mozilla :: #content

10 Aug 2017
01:08Caspy7can I ask about the status of Quantum DOM? Is it still planned for 57?
02:45kanrumystor: ipdls are compiled in one pass so we actually has some global knowledge about that, I think..
03:01mystorkanru: hmm, neat
03:12kanrumystor: oh, just forget what I said :) only imported ipdls are read togther.
14:43jdmmmm, surprise reconfigure when building
16:22mystorIs anyone else having problems with searchfox just giving an empty document when searching?
16:36smaugmystor: seems to work here
16:36mystorsmaug: ok - yeah, it's working again for me now :-/
16:37mystorsmaug: I was trying to look up some wrapobject methods for the thing I ni?ed you on and couldn't
16:38smaugmystor: I'm mostly worried about the extra QI calls there ;)
16:39smaug(getting used to this irccloud clearly takes time, but Chatzilla isn't working with Nightly)
16:39mystor:-( that stinks
16:39mystor(irccloud sucks though, prepare to be disappointed)
16:39mystorsmaug: Fair, for some reason I had that method in my head as a fairly cold one where you don't have to be efficient
16:40mystorsmaug: Do you mind taking over review on that one?
16:53smaugmystor: I guess I could, though I have queue closed right now
16:54smaugjust tiny bit reviewing overload
16:54mystorsmaug: Fair
16:54smaugmystor: not sure how hot that code is
16:54smauggetElementById is rather hot
16:57mystorsmaug: fair
17:14bzmystor: which patch are you talking about?
17:15* bz hears getElementById, wrapobject, performance, gets interested
17:17bkellybz: sorry to ask, but do you think you will have time to review those two streams patches today or tomorrow? one is just an explanation of why the patch you already looked at should be safe and the other is just adding CC to a class
17:17bkellyI'd really like to get that bug landed (pref'd off of course)
17:19bkellyeven if the multi-global stuff is not resolved yet
17:19* bkelly goes to get lunch...
17:19mystorbz 834208
17:32bzbkelly: Yes, I am aiming to do it today
17:34bzmystor: hmm... where does getElementById come in?
17:34mystorbz: not sure
17:34mystorbz: I was in a meeting for part of the conversation and wasn't able to give it my full attention :S
17:35* bz ponders whether to pile on to this review, because this code has some not-pointed-out-yet issues
17:36mystorbz: g for it
18:07bzfrodynj: ping
18:07bzfroydnj: ping
18:11froydnjbz: pong
18:12firebotBug 1389009 NEW, stylo: check whether stylo performs large infallible allocations
18:12bzfroydnj: Can you point me to where the rust/jemalloc integration bits are?
18:13froydnjbz: there aren't any; Rust just calls the standard system allocation functions and aborts if those fail
18:17bzfroydnj: aha
18:17bzfroydnj: so it's going through jemalloc just because we tell ld that "malloc" is jemalloc's malloc?
18:17froydnjbz: right
18:17bzalright, thanks
18:41* bkelly is back.
18:44bkellyjdm: do you know if amy is on twitter?
18:44jdmbkelly: I do not
18:44bkellyI wanted to give her a shout out about cookie service
18:53smaugdo we have some telemetry data about how non-sync cookie handling has helped?
18:53bkellysmaug: there is some data in the bug saying its about 2000% better in profiles
18:53bzbkelly: done, sorry for the delay
18:53bzbkelly: I thought there was more to do there than there really was. :(
18:53smaugonly 2000% :)
18:53bkellybz: no problem... sorry for pinging you... and thanks!
18:54bzbkelly: no worries about the ping. Thank you for picking up these bits.
18:54firebotBug 1331680 FIXED, Consider not doing sync IPC for document.cookie getter/setter
18:54bzbkelly: that bug is ... not so readable. :(
18:55bzIt's not your fault
18:56jdmI refer to it as the sync ipc cookie experience
18:56bz"went from 314ms to .1168ms"
18:58bkellyI guess there is still some jank in some places:
18:59firebotBug 1232638 NEW, IPDL::PCookieService::RecvGetCookieString causing janks
19:01jdmseems unlikely
19:01jdmthat was filed 2 years ago
19:05smaugjdm: since you're there, how does async cookie handling work?
19:05jdmsmaug: what do you mean?
19:05smaugjdm: I mean, how did we make it async
19:05jdmsmaug: we keep a duplicate hashtable in the child and broadcast updates from the parent
19:06jdmdescribing it as async is misleading, for that reason
19:06smaugright. Ok, not-using-sync-ipc then
20:14smaugmccr8: wasn't there a bug to cut chrome->chrome leaks
20:14* smaug can't find it now
20:15mccr8smaug: yes! but basically it broke all devtools and they didn't care enough to fix it. erahm was working on it.
20:15firebotBug 1276366 UNCONFIRMED, Remove support for chrome -> chrome window leaks
20:32erahmwell it broke session restore too and some other rando stuff
20:36mystorfroydnj: Would you be OK with me splitting that first part into a few parts which each make one of the main changes, but where the intermediate states totally don't compile?
20:36froydnjmystor: as long as the parts get squashed back into one at commit, sure
20:37* froydnj has done that before, super convenient
20:37mystorfroydnj: The reason why that patch is relatively large is because changing HangStack causes ThreadStackHelper to break, and then you need to rewrite that, and then... and then...
21:12bkellybz: our webidl does not support default `= "value"` syntax in dictionaries?
21:19bzbkelly: it totally does
21:19bkellyis that new?
21:19bzbkelly: no....
21:19bkellywell ok, then
21:19bz has a bunch
21:19bkellyreview comments it is
21:20* bz wonders what bkelly is reviewing
21:20bkellybz: bug 1353636
21:20firebot ASSIGNED, Change ServiceWorker useCache to an enum and default to "imports"
21:21bzI wonder why they didn't just copy the spec idl
21:47billmbholley: ping
21:56bholleybillm: hi
21:56billmbholley: hey, sorry, nm. I got an answer to my question from someone else.
21:56bholleybillm: k
22:28smaugmstange: is there some easy way to see page loads in profiler UI?
22:28smaugI guess I need start of page load and when load event fires
11 Aug 2017
No messages
Last message: 67 days and 22 hours ago