Electrolysis/Meetings/2015-10-29

jimm

bug 1164508 (m8, fix UX related to hung scripts and plugins) - Landed last week but got backed out. A new test I've added experienced a random failure due to left over notifications from a different test. Currently trying to track the source of this down via try.
bug 1152049 (m8, plugins aren't clipped by scrollbars) - Making progress on this. Need to reshuffle the order of things in the CompositorParent::ComposeToTarget method since a type of clipping specific to scroll isn't being applied prior to when we process plugin clipping. Looks a little risky but mstange thinks it'll be ok.
bug 1153156 (m8, mousewheel scroll speed is not respecting custom system settings) - resolved WFM. Seems to be working, ni'd to masayuki to confirm.
bug 1129174 (m8, Plugin hangs can cause slow script UI to appear) - resolved WFM.
bug 1190364 (m8, plugin info sometimes fails with e10s) - no progress.

bug 1217876[e10s] nsIXMLHttpRequest clients in a frame script don't get any visible AuthPrompt when needed
Experimenting with Web APIs (e.g. WebSocket) in WebExtensions
Set up a local MozillaBuild env to start hacking on WebExtensions API
Tails Download & Verify Extension final rush

bug 1156742 - print to xps with low integrity sandbox - found a memory leak when playing the RecordedEvents, but when I fixed it it caused a crash with fonts. The problem was that we use a pointer to the data as part of the key when creating the custom font and so when that data was deleted the key became invalid (even though we'd already copied the data). As I was fixing this I realised that it was a similar issue that was causing the problem with recording WebFonts, so now I have that working as well. Should have patches up for review tomorrow or Monday.

M8 bug 1217190 - Printing is broken with e10s on OS X since a43f6bb86588
- I have a fix... but I'm not entirely sure why it fixes things. For some reason, that clang upgrade seems to affect the value of a bool when sent up to the parent as an out param. Looking into why to make sure there's not something really horribly busted here.
M8 bug 1209689 - Crashed tab indicates all tabs have crashed and every tab loads the crashed tab page
- Initially misunderstood the bug, but thankfully the work I'd done is required already to do this properly. So no waste.
- I have a patch, just need to address some more edge cases.
M8 bug 1171708 - [e10s] Stop using CPOWs for window closing
- Talked to billm earlier, and this seems pretty straight-forward. Wrote some experimental code, and I've found where the trickiness is. Currently designing a solution.
M8 bug 1177310 - [e10s] Stop using CPOWs on application shutdown
- Have not started this yet
I've been bombarded with interview stuff, so sorry if I've been slow to respond to things
Lots of reviews.

https://wiki.mozilla.org/Electrolysis/Test_Coverage#Individual_Test_Issues
- will track test my fixing progress here. Goal is to fix 12-15 test cases this quarter, focused on general directory.

bug 1171215 (m8, websockets don't send cookies if third-party cookies are disabled) - Patch that passes tryserver attached and waiting on review.
bug 1214666 (m8, about:looppanel doesn't work in remote browser) - Closed WFM (INVALID, really) after investigation.
bug 1113196 (m8, decide what to do with triggeringPrincipal in content processes) - Gave to tanvi since she's writing an equivalent patch in other code and has been working on figuring out what the right thing to do is.
bug 1147276 (m8, Google Calendar autocomplete breaks in e10s) - Closed WFM after discussing with cpeterson.
Sent email with tips for making tests e10s compatible (see also Electrolysis/e10s test tips).
- Will start a series of blog posts with examples.
Up next: getting Windows e10s tests green on try.

Strawman Rollout Plan:https://public.etherpad-mozilla.org/p/e10s-rollout
Update on crash data: bug 1218528 - [e10s] Provide a way to filter Socorro top crash tables based on DOMIPCEnabled annotation
Decision to decouple add-ons warnings/UX from e10s; perf team has more time to focus on measurements
Working on Beta Opt-In for 43.0 Beta 1 is a bit of a push for Felipe/Testing but we should be good for Beta 2. RelMan isn't really supportive of this so I need to work through the list of concerns and burn them down. 1218484 bug 1218484 - (e10s-beta) [Tracker] Happenings for e10s Beta
Vishy will be going through 'product wanted' tomorrow-friday
Need to land on A11y decision; hoping to make that an email
Working to port mana status to wiki for Friday
Working to get full list of per measurements included
Aurora Telemetry Experiment Happening; study ready ~Nov 9th if not sooner
We should just use public etherpad for this meeting, no?

bug 1196975 (m8, - stop using shims in the SDK) - reviewed Dave's patch about remote page-workers. started rebasing patches based on it, and adressing review comments. Dave also pointed out a case where we might hit shims unnecessarily, I have a patch for that too. Currently fixing test failures.
Have been collecting SDK bugs that we should dust off.

e10s Telemetry experiment is running this week
- http://bsmedberg.github.io/telemetry-experiments-dashboard/
- http://poiru.net/experiments.html
Birunthan is re-running Roberto's e10s vs non-e10s analysis on the preliminary experiment data
Session restore test is e10s compatible now
about:telemetry shows child process telemetry (will land today)

can the e10s team file some crashes from crash-stats? (search on dom ipc enabled) Many of us are traveling this month and next.
opt-in test in beta vs. telemetry questions