Page MenuHomePhabricator

Clarify future of discourse-mediawiki.wmflabs.org: Make the website read-only
Closed, ResolvedPublic

Description

Copying from https://lists.wikimedia.org/pipermail/wikitech-l/2020-July/093642.html :

In January 2018, an experimental instance of the Q&A platform "Discourse" was set up for developer support at https://discourse-mediawiki.wmflabs.org . See https://phabricator.wikimedia.org/T180854 .

Since its launch, the usage of this instance has declined: https://phabricator.wikimedia.org/T247010#6270883
On average, there are two users who are engaged daily.

Other support venues which exist in parallel are much more active, such as IRC, mailing lists, [[mw:Project:Support_desk]], [[meta:Tech]], local technical village pumps, newsletters, Zulip for outreach programs, or other third-party sites such as mwusers.org, StackExchange.com, or Reddit.

Discontinuing the least frequented venue (which unfortunately is Discourse though it seems to be a great communication platform) would keep us from further fragmenting our conversations and efforts.

Planning, implementing, establishing and running a centralized Wikimedia developer support venue on production infrastructure (puppetization, data backups, uptime, etc) would require significant technical and "social" capacity. If Discourse was to be further evaluated, https://phabricator.wikimedia.org/T155678 and its subtasks list part of the remaining needed work.

I propose to make discourse-mediawiki.wmflabs.org read-only by the end of August 2020.


Random other links:

Event Timeline

I have boldly updated wiki pages about the somewhat unclear status of mediawiki-discourse.wmflabs.org and removed links from some other pages (e.g. outreachy/gsoc, TechEng team, etc)

Related but not related but related indeed ;)

https://discuss-space.wmflabs.org/t/responding-to-questions-about-a-future-discuss-space-owned-by-volunteers/3367

Basically, the answer to the questions posed in these task might or might not be influenced by events in Discuss Space, if eventually a team of volunteers takes over. Other than that, my involvement in both Discourse instances was motivated by the hypothesis of having a platform supported and maintained by my team at the Foundation. Since this isn't happening, I'm transitioning out as admin in both instances.

PS: Should the future of discourse-mediawiki.wmflabs.org be clarified at discourse-mediawiki.wmflabs.org?

if eventually a team of volunteers takes over.

@Qgil: From my personal point of view (not speaking for any team), things to sort out first are how not to have production data in a non-production instance, backups, proper puppetization, technical responsibility. And if a team of volunteers could sort out all these things themselves.

PS: Should the future of discourse-mediawiki.wmflabs.org be clarified at discourse-mediawiki.wmflabs.org?

There is a banner... In addition, any admin could update https://discourse-mediawiki.wmflabs.org/t/welcome-to-wikimedia-developer-support/8

Heads-up to @Tgr as he seems to be the user who is most active in that instance lately.

19 users logged in in the last 30 days. 77 users within the last six months.

discourse-users.jpg (263×415 px, 35 KB)

Community size seems small compared to activity on wikitech-l@, mediawiki.org support desk, IRC activity. Especially with regard to community fragmentation.

Aklapper triaged this task as Low priority.

I have published a proposal to make this instance read-only by August 2020: https://lists.wikimedia.org/pipermail/wikitech-l/2020-July/093642.html

Space has been canceled

I get a 404 for this link.

What's discourse.wmflabs.org, BTW? 🤔 I couldn't find any information about it on a quick search in meta.

@Kaartic: Thanks for catching that. Looks like URLs broke. I've updated the link in the task description to one that seems to work.

I get a 404 for this link.

It used to be a post about the closure of discourse.wmflabs.org. Not relevant to this task.

What's discourse.wmflabs.org, BTW? 🤔 I couldn't find any information about it on a quick search in meta.

You can find it in the page history of https://meta.wikimedia.org/wiki/Discourse if you are curious. It's the initial Discourse instance, it was set up by @AdHuikeshoven five years ago, IIRC. It was mainly used for testing.

