[00:58:49] James_F: So, I have an old note here to make use the (remarkably simple and dull) guts of json-lint and use them in banana-checker. [00:59:08] But I see it seems eslint could do this instead. Which seems easier to manage indeed. [00:59:34] in that, it's more re-usable that way. [07:08:56] https://codesearch.wmflabs.org/core/?q=%40noinspection&i=nope&files=&repos= what tool is @noinspection used by? [09:44:21] hey folks, and especially duesen if you're here (but maybe you aren't), can someone look at https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/502150/ and give a thumbs up or what else is needed? I've live patched hosts for now [09:44:48] but that wil go away with the next deploy of course... so I'd like to get this merged and backported before today's deploy [09:45:12] ...because abstracts on enwiki (and I think commonswiki too) are broken without it [12:07:36] apergos: Ideally those pages should be moved to accessible titles, as you mentioned in T220424. https://www.mediawiki.org/wiki/Manual:NamespaceDupes.php seems likely to handle it for the cases mentioned in that task and in T220316. [12:07:37] T220424: XmlDUmpWriter::writeRevision sometimes broken by duplicate keys in Link Cache - https://phabricator.wikimedia.org/T220424 [12:07:38] T220316: XmlDumpWriter::openPage handles main namespace articles with prefixes that are namespace names AND are redirects incorrectly - https://phabricator.wikimedia.org/T220316 [12:08:13] right now (as usual) I'm looking for the short term fix so dumps can run; I've live patched he hosts but [12:08:31] in an hour during the next deploy those fixes will get overwritten [13:45:45] apergos: i was in meetings, sorry. [13:45:49] gave a +2 now [13:45:55] thanks! [13:46:09] I'll see if i can get someone to backport and deploy it before the train later today [13:46:17] * apergos looks at the swat schedule [13:48:44] hm no more swats today except puppet. but still if I can get the backport in, that should be enough, it will go around with the train [14:57:52] anomie: the READ_NEW patch doesn't need to get on the current train, right? [14:58:49] tgr: No, we already set the config explicitly on all WMF wikis. The main motivation is so that CI starts using the more modern setting when running tests to help guard against people introducing new breakages. [15:00:03] hm, I guess that means there is no point in waiting for the branch cut, either, since it gets overridden anyway [15:15:38] Krinkle: Hmm, yes, that could very much work. Did you file a task/issue? [15:16:53] James_F: i recall seeing a PR in github from you working on this [15:17:34] Krinkle: There's https://github.com/wikimedia/banana-checker/issues/22 but no activity since 2015. [15:18:09] Krinkle: And I filed https://phabricator.wikimedia.org/T220036 last week for giving up entirely on JSONlint. [17:02:33] James_F: ok found it in browsing history - https://github.com/wikimedia/eslint-config-wikimedia/pull/167 [17:02:48] Krinkle: Oh, yes, that. :-) [17:12:34] <_joe_> anomie: I took a stab at doing what you suggested for T219279, see https://gerrit.wikimedia.org/r/c/mediawiki/core/+/502546/ - I fully expect it to be completely non-working now, I might ask you a review when I feel more confident about it [17:12:37] T219279: Some pages will become completely unreachable after PHP7 update due to Unicode changes - https://phabricator.wikimedia.org/T219279 [17:12:51] <_joe_> my php is rusty as hell, and I've never really messed with core [17:24:58] _joe_: At first glance it seems ok, if a bit overcomplicated when you could just directly map strings to strings instead of doing mb_ord(), mapping number to number, and then doing an equivalent of mb_chr(). [17:31:07] cscott: btw, intentional that Parsing-Team is in subscribers as opposed to tags? (e.g. no workboard but e-mail notifs). Just checking :) [17:31:25] at https://phabricator.wikimedia.org/T216664 - noticed when browsing the board for errors that don't have a team tag. [17:43:17] <_joe_> anomie: I assumed we wanted no utf-8 characters in the code, but now I see it's not the case :P [18:20:13] James_F: kostajh: fresh-node10 bumped to 0.5.0 which fixes the Z_DATA_ERROR/'distance too far back' error you may've experienced on various repos previously. https://github.com/Krinkle/dotfiles/commit/b7dd4445a8c7 [18:21:34] ('networked-%' can talk to e.g. localhost: ports outside the container, 'root-%' means you'll log into the container as root, nothing scary :), e.g. when wanting to install additional apt-get packages or something ) [18:22:16] Krinkle: Thanks. [18:22:41] I don't remember seeing that error, but thanks anyway :) [18:23:47] kostajh: It was… not fun. We broke jsduck in half the repos. Whoops. [18:30:19] yay [20:27:29] Is there a way to properly use wfMessage() within a service? errr... what service can be injected the provides the same functionality? [20:31:35] davidwbarratt: The MessageLocalizer interface (also implemented by IContextSource) has a msg() method. I don't see a service for MessageLocalizer though, and IIRC duesen thinks IContextSource is too much of a kitchen sink to use with services. [20:32:27] yeah I was having trouble finding a service that I could inject rather than using the global function [20:48:31] shall I create a ticket for it?