Use twinkle-enwiki on enwiki

[siddharthvp]

The localisation project is likely to be completed soon. Opening this issue to discuss the procedure for migrating the existing twinkle to the localisation-based setup.

@MusikAnimal @Amorymeltzer

I'll try to write more integration tests to improve the confidence in the new setup working correctly. But TBH writing integration tests is time-consuming and no matter how far we go, there'll be subtleties left to be manually tested. Since we don't have an army of QA testers at our disposal, we should decide on an efficient strategy for the migration so that it doesn't take forever to be deployed.

[Amorymeltzer]

I have no real intention of contributing for the foreseeable future: this stopped being collaborative or cordial months ago, and I simply don't need the daily agita from something that's no longer enjoyable or healthy.

---

As for the topic at hand, I suppose it'd help for buy-in, but enWiki is the one place this doesn't matter. Better to ensure it's actually capable of and willing to be used where it counts, or else it's for nought.

[MusikAnimal]

It is upsetting to lose @Amorymeltzer as a maintainer. I think @siddharthvp is a talented engineer who has done some very superb work, but I must agree that much of the collaborative spirit seems to have been lost since the localization project started. At #1144 (comment) I was hinting that consensus was needed among maintainers before moving forward. I recall we got the go-ahead from @atlight on the grant proposal page on Meta, and there was obviously community consensus, but I'm not sure consensus was achieved among those who are/were current active maintainers (which I don't consider myself as). There lies the problem.

But, the work is largely done now, and while it comes at a cost of losing one of the best maintainers Twinkle has ever had, we should move forward. I want to thank Amory for all the hard work and exemplary leadership they have shown since they started contributing in 2018, and I hope one day they'll want to contribute again. I also want to thank Siddharthvp for all the amazing work he has done. It is a shame we can't have both of you :( I don't want to point fingers or take sides, but I will say that moving forward we need to be more strict about enforcing the Code of Conduct. And no, I don't mean me taking action in my role as a CoC Committee member (which I can't do anyway), but an active commitment from everyone who contributes here to be open to other's input and stay true to our values of fostering an enjoyable, collaborative space for everyone. That is far more important than things like code quality or engineering best practices. What we don't want is to end up with a heavily used gadget without an active maintainer.

That said, I don't know if Twinkle needs a de-facto "lead maintainer", so to speak, but if it does, that would now have to be Siddharthvp. @siddharthvp, I hope you're okay with assuming this role. I don't have the time to contribute heavily right now, or even help but so much with code review, but I will offer to handle the interface-admin side of things for the time being. I agree with Amory that other wikis should be the focus for this rollout. I would suggest finding a candidate wiki, and installing the new Twinkle as a separate "beta" gadget and have the community try it out. We can take the same approach on enwiki. I have global interface editing rights, so I can assist any willing community with installing the beta version. I don't have much time to do the outreach part, though, so hopefully you're okay with handling that. I did this for MoreMenu and it was a painful few months.

[siddharthvp]

I did exchange a couple of friendly emails with Amorymeltzer, so I am surprised by the tone of the comment here. However, I think Amory was talking about not contributing w.r.t this issue (which they indicated in emails as well – because of being busy IRL), not all of twinkle. In case this is wrong and MusikAnimal is right, I'd be sad too.

... that is far more important than things like code quality or engineering best practices

I agree. I took exception with #1349 primarily because it's a 2500-line PR created without any prior intimation necessitating much changes in localisation rewrite. This is not what I expected from someone who's so vocal about being collaborative.

I'm not sure consensus was achieved among those who are/were current active maintainers (which I don't consider myself as). There lies the problem.

Agreed. The discussions I opened (#1220, #1226) were all inconclusive and the only options were (i) wait indefinitely for people to comment in them, or (ii) go ahead with my own ideas. Even (i) was no option since I had some obligations to keep up with because of the grant, so there was really no choice.

[Amorymeltzer]

I think Amory was talking about not contributing w.r.t this issue (which they indicated in emails as well – because of being busy IRL), not all of twinkle. In case this is wrong and MusikAnimal is right, I'd be sad too.

He is. I am busy, which is perhaps the catalyst but not the reason. I'm not willing to spend my valuable free time engaging when I no longer enjoy it and doing so leaves me actively unhappy.

