(Tentatively)

Hi @Johan, we're getting close to loading this on the beta cluster (<lang>[.m].<project>.beta.wmflabs.org) and I wanted to give a heads up that we'll be interested in (re)starting outreach. A couple of tangible things we were considering, was curious about your thoughts:

• Would it make sense to identify a small wiki or two with a couple editors who might be interested in trying this first in production? We anticipate we should be able to ship to the wikis in general, but did want to smoketest in prod on a wiki or two and work out any obvious issues there with interested people.
• @JKatzWMF and @Nirzar (and maybe @Tgr or another engineer or both) will likely be the ones working out more user-facing documentation on writing mobile friendly styles for templates using this new tech, beyond the basic examples. They also will likely have some additional outreach and possibly a re-assembling of a working group. I won't be as involved as we start shifting into this phase of the project.

I'll make a comment in the other ticket around the guidance (T164879: TemplateStyles style guidance) about this ticket for anyone scrolling through the history later on.

Would it make sense to identify a small wiki or two with a couple editors who might be interested in trying this first in production? We anticipate we should be able to ship to the wikis in general, but did want to smoketest in prod on a wiki or two and work out any obvious issues there with interested people.

This is very rarely a bad idea. (:

FYI, TemplateStyles is now live in Beta Labs. See, for example, https://en.wikipedia.beta.wmflabs.org/wiki/TemplateStylesTest.

In T164927#3330216, @dr0ptp4kt wrote:

Hi @Johan, we're getting close to loading this on the beta cluster (<lang>[.m].<project>.beta.wmflabs.org) and I wanted to give a heads up that we'll be interested in (re)starting outreach. A couple of tangible things we were considering, was curious about your thoughts:

• Would it make sense to identify a small wiki or two with a couple editors who might be interested in trying this first in production? We anticipate we should be able to ship to the wikis in general, but did want to smoketest in prod on a wiki or two and work out any obvious issues there with interested people.
• @JKatzWMF and @Nirzar (and maybe @Tgr or another engineer or both) will likely be the ones working out more user-facing documentation on writing mobile friendly styles for templates using this new tech, beyond the basic examples. They also will likely have some additional outreach and possibly a re-assembling of a working group. I won't be as involved as we start shifting into this phase of the project.

Any CLs will need such documentation in place before being able to evaluate further. Has this work started anywhere?
@Tgr, @Anomie, your extensive community experience begs a question: let's start with huwp and enwp? ;)

I'll make a comment in the other ticket around the guidance (T164879: TemplateStyles style guidance) about this ticket for anyone scrolling through the history later on.

In T164927#3379026, @Elitre wrote:

@Tgr, @Anomie, your extensive community experience begs a question: let's start with huwp and enwp? ;)

As much as I think it would be a very good place to get feedback, since there are many active editors there experienced in CSS, enwiki tends to hate being a testbed. If we want to use enwiki as a test, we'll need to first ask on the Village pump there (probably the 'technical' section, and post notices of the discussion in the 'proposals' section, at the Administrators' noticeboard, and MediaWiki talk:Common.css) and accept it if they say "no". I'll be happy to help draft the request and handle responses.

@bd808 mentioned wanting it on wikitech. mediawiki.org might also be a decent place for testing. In both cases, though, I don't know that the community size/activity is sufficient to get good feedback.

It all depends on how big a community we want. I've been thinking mid-sized wikis, which would give us a couple of first rather intimate discussions given that they'll most likely just have a couple of template editors at this level.

I defer to @Johan and @JKatzWMF on community outreach for the content sites.

Regarding wikimedia.wikitech.org (@bd808 suggestion), mediawiki.org (@Anomie suggestion), and maybe wikimediafoundation.org (wouldn't it be nice to style the home page and org chart?), @JKatzWMF any interest in us deploying there, even if not doing outreach? What do you think about outreach in those areas?

On wikitech I would like to use the extension to replace the .deploycal* and .mw-tpl-portal* styles we have in https://wikitech.wikimedia.org/wiki/MediaWiki:Common.css. It may also be useful for future documentation beautification work on that wiki.

(FYI the WMF wiki is possibly not worth working on much ATM, since Comms is actively working on deciding what to do about that site.)

If part of the point here is to start having conversations with template editors who edit templates that will be used by the non-technical editor community, I'm not sure mediawiki.org or Wikitech will take us in that direction. That's not an argument for not doing them as soon as possible if there's reason to, but they're not exactly the same kind of environment as the typical Wikimedia wiki.

@Elitre good point, I remember hearing about that, but don't recall the particulars.

@Johan, agreed. I think for our internal and engineering community purposes it would be nice to afford this option on mediawiki.org and wikitech.wikimedia.org as a number of users on these wikis spend lots of time in CSS land and would probably enjoy the opportunity to work on the technology on a production internet facing site a tad earlier.

In T164927#3379026, @Elitre wrote:

@Tgr, @Anomie, your extensive community experience begs a question: let's start with huwp and enwp? ;)

Huwiki is generally open to testing new things so I don't think I would have any problem getting consensus; I am less sure if there would be anyone willing to spend a lot of time converting templates. Without someone actively testing, there is not too much point in enabling it early. I'll ask around.

Enwiki has a much larger template maintainer community so it would no doubt provide great feedback. OTOH part of the reason of starting with smaller wikis is to find out what kind of anti-abuse (or even anti-accidental-misuse) features are needed, and limit potential trouble until that happens, so testing on enwiki is somewhat risky IMO.

