mozilla :: #devtools

11 Aug 2017
01:27KWiersozer0: I had to back you out because your change was conflicting with a merge from m-c to inbound :(
01:27KWiersofeel free to rebase and reland whenever :)
01:28zer0I merged before push, sigh.
01:28zer0Should I do it now?
01:29zer0KWierso: ^
01:29KWiersogo for it, I'm done :)
01:29KWiersoit was just a one-line conflict, but I didn't trust myself to resolve it correctly
01:31zer0KWierso: I don't get any conflict. Which file was the issue?
01:31zer0I fetched from inbound, should I fetch from m-c?
01:31KWiersoit was devtools/client/webconsole/webpack.config.js
01:31KWiersoyou should be fine on inbound
01:32KWiersoif you don't hit a conflict, feel free to just reland it :)
01:32KWiersothat'll be future-me's problem :P
01:32KWierso|afkcurrent-me is now afk :P
01:33zer0I'm going to bed too
01:37zer0(ok, got the conflict)
08:55soleGood morning everyone! did you see about the vulnerabilities found in git/hg/svn?!topic/
09:00jdescottesah nope, thanks for the heads up sole
09:02solebrew upgrade if in mac will set you up \o/
09:03soleapt-get update? in linux? I always forget
10:08julienwsole, it depends on your linux :D
10:08julienwsole, just say "upgrade" ;)
10:09julienwthx for the heads up !
10:11julienwlooks like my Debian doesn't have an updated package for mercurial yet
10:11julienwbut I use it for Mozilla only anyway
11:56zer0folks, what happened to the webconsole-listeners?
11:57zer0ok, it seems there was a folder refactoring.
11:58julienwpretty sure this is ochameau's work
12:06zer0I hope not, since he was the reviewer of a bug that uses this module. :P
12:08firebotBug 1382968 FIXED, console calls shouldn't be piped to the parent until it is any useful
12:11jdescottesthe files just moved I think? is this an issue?
12:13zer0I just got sudden failures in tests that I wasn't expecting.
12:17nchevobbeupdating hg messed up extensions (AttributeError: 'firefoxtreesrepo' object has no attribute 'opener' when I try to pull) :/
12:20jdescottesnchevobbe: try ./mach mercurial-setup maybe ?
12:20jdescottesI remember my firefoxtree extension got messed up during a previous hg update
12:21jdescottesbut I can't remember how I fixed it :)
12:21nchevobbeah thanks, I was looking for that command
12:24nchevobbejdescottes: thanks, I think that fixed the issue :)
12:24jdescottesnchevobbe: \o/
12:59jdescottesI feel like my `hg wip` is going crazy though oO, I keep seeing remote changes being displayed as local ones. I have to pull inbound|beta|etc... (even if no changes are pulled) to make it behave
13:12nchevobbejdescottes: seems to work fine for me
13:13nchevobbethe color-scheme nicely changed even :)
13:13jdescottesnchevobbe: ok thanks for checking. In my case it seems to be triggered when I do a rebase
13:24nchevobbeoh, bye bye TabGroups
13:48solejdescottes: found this post on code coverage from marco
13:48soleI have a feeling he's on holidays
13:48soleBut it's a cool explanation I feel
13:49jdescotteslooks like a great read, thanks for the link!
14:07soleI think I need to digest it more but it can set us in the right direction. I love all the data on numbers too
14:29solezer0: what's the situation on the patch? Any update on the repush?
14:30zer0sole, un
14:30soleI don't know how to read this - is it in m-c?
14:30zer0sole, the bug has two patches (two commit)
14:30zer0one is landed the one in the link above
14:31solethis is the one that changed the paths right?
14:31zer0the other one is reviewed and I'm checking try. Got failures for unexpected changes in the tree, fixed, re-push just on linux, got green, now I'm waiting try for the whole platforms.
14:31zer0yes, it's the one that change the current "event-emitter" in "old-event-emitter".
14:32solerighty ok!
14:32solethanks zer0
14:33zer0sole: np! I'll keep you posted
14:34soleI'm itching to fix more bugs and to move the project forward... we had such a nice momentum 2 weeks ago
14:35zer0I know. :) But no worries! This bug will be fixed before I'll leave in PTO, and when I'll be back in a week I'm pretty sure you folks fixed all the events bug. :)
14:56Honzanchevobbe: does devtools/client/webconsole/new-console-output/test/mochitest/browser_webconsole_console_dir.js pass for you?