Aklapper raised the priority of this task from Low to Medium.Jul 27 2020, 7:44 PM

I have published a proposal to make this instance read-only by end of August 2020: https://lists.wikimedia.org/pipermail/wikitech-l/2020-July/093642.html

No feedback in this task and no feedback on the mailing list so far. Looks like nobody cares (which is fine).

Moving to read-only is fine by me.

Why was/is the discourse instance not promoted more?

Also see related T212653: Promote "Wikimedia Developer Support" in the current channels where developer support is provided.
It has been promoted. Could you elaborate where and how you would have expected "more"?

I have not heard of any complaints, hence I will proceed and make the website read-only soon.

Aklapper renamed this task from Clarify future of discourse-mediawiki.wmflabs.org to Clarify future of discourse-mediawiki.wmflabs.org: Make the website read-only.Sep 1 2020, 6:39 PM

It has been promoted. Could you elaborate where and how you would have expected "more"?

Personally, I came to know about it by chance, I haven't seen any promotion of it. If it was promoted then unsuccessfully.

@Aklapper Two feedback on the discourse notice:

  • The font is huge, I think half the size would be sufficient for readers to notice.
  • The recently launched Mattermost chat can serve as an easily accessible alternative (unlike IRC) or even continuation of the discourse instance, therefore I hope you could add that to the notification also:

https://chat.wmcloud.org/
https://meta.wikimedia.org/wiki/Wikimedia_Chat
https://lists.wikimedia.org/pipermail/wikitech-l/2020-August/093775.html

Thank you.

Hi,

  • The font is intentionally huge, as people tend not to read banners. (This is based on my personal experience in my volunteer time that the banner that I put on https://bugzilla.gnome.org is not read by everyone and that people continue to create tickets over there. The next level of annoyance would be a popup overlay.)
  • Apart from Mattermost we also have Zulip for outreach programs, we have the #wmhack group on Telegram which is used by quite some folks for general questions not Hackathon related, and probably many other places, so I intentionally went for established places which have a larger number of users now.
  • The font is intentionally huge, as people tend not to read banners. (This is based on my personal experience in my volunteer time that the banner that I put on https://bugzilla.gnome.org is not read by everyone and that people continue to create tickets over there. The next level of annoyance would be a popup overlay.)

I see. Yes, the personal experience part about ineffectiveness came through ;-) Anyway I would suggest a bit reducing the size and adding more content.

  • Apart from Mattermost we also have Zulip for outreach programs, we have the #wmhack group on Telegram which is used by quite some folks for general questions not Hackathon related, and probably many other places, so I intentionally went for established places which have a larger number of users now.

All these are quite hard to find. I recall spending hours looking for places where I might get answers to my questions. In the end my questions remained unanswered and I haven't found these places either.
From my personal experience: it would help to add all these 3 you've listed (I'm not against adding the rest 😄). People will choose the one that's closest to their area of interest and/or the most familiar/accessible of these tools.
I think a proactive approach would be beneficial here: the only way to transition from outdated, but very well established solutions toward modern, more effective and convenient ones is to promote those, thus being less established is extra reason to advertise them, otherwise the fate of discourse might repeat itself.

From my personal experience: it would help to add all these 3 you've listed (I'm not against adding the rest 😄). People will choose the one that's closest to their area of interest and/or the most familiar/accessible of these tools.

The more choices you offer, the more confused I'm going to be which choice to take. I might post in all places, which will fragment conversations and duplicate efforts of people replying. This problem already exists and I'd like to not make it bigger.

I think a proactive approach would be beneficial here: the only way to transition from outdated, but very well established solutions toward modern, more effective and convenient ones is to promote those, thus being less established is extra reason to advertise them, otherwise the fate of discourse might repeat itself.

I agree; this first requires a plan and broader discussion and agreement what to promote, and what to replace. The reality is usually https://xkcd.com/927/ ...

