[08:54:48] Pchelolo: sure, thanks! [14:17:36] given serviceops is committed to php 7.4 update next Q, could PET have a look on T291127 at some point please? [14:17:37] T291127: MediaWiki session management does not work on split php7.3 & php7.4 cluster - https://phabricator.wikimedia.org/T291127 [14:23:12] lol. I see brad filed a followup to the upstream PHP bug https://bugs.php.net/bug.php?id=79174 [14:24:13] the followup bug fixed issues with 7.4 working independently, but as far as I can see it's incompatible with 7.3 [14:26:29] Yeah, I wasn't saying it was the fix, just that the report came from him [14:29:51] You've tagged their workboard.. It should hopefully get triaged at their next meeting [15:21:25] Are we still using PHP's session storage or does sessionstore alleviate the problem for us? [15:32:20] We still store username stuff (CA) in cookies [15:32:35] It would look like (based on what's attached) that this only really fails for users with spaces in their name [21:06:02] legoktm: hey hey, is this [wip] for any specific reason? https://gerrit.wikimedia.org/r/c/mediawiki/core/+/720143 [21:06:23] you know why I'm asking 😈 [21:10:02] Amir1: the code is ready, but for some reason it wasn't properly parsing metadata when I was trying to test it locally [21:10:42] I think I didn't configure MW properly for djvu? it was never even reaching DjvuImage::retrieveMetaData() [21:10:54] but I ran out of time yesterday to debug further [21:11:17] sure, you have more than enough in your plate. Take your time. [21:11:27] just ping me once I'm okay to proceed [21:14:09] mostly just want to make sure someone tests it with all the different configuration variants before it lands...will try to get to it tomorrow [22:47:02] TimStarling: could use a second pair of eyes the output compression fix, given the astronical odds that you and Aaron both made a /different/ off-by-one error two lines apart in the same method during the same stack of commits. - https://gerrit.wikimedia.org/r/c/mediawiki/core/+/721054 [22:47:15] but I suppose it's possible, and it does seem to check out from local testing. [23:05:01] eh, we're both humans [23:07:39] really code review is meant to pick up this kind of thing [23:07:45] and testing [23:09:27] I would just merge it and do the nitpicks as a followup, you know Aaron often takes a long time to get back to CR comments [23:15:09] TimStarling: ack, I just want to test the negative case as well, which one did you have in mind where e.g. it is meant to no-op? [23:20:43] not sure I understand, are you asking me how to cover line 52? [23:21:15] according to the coverage report, none of it is covered -- we don't really have a framework for client-server tests [23:32:11] TimStarling: manual testing I mean [23:33:58] Special:Export should enter it, right? [23:34:09] yes [23:35:15] Special:Export should call wfResetOutputBuffers() which should immediately result in a call to OutputHandler::handle() [23:35:36] which should hit the no-op case, it shouldn't send headers