Electrolysis/Meetings/2016-01-21

mconley

M8 bug 1238180 - Unsafe CPOW usage in nsContextMenu.js, browser.js for pageInfo for data URIs opened from context menu
- Fixed, landed on fx-team. This will break Page Info for sidebar browser, but sidebar browser appears to just be busted in general for many things.
M8 bug 1233803 - Get rid of of sessionHistory CPOW sent up via WebNavigation:SetHistory message
- Fixed, landed in mozilla-central, just uplifted to aurora.
M8 bug 1233497 - Make unsafe CPOWs from non-addon browser code throw by default
- We've decided to allow CPOWs within Task.jsm and Promise-backend.jsm for now
- Sent email to dev-platform and firefox-dev about this bug.
- Last Try push, these patches seem to have caused an intermittent to go perma-orange on dt6 (bug 1240044). There's a patch in that bug which I will now test.
Perf P1 bug 1223780 - [e10s][telemetry] regressions in EVENTLOOP_UI_LAG_EXP_MS (3 metrics: parent, children, parent + children)
- vladan argues that we care about that probe from the content process as well before we can close this one out. bug 1235908 makes that work, so I guess the position is to gather more data from the next Telemetry experiment to determine whether or not there's really a problem here.
Perf P1 bug 1195295 - content-sessionStore.js sends a sync message to the parent in SyncHandler.init
- Still waiting on bug 1220517 to get fixed. :ting and :froydnj are discussing potential fixes.

Finished converting newtab tests, waiting for review
Checked in a bunch of other fixed tests (places, search)
Expanded the browser-chrome test-tips page with more details (https://wiki.mozilla.org/Electrolysis/e10s_test_tips)

bug 1240164 (5% Win tpaint/sessionrestore regression caused by bug 1096804) - resolved. turns out taskbar thumbnailing code regresses these tests. we've lived with this in non-e10s since the original code landed. filed follow up bug 1241537.
bug 1232181 (M8, Replace plugin windows with static window captures when scrolling) - wip, currently figuring out gfx layer mechanics associated with the plugin layer.

Working on the 45beta experiment
"<kats> rough approximation: http://mzl.la/1ZOODfD - there's issues with scroll snapping, scrolling the sidebar in gmail, windowed flash plugins, a number of issues on HiDPI displays, an RTL bug"

recent test case work
- bug 1221159 - re-enabled a few working cases
- bug 1237713 - fixed this test case but need to circle back to modernize it once I have a better grasp of how to do that.
- bug 1240750 - fixed and modernized
- looking at others to work on per the e10s tests spreadsheet.
keeping up with needInfo's and pre-triage work

bug 1238964 - Printing via parent doesn't handle paper sizes correctly - patches up for review.
bug 1240437 - PushLayer and PopLayer APIs are not implemented for DrawTargetRecording - Bas said he might pick this up, but I'll probably start looking at it tomorrow if he hasn't.
bug 1173371 - [e10s] Web page is not shown when launch Firefox from network drive on Windows - have patch for chromium sandbox to allow the use of network drives in policy and also patch for gecko to add read access to the policy for the Fx program directory if on remote drive. Tested on WinXP, Vista, Win7, Win8.1 and Win10.

3 days PTO next week

bug 1234675 (m8, - disable e10s on beta rollout for users with addons) - several runs... final version should be ready for review, will land it this week.
bug 1239353 (m8, - Tearaway tab from lodpi to hidpi screen) - awefull... Layout and boxmodell plus the platform specific magic we do for this lodpi hidpi case is scary and new to me. I asked for help, and will continue debugging but honestly I'm not sure how long this will take if I have to figure out what is broken.
bug 1236926 (m8 - Remove the "you are now testing e10s" prompt) - uplifted
Have been helping out others in various other bugs.

bug 1234049 - fixed and landed. aurora uplift requested.
bug 1109714 - rss reader fix uplifted to aurora.
bug 1237025 - have a working patch but have some small review comments to fix up before landing. should land by the end of today and this is my last M8 (for now...)
bug 1179732 - tresize issue. as previously reported, this looks like the entire discrepancy is due to a startup delay and I am working on this now.