Re-ran on testwiki:

There was a conflicting Wikimedia event for product managers Tuesday. A PM mentioned they would have liked to attend a session, but could not because of this.

Full log is at P6624. Summary below (the grep excludes gomwiki, due to unrelated T154623: Workflow ID null for some user talk pages on gomwiki). The other two errors are also unrelated, going back to P4715.

@Wong128hk We apologize for this issue. It is fixed now, but temporarily you will not be able to adjust the Beta Feature. We will post here when you can.

In T184670#3910724, @SBisson wrote:

1. Repair inconsistent user talk pages in production

I think at one point we didn't include JavaScript on secondary pages (which are intended for no-JS users anyway) to avoid such issues.

Hiding the preference until the fix may make sense.

I'm not sure why this truncates at all (it's a rendered version, not a user input limitation). If there's some reason to, it's way too short (and other untruncated lines are longer).

See my comment at T183319#3908502 , also relevant to "Building partnerships with other Open Source communities on shared interests (e.g. translation, audio, video)".

See position papers by Birgit Müller and myself.

How can we influence the larger OSS community? (Are there other external tools we should upstream to, or participate in, etc)
[...]
How can we outreach and market our open source code and tooling
Can this strategy help in scaling the contributor experience? Allowing more contributors participate in wider-impact tools for the OSS community at large?

In T171577#3561537, @Qgil wrote:

A related question is whether scripts/gadgets can be considered a good entry point for new developers.

Apparently caused by rMW14ee3f210782: rdbms: specify DB name and table prefix even for the local domain.

@aaron Am I right that having two databases with the same name (i.e. "enwiki where revision, page, etc. are" and "enwiki where blobs is") on the same machine requires two different MySQL data directories and ports/sockets?

loginwiki is special because almost no extensions run there. That makes it a good place to catch missing dependencies like this.

In T182241#3869755, @Mooeypoo wrote:

Yes, there's a funny (not) bug/feature in mw.msg regarding treating the number 0 as plural, so {{PLURAL:$1|change|$1 changes}} will be 'change' if it's only one, and '0 changes' if it's zero or bigger than 1.

We can use an epsilon. When displaying that drop-down, if the value is close enough (e.g. .01 hours away from the pre-defined value), use the pre-defined value (e.g. 1 hour) instead of a custom value (e.g. 0.99984 h).

I've edited the description to use @Schnark's proposal.

In T171563#3469057, @Schnark wrote:

Instead, I'd like to propose that JS/CSS pages outside MediaWiki: or User: namespace (or other namespaces like Gadget: where they are protected automatically) should be deprecated, and MW finally should stop to send them with a text/javascript MIME type even if requested so, and additionally set the X-Content-Type-Options: nosniff header to make browsers accept that and not to execute such scripts.

In T171848#3848321, @zeljkofilipin wrote:

@Mattflaschen-WMF go ahead and delete the tests. @Etonkovidova and I were just trying out Selenium for Echo. I have abandoned the patch.

In T181545#3850212, @Jdlrobson wrote:

I ended up proposing a new hook so that MobileFrontend can disable RCFilters for mobile views.

Are there any alternatives to doing it this way?

There are already permalinks, but those help you find the topic or post again.

I can reproduce it on my Galaxy S4 (Android, Chrome 63), which has hover support (when your finger hovers).

Let's also change all the other 'comment' uses to 'post'.

@Etonkovidova @zeljkofilipin Were you working on this, or should we just go ahead and remove them per the meeting?

It has gotten flakier lately, but I think that is due to Beta instability, not any issue with GettingStarted or the test.

@Etonkovidova @SBisson @Mooeypoo @Catrope I would like to make an exception and migrate this one, since it is actively monitored by @phuedx and I, and the test has caught a serious outage in the account creation workflow (though the response to that should have been better).

In T132650#3185853, @Steinsplitter wrote:

In T132650#3185630, @Fastily wrote:

How could this possibly work for Wikis that permit fair-use? Concerned there will be *many* false alarms.

I think we talk about non fair use wikis here.

I agree with @Etonkovidova.