[16:38:32] bd808: Do you think we should allow shallow clones of core in mw vagrant? [16:38:38] As some config option [16:39:17] we have talked about that before. The negative is that pushing a patch to gerrit requires a deep clone [16:39:36] so as a dev platform full clones are better [16:40:01] but for getting up and running quickly our bloated mediawiki/core repo is not funn to bootstrap [16:41:11] Ah. It does? [16:41:20] I didn't realise that [16:41:54] I *think* it does [16:42:05] maybe some testing to make sure would be good [16:42:32] because if it doesn't then I'd be all for figuring out how to make the initial clones shallow [16:42:52] hm... if you don't use git review, it might work [16:42:56] but I haven't tried [16:43:04] Not some bug related to ooold git verisons? [16:43:28] anomie: About? (unrelated to above) [16:43:41] Reedy: ? [16:43:51] bd808: Are you talking about pusing via git-review or via plain git [16:43:52] Any idea why https://en.wikipedia.org/w/api.php?action=query&prop=links&titles=User%3aOd+Mishehu%2friver+stubs%2fM&pllimit=max&rawcontinue=1 returns no results? [16:43:58] https://stackoverflow.com/questions/6900103/why-cant-i-push-from-a-shallow-clone [16:43:59] * anomie looks [16:44:02] hoo: git-review [16:44:32] anomie: nvm [16:44:42] I just did a proper purge via the api with forcerecursivelinkupdate and it works [16:44:48] Wonder why normal purge via the page didn't fix it [16:45:14] Some caching bug I guess [16:45:57] Reedy: My guess is that pagelinks was empty for some reason. A normal purge doesn't fix that, but an API action=purge with forcelinkupdate will. So will a null edit, usually. [16:46:53] hoo: teaching new MW contributors to use git-review is hard enough. Learning how to do a git push to gerrit without it is not worth trying IMO [16:47:39] true that [16:48:20] I, for one, welcome our new Diffusion overlords [16:48:57] May I quote you on that once we are actually using it :P [16:49:14] No, you know full well I'm trolling [16:49:15] :P [16:55:26] anomie: thanks [16:56:38] I predict that we all hate Diffusion for 12 months, like it for 2 months, and then start whining about code review in general again [16:56:56] You mean Differential? ;) [16:57:17] * Reedy slaps legoktm [17:01:03] * anomie wonders why it's not called Diphpherential [17:03:02] haha [17:03:04] * legoktm quips [17:14:34] anomie, tgr: we are getting APi log data into kafka finally [17:14:46] bd808: Cool [17:14:54] \o/ [17:15:05] Now to find out if oozie is pulling it down to stat1002 [17:15:59] I was wondering whether the next statement was going to be that kafka was losing it somewhere. [17:27:45] csteipp: does https://phabricator.wikimedia.org/T130700 mean OATHAuth is going to go ahead without AuthManager? [17:27:45] legoktm: Yes, ish. [17:27:45] ish? :P [17:27:45] We're going ahead with the ugly solution that really shouldn't be done, but knowing that we'll clean up the tech debt with AuthManager. [17:28:34] hmm, okay [17:30:53] the point being to allow stewards to see what the UI/UX will be like as I understand it [17:32:31] Yep, we want to 1) make sure we're keeping our promise to the sysops who we told we would be deploying it this quarter; to ease the pain of getting them to adopt it. And 2, being able to work out the process issues with stewards/staff in an initial trial. Which is going to be much more work than the technology. [20:31:13] lolz -- http://www.theregister.co.uk/2016/03/23/npm_left_pad_chaos/ [22:00:17] urandom: they should've used python https://pypi.python.org/pypi/left-pad/0.0.3 :P