14:57nchevobbeHonza: yes it does
14:57nchevobbeDo you see issues with you patch ?
14:58Honzathis test fails, but I tried it without my patch and it still fails
14:59nchevobbetried with --setenv MOZ_CHAOSMODE=true , still fine
15:16TazmainHi all, is there a way to see why something like a css animation, is running very slow and sluggish in firefox, vs chrome ?
15:18soleTazmain: the performance profiler can help you! julienw or gregtatum know more
15:20Tazmainsole, do I need to run that perf-html in firefox
15:22soleTazmain: you install an extension (that website will guide you) that will let you capture data while the slow website is running
15:22solethen you can inspect the recorded data in detail (it will let you see where the time is spent) or you can file a bug with that data if it's too much, and we'll try to make it fast
15:23Tazmainokay how do I read this exactly /
15:23Tazmainmain thread spent 3890ms on GC Major , is that a lot ?
15:24Honzanchevobbe: one more thing, can you run npm link in webconsole?
15:24soleTazmain: that's why I said if it's too much we can try and investigate
15:24Honzanchevobbe: is it all green for you?
15:24Honzanchevobbe: sorry, npm test
15:24soleTazmain: it does sound like a lot of time yeah
15:25solebut would need to look at it
15:25Tazmainsole, okay where do I upload this to? Should I use the gecko one ?
15:25nchevobbeHonza: Oh, no it's not, I seeexceptions due to openLink
15:26Honzanchevobbe: ok, I am seeing the same
15:26nchevobbeTazmain: you have a share button, it will give you a shareable link
15:27Tazmainnchevobbe, oh thought I had to open a bug tracker
15:27Tazmainhere is the profile
15:27TazmainI meant
15:27Honzanchevobbe: not sure what patch caused that, but must be pretty new
15:27Honzanchevobbe: I just pulled from m-c
15:27nchevobbeHonza: it's the URL one
15:27Honzanchevobbe: ok
15:27nchevobbeit's my fault
15:29nchevobbeHonza: so it appears to be all in evaluation-results.test.js
15:30nchevobbewe should add serviceContainer as a prop to every EvaluationResult call
15:31nchevobbeThe serviceContainer shim is already required in the file
15:31nchevobbeHonza: I'll file a bug and fix this
15:32Honzanchevobbe: I see, ok, thanks!
15:32soleTazmain: let me have a look
15:38soleTazmain: is the website public? it's hard to know what to look at :)
15:39solei.e. the code that looks so slow for you in firefox
15:41nchevobbeHonza: should be better with
15:43Honzanchevobbe: all green again, nice and quick!
15:47nchevobbeHonza: cool, I pushed to autoland
15:50Honzanchevobbe: I don't understand your last comment in bug 1385791
15:50firebot ASSIGNED, "Requests", "XHR" and "CSS" filtering are buggy
15:50Honzanchevobbe: should I re-generate the stubs?
15:51nchevobbeso, if you look at
15:51nchevobbewe assign a level of "LOG" for debug messages
15:51nchevobbeit should be DEBUG
15:52nchevobbethat's why you had to test message.type for matchDebugFilter (and not message.level like in others messages)
15:52nchevobbeso, we should make this line `debug: levels.LEVEL_DEBUG,`
15:53Honzanchevobbe: ah, I see
15:53nchevobbeand generate the stubs again so the correct message.type will be in the stub
15:53nchevobbethen, make the matchDebugFilter check message.level (or roll back to a single matchLevelFilter, which I found nicer)
15:53Honzanchevobbe: why 'count' is level 'debug'?
15:54Honzalog coming from count should be visible when 'log' filter is on, no?
15:55nchevobbeI think it would make more sense too
15:55Honzanchevobbe: cool
16:01Tazmainsole, you can login if you want
16:02Tazmainsole, the code that is making it slow is some css 'animation: dash 7s linear infinite;' that is applied
16:06soleTazmain: I think if you could provide a reduced sample, that would help (like a codepen that just applies that animation to an element... and is slow in firefox but not in chrome)
16:07Tazmainsole, hmmm, will look into it. otherwise you can just use my login lol
16:09Tazmainwell g2g
16:38tromeyI wonder if we could finally get rid of the old style editor actor
16:39tromeythe comment in the style editor says it is from before firefox 29
16:46soletromey: wait.. there is a NEW style editor actor??
