(10:01:49 AM) mvolz: [13:55:14] mobrovac: I think we need to chat about response times. I'm not sure how to go about fixing this but in the morning citoid was basically unuseable for me because the extension kept timing out: https://phabricator.wikimedia.org/T162886
(10:01:49 AM) mvolz: [13:55:25] It's a really bad user experience.
(10:01:49 AM) mvolz: [13:55:42] Is there a way maybe we can force citoid to respond within a certain amount of time?
(10:01:49 AM) mvolz: [13:55:55] That matches the time out of the extension?
(10:01:49 AM) mvolz: [13:56:46] I'm not sure how I would begin to go about doing that. 
(10:07:20 AM) mobrovac: mvolz: all the services in that DC alerted today, so it might not be citoid-specific
(10:07:34 AM) mvolz: hmm, okay :)
(10:08:35 AM) mobrovac: mvolz: but generally speaking, citoid's problem is mostly zotero, as it takes a long time to respond
(10:09:09 AM) mvolz: yeah, I'm not sure what proportion of that is coming from pubmed though
(10:09:35 AM) mvolz: we make a request to pubmed for every item that has a doi, in order to add a pmid to it
(10:09:43 AM) mvolz: I was thinking of configuring it to disable it.
(10:10:00 AM) mvolz: what do you think? Worth trying? Is there a way to benchmark this somehow?
(10:12:22 AM) mobrovac: mvolz: hm but if we disable that then we'll start getting complaints (again), as the expected functionality would not be there
(10:12:53 AM) mobrovac: mvolz: one idea could be to make parallel requests - zotero, htmldata and pubmed
(10:14:50 AM) mvolz: well, imo the pmcid and pmc when you already have a doi is extraneous. But I guess that's why I was wondering if we could benchmark. See if it saves us time
(10:15:03 AM) mvolz: The problem with parallelising it is that we can't always do that
(10:15:19 AM) mvolz: If we have the doi to begin with then possibly
(10:15:26 AM) mvolz: but if we get the doi from scraping not so much
(10:16:00 AM) mvolz: But yeah let's try paralellising it when we can instead of doing it in exporter maybe.?
(10:16:30 AM) mvolz: And then if we get the doi too late just skip it adding the pmid? 
(10:17:16 AM) mobrovac: sounds like a plan
(10:17:34 AM) mobrovac: if we get the doi late, that could be a config option
(10:17:40 AM) mobrovac: and then we play with it