[10:10:40] lunch [13:21:08] hmmm, looks like we had quite a few alerts for WDQS in CODFW, but they've cleared [13:25:19] inflatador: yes... seems like codfw is struggling with mem, old gc rate is high and jvmquake is probably doing its job [13:26:40] o/ [13:56:56] ACK...looks like the CODFW lag is OK so I think we can roll with it for now [13:57:42] yes... if this is too noisy I'll take a look at the queries [13:57:58] nah, I think we're OK for the time being [13:58:47] \o [14:03:32] o/ [14:08:18] dcausse: thanks for the review on the DWIM patch. I always appreciate it when you pick the nits! [14:11:16] Trey314159: I rarely find things obviously wrong in your code so the nits are the only thing I can actually find :P [14:11:28] but as always feel free to ignore them! :) [14:12:21] I'm a better computer scientist than software engineer, so I appreciate both the compliment and the suggestions! [15:48:30] hmm, curiously per-request namespaces list is always empty on commonswiki. I can look at special:search requests that use ns0=1 in the url, but curious namespaces is missing [15:48:41] (empty in the event.mediawiki_cirrussearch_request logs) [15:56:54] break, back in ~30 [16:32:16] dinner [16:46:48] back [16:51:46] hmm, deciding weights is annoying :P Looking at commons queries on web that return galleries, mean gallery position is 8.66. A weight of 0.9 moves the mean to 10.9, 0.5 gets 13.79, 0.2 gets 17.68. How far down are we trying to push them? unclear... [17:30:27] * ebernhardson reads the original user request, and finds what they actually care about is the relationship between galleries and categories, at least gives a benchmark [17:57:16] lunch, back in ~40 [18:00:25] * ebernhardson wonders what we should do something about that flakey smoke test in browser tests [20:22:28] sorry, been back awhile but need to step out for ~90m [21:54:06] back