16:47tromeyyeah, the old one is StyleEditor and the new one is StyleSheets
16:49tromeydo we have a way to tell when a new style sheet was added to the document?
16:49tromeyI am looking for a method like this somewhere and not finding it
16:51soletromey: so whenI go to the style editor tab... which actor is being used?
16:52tromeysole: it depends on what you are attached to, but unless you are connected to a very old firefox, you get StyleSheets
16:53tromeyjdescottes: aha, my problem is bug 1224558
16:53firebot NEW, StyleSheetsActor should use the parent tabActor to retrieve the list of windows and react to new/rem
16:53tromeywhat a rabbit hole source maps are
16:53jdescottesooooh I never fixed that
16:54tromeydo you have any preliminary hacks?
16:56jdescottestromey: no sorry :(
16:57tromeyno worries
16:57tromeywell, I am mildly worried, like always, but it'll be ok
16:58gloh, this would probably make moving the style editor into the debugger easier
16:58gli assume
17:00julienwsole, tazmain left but I think he probably animated things that animates badly
17:00julienwsole, reflows took forever :)
17:02solejulienw: I will admit I can't read profiles very well (yet) that's why I was pinging you two
17:03julienwyeah I missed the ping
17:04julienwsole, my first step for lags is identify the long red lines at the top
17:04julienwclick on it
17:04julienwand then focus on this
17:04solelet me see...
17:05soleI see .. part of it
17:05solejulienw: I have to go soonish
17:05soleI will ping you next week :)
17:05julienwsole, then here I see various things, like long GC and CC (this is the garbage collector) and long "display list" and "layer building"
17:05julienwlong display list and layer building means expensive reflows :)
17:05julienwlong GC and CC means a lot of objects are created
17:06julienwdifficult to know more without the code
17:06solejulienw: yup
17:07solethat's why I was asking for more info...
17:07soleAAAANYway I'm leaving for today ... I'll ping you monday!
17:07solejulienw: au revoir! or a bientot! or something :)
17:25julienwbye :)
17:25julienwa la semana proxima !
17:25julienwI need to practice my spanish before the lessons start in september
17:52wedrHello, does anyone know how to change the "aurora" branch shown here to the "beta" branch?
17:52wedrThis is the latest version right now.
18:04glwedr: you would need to download beta
18:15wedrgl: So it is not possible to keep the Dev Edition, despite that it is build for developers?
18:15glwedr: beta and deved are really separate branches and builds
18:16wedrgl: Oh? First time hearing this.
18:16wedrI had the impression beta and aurora (the dev builds) are to be merged or something.
18:16glwedr: yes indeed it's actually under a release cycle
18:18wedrOh ok
18:19wedrI'm just hoping the Beta build doesn't override my stable Firefox build.
18:20glwedr: it will not override your stable firefox build
18:35tromeyjryans: which channel should I use for stylo questions?
18:35jryanstromey: #servo is best
18:37wedrgl: Thanks.
18:40wedrgl: I see no differences between Beta and stable Firefox. They even use the same installation paths. :/
18:40wedrThis is pretty bad...
18:42wedrah, seems like the link you gave mixed up the url.
18:42wedrok, my fault for jumping to conclusions.
21:39jimbbsmedberg asks:
21:39jimb"I've been trying to find out some technical details about the devtools plan, but my initial understanding is that they are trying to target stable web/webextensions/debugger API surfaces, and so they *shouldn't* be affected by gecko internals changes. But I'd be a lot more comfortable if that were in writing as part of the devtools plan."
21:40jimbCould someone here reply to that thread and give him a summary of what privileged APIs the broken-out devtools are going to use?
21:40jimbI believe it's restricted to what we need to connect to the server, with the server remaining the only thing that has its hands dirty with lots of Gecko-specific APIs, and hence remaining in-tree.
21:40jimbbut I'm not positive
21:41jimband misinforming is worse than leaving someone with the correct impression that they don't know something
12 Aug 2017
No messages
Last message: 12 days ago