mozilla :: #treeherder

16 Mar 2017
11:01bcFailure classification tab on staging is looking nice..
11:13emorleyjgraham: ^ :-)
11:17jgrahambc|afk: \o/
12:17bcjgraham: Ok to play around with the classification tab on staging? I won't screw anything up by using it?
12:30emorleyit's completely separate from prod, so go ahead :-)
14:09jgrahamIs the procedure now that you have to r+ on GitHub and bugzilla?
14:28wlachjgraham: it appears so
14:29wlachI'd rather we ditch the github review requirement tbh
14:30jgrahamI certainly think that having to do two things is very annoying. And I guess there wouldn't be much support for ditching bugzilla
14:31wlachyeah, I don't think so. I find bugzilla's review request and needinfo pretty core to my workflow at this point
14:31jgrahamHaving a single dashboard of "this is what you have to do" is pretty useful
14:32jgraham(it's just a pity that bugzilla has such bad workflow around setting and clearing the flags)
14:33jgrahamWell clearing needinfo is fine, but setting and requesting r is unnecessarily hard
15:03wlach|mtgyeah it is
15:05emorleyI've turned off the GitHub review requirement for now, until we get a bot that can update the Buzgilla side for us
15:06wlachthanks emorley
15:06emorleythe feature I was more interested in was "Require status checks to pass before merging", but that can be enabled separately
15:06jgrahamemorley: Thanks (I apparently said "very" annoying but was thinking "somewhat" annoying, sorry about that)
15:06emorleyit had been annoying me too to be fair
15:06emorleyI enabled it to see how it would work out
15:06jgrahamYeah, the status check check is good
15:07emorleyand it was fine for longer reviews (where you're going to be commenting on github anyway)
15:07emorleybut for shorter it's a lot of overhead
15:27wlachemorley: I can see the pulse queues fine now, thanks!
15:27emorleyyw
15:52wlachjgraham: do you have a tracking bug for ui nits related to the new autoclassify panel? if not, could we create one? I'm tempted to file some bugs, but don't want them to get lost
15:59herokubotdeployed 584bc84 to treeherder-stage
16:02herokubotdeployed 584bc84 to treeherder-prototype
16:02jgrahamwlach: Don't have a tracking bug, but I can file one
16:03jgrahamhttps://bugzilla.mozilla.org/show_bug.cgi?id=1347974
16:03firebotBug 1347974 NEW, nobody@mozilla.org [Meta] UI improvements for the Failure Classification panel
16:09jgrahamSo the problem is bug 1347726 appears to be that when an element gets disabled it eats key events
16:09firebothttps://bugzil.la/1347726 NEW, nobody@mozilla.org Clicking "save all" button in new autoclassify UI breaks n/p shortcuts
16:09jgrahamThose buttons have prevent-default-on-left-click, and I can update that to element.blur(), but I don't know if that would have adverse effects elsewhere
16:09jgrahamHas anyone got another suggestion
16:18jgrahamemorley: https://public-artifacts.taskcluster.net/EnHkkQP8RN-7mJf9ZHo4Ug/0/public/test_info//reftest-no-accel_errorsummary.log seems to be entirely ahal's manifest additions
16:18jgrahamNo idea how it could take 57s to download though
16:19jgrahamSame with the other example
16:19jgrahamIf that's what's taking all the time then compression seems like a good first step
16:29emorleyjgraham: yeah I think it's a combination of (1) queue.tc.net was having a bad day, (2) no gzip, (3) the unfortunate redirect situation for queue.tc.net urls, (4) the file being large due to the recent additions
16:29emorleyI'm filing a bug for (3) shortly
16:30emorley400ms to redirect for a 40ms download :-/
16:30emorleywell 70ms
16:30emorley(not for that URL, looking at another)
16:34jgrahamemorley: OK, thanks. I infer that there isn't much for me to do in the short term?
