Electrolysis/Meetings/2015-02-12

handyman

• bug 1130051 [e10s] Path Intersections of Paper.js demo is slow/laggy with e10s
  • billm and kats recommend the ‘revamping IPDL compress message’ solution
• bug 1124728 - Frequent OSX 10.6 T-e10s "timed out after 3600 seconds of no output"
  • started
• bug 1103248 - E10S - Flash Player Right-Click Context Menu Does Not Display
  • turned into Bug 1130976 - [e10s] Flash context menu causes CPOW deadlock with AdBlock Plus
  • Back-burner. This is a complex deadlock issue.
• bug 1118575 Far seek on YouTube spins forever in dev edition - but does not freeze the browser
  • Talking to media team.
• bug 1075670 [e10s] event.screenX and event.screenY is wrong
  • worked with jimm to finally make this land (patch-application failures)

gw280

• Did some bugzilla housekeeping - want to try and keep better track of all the spinner-related issues as there are a lot of them. Tracking bug: 1131167
• Currently looking at https://bugzilla.mozilla.org/show_bug.cgi?id=1066531 (tab switching feels sluggish). Can reproduce locally, and it's visible even when we're below the 300ms threshold (we lag by about 2-3 frames when switching the tab). The main issue here is that we're flipping the tab in the chrome before we render the new tab in the content area, anywhere from 2-3 frames before to a noticeable amount of time (I've seen it as much as 1/2 second).
• My profile of the above is a degenerate case where there were a lot of treeherder windows open and doing a lot of JS that was chewing up the content process and causing paint delays. We can't do anything about the computation contention but we can ensure that the tab switch and the spinner are shown atomically.
• I think the main issue here is that we're not gating the tab flip on mozAfterRemotePaint, but we are gating the tab content switch itself. I'm going to need to dig deeper into tabbrowser.xml as I don't think the current machinery allows for us to do that particularly easily.

==ejpbruel== (DevTools)

• Will join the meeting 15 minutes late due to scheduling conflict
• Not much news to report this week.
• Suggest upping bug 1093535 to M6

 * This bug is preventing debugger tests from being enabled. Although not yet reproduced in the wild, the behavior tested by those tests is probably broken, and looks like a race condition waiting to happen. Would like to have this fixed in Aurora.

gabor

• Implemented/landed GetPrototypeOf hook for CPOWProxyHandler for Bug 1090592
• Worked on Bug 1072350, patch is coming soon...
• Struggled with Bug 1118457, not too much progress
• Addon compat. talk with Dave and Bill

jimm

• bug 1103177 (native plugin windows torn out of tabs) - fixed
• bug 1030137 (touch scroll doesn't work with e10s) - fixed
• child aborts (KillHard bug 1116884, child aborts bug 1130734) - finished up with this.
  • Currently there are no major crashers in the wild.
  • daily automated report on child process aborts: http://www.mathies.com/mozilla/bug1116884.txt
  • Bob Owen just fixed bug 1094953 - the last major non-child abort content crash. Those signatures are now falling off the three day nicely.
• bug 1095761 (Plugin window focus routing) - wip

Mossop

• Almost done fixing SDK page-mods to not use shims or CPOWs
• Working with a 1password developer to identify their problems

mconley

• M5 - bug 1047603 - [e10s] Non-remote tabs in e10s windows do not handle target="_blank" or window.open links properly.
  • Landed and in today's Nightly. Finally.
  • Could use some QA help to verify that it fixes all of the (many) dupes
• M5 - bug 1077168 - Installing an app on marketplace fails with e10s
  • Got reviews, landed, and bounced from test failures. Have to re-jig a few things to make sure I don't re-open a security hole (bug 771294). Have a way forward, just cleaning up patches now and I need to clear some changes with the people that maintain the Webapps framework under dom/.
• M5 - bug 1090448 - Make e10s printing work on Linux
  • Patches got reviews from karlt, which I'm going over now.
• M5 - bug 1066812 - e10s: child process continues running if parent process crashes
  • This landed and then bounced because of test failures.
  • I've instrumented the patch to dump a log to tell the story of process shutdown which will hopefully shed some light on what's going on.
• M5 - bug 1090439 - PPrinting calls from child to parent via ShowProgress and ShowPrintDialog should not be sync
  • Have a proof-of-concept that works nicely, but I need to clean this up and rebase it on top of the changes that karlt is suggesting in bug 1090448.

blassey

• bug 1071880 - Notify user of addons that are slowing their browser down significantly - bounced again, this time for talos regressions

smacleod

• Landed ContentTask.spawn() for testing Bug 1107609
• Patch up to make about:privatebrowsing load remote Bug 1079245
• Working on getting all the privatebrowsing tests running in e10s
• Working on Bug 1094170 - seeing some strange network stuff, might be a deeper problem

mrbkap

• More work on bug 1117337 (do not load: crashes content process on load) fixing crash serializing URIs.
• Looked into my m5 bugs, one WFM.

felipe

back to e10s land! review list clear, send review requests my way

• bug 1105048 - ctp ui doesn't show up in google hangout's popup windows

« previous week | index | next week »