engineers on the team. Taking on this project will easily make our work stretch into March/April next year. We have the new wishlist coming up next month too. -- NKohli (WMF) (talk) 23:44, 7 November 2018 (UTC)
A delay is fine, but complete abandonment would be rather soul-crushing, especially given the massive potential benefits and impact this has. I can provide a lot of support/testing/feedback if we can test that on enwiki. This is a project that needs a lot of high-level design decisions, I gave what I think are solid suggestions to explore. If you have a per-wiki declaration of how workflows are setup + localization, combined with per-topic customizable subscriptions, that's something that would scale extremely well. How to achieve that (Extensions? A recentchange-like output that only report when a page enters/exists a workflow? A custom report by a WMF-operated bot?) is beyond me, but once that's designed, I can give a lot of actionable feedback. Headbomb (talk) 21:12, 15 November 2018 (UTC)