Agreed. The discussions I opened (#1220, #1226) were all inconclusive and the only options were (i) wait indefinitely for people to comment in them, or (ii) go ahead with my own ideas. Even (i) was no option since I had some obligations to keep up with because of the grant, so there was really no choice.

This is actually a pretty clear example of what I'm talking about: It's not those issues, it's the thought process immediately after. I don't wish to put words in MA's mouth, but he linked to his unanswered comment from well before those. The issue began with #129 (comment) where, with essentially no discussion, input, or planning with anyone else, much less other active participants, the grant was immediately filed. That was the beginning of the end, as once that began, everything could be justified by saying it "was no option since [you] had some obligations to keep up with because of the grant, so there was really no choice." Saying "there was really no choice" is not good-faith collaboration, it's my way or the highway.

#1349 is not a directed attack; if you want to OSE me about it, fine, but it's a matter of scale. There is a big difference between "proposed additional class to add much-requested features" and "entire project-spanning rewrite, restructure, and redesign." I want(ed) to talk to folks, including you, about my proposal, and indeed had intended to redo a fair bit of it per your suggestions; nothing even close to that took place over the entire half a year of this. #1349 is closer in scale to #485 IMO, but again, my issue isn't the code or the concept itself, it's the behavior and ownership around it. The project has crystallized around one person's singular vision; it's nice you got paid for it, but I don't need to be bullied into being a part of it.

[siddharthvp]

The issue began with #129 (comment) where, with essentially no discussion, input, or planning with anyone else, much less other active participants, the grant was immediately filed

Please note that I brought up the idea 8 months prior in which I specifically asked you and MA if you were ok with it. Your response was affirmative. As far as planning is concerned, I did not have a plan myself. The whole point of the grant was to serve as a driving force, because all of us only edit enwiki, there is no other incentive to work on a long-term localisation project.

Saying "there was really no choice" is not good-faith collaboration, it's my way or the highway.

That's taking things out of context. By the very meaning of collaboration it needs multiple people. At #1220, it took you 12 days to reply when my own reply earlier had come within a day, while you were quite active on this very repo. That made it clear to me that you had no intention of collaborating. As for #1226, I closed it ONE MONTH after the last comment by me didn't get a response. If you're saying that you wanted to discuss things but I didn't give space for that, unfortunately the facts show otherwise. Even if there were no deadlines, one month would be considered sufficient time to wait before a discussion can be treated as over.

I always said I am open to redoing things if someone could come up with a better idea.

There is a big difference between ... and "entire project-spanning rewrite, restructure, and redesign."

Twinkle-core is a separate project. Obviously I do wish it be used on enwiki. I don't think I'm wrong to hope that. I always anticipated that you would oppose it. It seems you took my communication too literally – the point of sending them was to check what you thought about the idea. I totally knew that you would never contribute to the twinkle-core project.

#1349 is closer in scale to #485 IMO

Maybe. The difference is that #485 was discussed it at #473 for two weeks. It didn't take anyone by surprise.

[Amorymeltzer]

I've no need or desire to continue bickering with you. As I've stated, I intend on entirely disengaging from active maintenance, both here and on-wiki, at least until I feel it's worth my time to do so. I've neither reviewed nor trust anything since c454d89.

[siddharthvp]

I think it's worthwhile discussion, not bickering – though naturally you don't want to continue it as it isn't showing you in good light.

[atlight]

naturally you don't want to continue it as it isn't showing you in good light.

Comments of this nature are not constructive; please try to focus on the technicalities of this issue.

It would be preferable to do a test deployment on testwiki and/or en.wikipedia.beta.wmflabs.org first, which can remain there for an indefinite time until we feel that as many lingering issues as are going to be found have been found. I can assign user rights where needed for this (and I imagine @MusikAnimal also is able to assist in this way).

I don't see that there would be enormous motivation for enwiki users to help with testing. It might be better to try recruiting testers by looking for keen users from other wiki communities where old, semi-broken Twinkle copies are relied upon. Even if it is the enwiki version they are testing, they will be assisting to shake down the infrastructure that will allow Twinkle to come to their wiki in their language.

[siddharthvp]

Will start a new discussion later as this one is clearly derailed off track.