The more choices you offer, the more confused I'm going to be which choice to take. I might post in all places, which will fragment conversations and duplicate efforts of people replying.

But you aren't the target audience ;-) Volunteers - like myself - have a more direct view of what can be achieved on each forum.

The multiple channels on IRC are already fragmented. While it makes sense to someone participating for years, there is no guideline which channel to use for a newcomer, whom to ask, etc. Also, IRC is so troublesome I'd suggest it only as a last resort for a newcomer. Just setting up a cloak takes hours.

These forums each have a different set of active contributors, the discussions that can take place widely differ. In my experience not fragmentation is the problem but to find an actual forum where someone can and cares to answer my question. Usually I try the alternatives in sequence until the topic sticks at one place. The issue is not that multiple fragmented discussions would take place, but to just find one place where a question would be answered.

For this reason I imagined a thematic listing of the alternatives that highlight the main focus of each would give better coverage and possibly help people find the right forum. For ex:

The issue is [...] to just find one place where a question would be answered.

I agree, with a stress on "one". Hence adding more places itself does not solve the problem, but a plan to consolidate places would.

The issue is [...] to just find one place where a question would be answered.

I agree, with a stress on "one". Hence adding more places itself does not solve the problem, but a plan to consolidate places would.

That's the opposite of what issue I meant which is: finding more than zero places that solve the problem.
I doubt a plan to move all developers from the inaccessible IRC to any of the alternatives would be a simpler plan than for ex. the move to GitLab. Probably much harder, so I prefer to stay on the groups of what's possible. Which is (continued from previous comment) a listing similar to:

The descriptions might not be the most accurate as I'm not intricately familiar with all, you can do much better than this. I hope this explains what would be helpful from an outsider perspective.

There's a list in the sidebar of the support desk (currently IRC, mailing lists, Stackoverflow, and mwusers), which I think could probably be expanded to include all active venues even if they're not massively popular.

@Aklapper Would it be possible to add the following links to the notification on https://discourse-mediawiki.wmflabs.org/ ?

* [Wikimedia Chat (Mattermost)](https://chat.wmcloud.org/)
* [MediaWiki unofficial (Discord)](https://discord.gg/ZrV2Ex9)
* [Outreachy chat (Zulip)](https://wikimedia.zulipchat.com/) 
* [Hackathon chat (Telegram)](https://t.me/wmhack)

Yes, I remember. However that approach failed with mediawiki-discourse which remained largely unknown throughout the developer community, presumably also due to a lack of explorability. I think this time a more proactive approach should be taken, or we'll see history repeat itself.

The more choices you offer, the more confused I'm going to be which choice to take.

Whoever finds mediawiki-discourse is already searching for the right forum, it would not be found just randomly. IMO the best way to help them is by giving explicit, direct choices without the need to further traverse the web and make those choices deeper down the graph. This concern of confusion should be addressed by a detailed list with specific guidance similar to one I presented in T247010#6445709:

  1. Support Desk (official)
  2. IRC - Primary place to find WMF developers
  3. Mailing lists
  4. MediaWiki unofficial (Discord) - Run by wiki operators. Topics include wiki installation, operation, skinning, custom developments (details)
  5. Wikimedia Chat (Mattermost) - For movement-wide discussions, not just technical matters. Started in September 2020 (details)
  6. Outreachy chat (Zulip) - For Outreachy programs (details)
  7. Hackathon chat (Telegram @wmhack group) - For Hackathon related questions (details)

No, please see my previous comments. Throwing millions of links at people instead of 3 links solves no problem, to the contrary.

No, please see my previous comments. Throwing millions of links at people [...]

Note: Those are the comments I've responded to with a different POV. I believe a more detailed notification with explicit guidance is more helpful. However, we don't need to agree on this, the visibility and effect is probably quite small, not worth spending more energy on it.
Also note: "millions of links" is not a neutral or factual understanding of my contribution ;-)