@dr0ptp4kt Would be a potential way to do it, of course. If so, just to make sure we're on the same page, do we do something like this?

First step: mediawiki.org and Wikitech (I don't think the WMF wiki is a good idea, mainly because of what @Elitre pointed out above). Not a lot of outreach necessary. Mention it in the right IRC-channels, email to wikitech-l, more that level. Do this soon. Iron out any obvious problems.

Second step: A couple of mid-sized wikis (e.g. Hungarian Wikipedia, as suggested by @Elitre above if @Tgr thinks its a good idea). We mention it in the right local channels on the wikis, but concentrate on talking to active template editors.

Unless someone turns up and says "hey, I want that on my wiki" in case there's probably (with a few exceptions, e.g. maybe not the largest wikis first) good reason to let them have it on their wiki, as this will work much better if it's being driven by someone locally with our support.

@Johan that would work for me, although @JKatzWMF what do you think?

When looking for betatesters / early adopters, would it make sense to ask in wikitech-ambassadors? This in itself would be an exercise of explaining the feature and its rationale, and registering initial interest.

In T164927#3387765, @dr0ptp4kt wrote:

@Johan that would work for me, although @JKatzWMF what do you think?

Makes sense to me. Thanks!

Thanks @JKatzWMF - also, what do you think about the suggestion from @Qgil on wikitech-ambassadors for identification of interested early adopter wikis (i.e., those beyond mediawiki.org and wikitech.wikimedia.org)? Any parameters about how you'd want that positioned on wikitech-ambassadors if wikitech-ambassadors makes sense to you?

@Johan @bd808 @Tgr @Anomie would you feel comfortable trying to do a CL-led announcement that it'll be coming to wikitech.wikimedia.org and mediawiki.org, followed by deploying it there? I want us to be mindful of there being operational support in case of bugs; I for one have some vacation next week and part of the week following it (and wouldn't be a useful bug squasher, anyway).

In T164927#3392296, @dr0ptp4kt wrote:

I want us to be mindful of there being operational support in case of bugs

I have vacation the week of July 10-14 and the weeks of July 31-August 11.

In T164927#3392296, @dr0ptp4kt wrote:

[...] @bd808 [...] would you feel comfortable trying to do a CL-led announcement that it'll be coming to wikitech.wikimedia.org and mediawiki.org, followed by deploying it there?

There's really no community to warn on wikitech to speak of. We could announce on labs-l/wikitech-l/ops-l before or after the rollout. The early adopters would probably be me, Quiddity, and maybe Krinkle. We could try moving the styles for mbox/ambox into using it to give an initial test.

In T164927#3389630, @Qgil wrote:

When looking for betatesters / early adopters, would it make sense to ask in wikitech-ambassadors? This in itself would be an exercise of explaining the feature and its rationale, and registering initial interest.

(My2c) Generally speaking I'd rather keep this option for when we have no options, or need to encourage multiple wikis to volunteer. In this phase we could probably find the 2 candidates we need without bothering hundreds of people. Every message we send out has a cost in terms of attention.

Relevant: See @Nirmos' comment in T168808#3399400.

What is left to complete this task? It is not clear.

This task has been languishing and changing of hands. I for one would welcome a clearer description of this Readers-Community-Engagement request including the details suggested at https://phabricator.wikimedia.org/T157446

Dan and I spoke today. Here's a suggested (rough) plan. Open to suggestions and feedback.

There are currently two related tasks that need to be addressed.
First is restricting the feature to the Template: namespace. Creating Yet Another Standard (xkcd) on how to style content that is not a template could cause confusion down the road. T176269

Second is being prepared, in case of some emergency, that we have a quick way to deactivate the feature. Un-deploying extensions is a known process, but is more complicated than making a relatively simple configuration change. T176268

Meta wiki seems like a logical contender for the next non-content wiki. However the initial discussion was short and the community there would need to be approached again. T170863#3457418

The Swedish Wikipedia is interested in being the first content wiki to have this feature enabled. The conversation (my understanding of Swedish is limited) appears to be positive. T168808#3611698

While the tasks mentioned are being addressed I'll work on drafting an announcement for the Swedish Wikipedia. We'll then approach Swedish (and possibly Meta again) with a conversation about enabling the feature.

Assuming everything goes as anticipated, to build awareness, I'll post announcements to the appropriate talk pages (Village Pumps) and mailing lists to let contributors know of the new feature, reference documentation, and let folks know where to get help.

Additionally, it's recommended that as we work with the communities to enable this feature we use something like a notice on the Template namespace (English Wikipedia example) for a few weeks after enabling the feature to let logged-in contributors know of the feature - in case they miss the initial announcement. This is something I'm happy to help draft, but when possible would love to have community support for enabling.

After Meta and Swedish Wikipedia the Russian Wikipedia has shown interest, albeit with concerns over a particular technical issue. (T168333) If that can be resolved we can approach them next, with perhaps other interested wikis. If there is no clear interested communities we can approach similarly sized projects.

In T164927#3726517, @CKoerner_WMF wrote:

There are currently two related tasks that need to be addressed.

Is that "addressed" as in "discussed and potentially rejected", or "addressed" as in "we demand you do this, no matter how bad of an idea it is"?

I'm not demanding anything. Discussed and potentially rejected is a viable method of addressing the outstanding concerns.

If you're working on this then it probably doesn't belong to the backlog anymore :)

I will never be an agile scrum master. :)