17:14emorleycamd: Eli https://docs.google.com/document/d/1n2K7n1GSMLANbljgE8BvrzU5dwJQoem8DrA3ASEcNbE/edit
17:17emorleycamd: Eli: ah this is the correct link https://docs.google.com/document/d/1jY7-bCR-ra7jnC06YvGySe_Fqb8lw3dyizNOYPluyOk/edit
17:17emorleycamd: Eli: on there it lists other projects under that item, rather than this one
18:01herokubotdeployed dfb8263 to treeherder-stage
18:01herokubotdeployed dfb8263 to treeherder-prototype
18:38KWiersoemorley: ping
18:39KWiersoor camd: ping
18:39emorleyKWierso: https://blogs.gnome.org/markmc/2014/02/20/naked-pings/
18:40KWiersoI'm trying to fix a linting failure https://travis-ci.org/mozilla/treeherder/jobs/210806547 for one of my PRs
18:40KWiersothat first error wanting me to switch to ===
18:40KWiersobut in this case, I *do* want to use ==
18:41camdKWierso: you can use an eslint exception like:
18:41camd/ eslint-disable-line no-param-reassign
18:41camdexcept you'll put whatever the name of your error in place of no-param-reassign
18:42KWiersoany in-tree examples?
18:42KWiersogithub search isn't showing anything for 'eslint-disable-line'
18:42emorley99% of the time, having to add an exception is a code smell IMO
18:43AryxKWierso: https://dxr.mozilla.org/mozilla-central/source/browser/components/extensions/ext-contextMenus.js#214
18:43emorleyAlso, this question could likely have been asked on a PR, or just in channel without a ping
18:43camdKWierso: yeah, we almost never do it. But I have a couple in the mozilla/treeherder-manifest repo
18:43emorleyasync ftw :-)
18:44KWiersoemorley: true
18:44KWiersoin this case, I'm doing this: https://pastebin.mozilla.org/8982254
18:44KWiersotaking a user input that's expecting an integer (bug ID), and trying to make sure it's a valid integer
18:44KWiersoopen to smarter options :)
18:47KWiersoactually...
18:47KWiersoemorley, camd: unping
18:50jgrahamKWierso: /^[0-9]*$/.test("123") seems like it would be enough to validate an integer (I would suggest parseInt, but ofc it's insane)
19:35davehuntafter it's been running for a while, I eventually always get Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' (111) on Treeherder running locally in vagrant
19:36davehuntanyone else?
19:42emorleydavehunt: run `sudo service mysql status` -> I bet it says it's stopped, and I bet your machine still has 2GB RAM (check in virtualbox)
19:43davehuntemorley: I'll try that next time (a vagrant halt/up fixes it)
19:43emorleydavehunt: still check the RAM
19:43emorleyif it's 2GB, run `vagrant reload` to pick up the recent 2GB->3GB change (presuming your on or near master)
19:43emorleyyou're
19:44davehuntemorley: Base Memory: 3072 MB
19:50davehuntgtg, night all
19:56herokubotdeployed 296ecb1 to treeherder-prototype
19:56herokubotdeployed 296ecb1 to treeherder-stage
20:04emorleydavehunt: ah good :-)
22:25Elicwill: ping
22:55Eliemorley: you still around?
23:35camdEli: do you know how to redirect on a Route if some query params don't exist?
23:35camdI&#39;m looking at <Route render=... But it seems pretty clunky. Maybe it&#39;s the only way though...
23:37Elicamd: <Redirect to=&quot;&quot; />
23:37Elihttps://reacttraining.com/react-router/web/api/Redirect
23:37camdthanks, looking. :)
23:40camdok, yeah, that&#39;s about what I was seeing.
23:40camdEli: thanks
23:40Elinp :)
23:42Eliyou can also tweak the regexes used to match the routes
23:42Eliusing the exact prop, and then sending to another route that matches
23:42camdoh yeah, I forgot about that. That might work well, too
23:43Elihttps://github.com/mozilla-rpweb/treeherder-manifest/blob/master/src/index.js#L18-L19
23:43camdThough I *try* to avoid regexes as much as possible... :)
23:49camdEli: actually, I should redirect if they provide a repo and/or revision that doesn&#39;t exist. Do you know how to redirect in just the class?
23:49camdor do I need an if..then in the render() function?
23:50camdI know there is/was a ``transition`` object that was passed as a param to some functions... But v4 changed a lot :)
17 Mar 2017
No messages
   
Last message: 190 days and 8 hours ago