Electrolysis/Meetings/2015-12-03

« previous week | index | next week »

mconley

• M8 bug 1177310 - [e10s] Stop using CPOWs on application shutdown
  • Final reviews are in, and has landed on fx-team.
  • Hoping I can get some QA eyes on this, since we don't have an amazing test story for full browser restart (yet - see bug 1228120). Either manually banging on it, or helping me keep an eye on Bugzilla and the support channels about any lost sessions or crashes on shutdown.
• M8 bug 1159827 - [e10s] We always show the spinner when closing a tab
  • Unable to reproduce this. Blocked waiting on STR.
• M8 bug 1227312 - Investigate: many more content process crashes than aborts should not be possible.
  • Reasonably certain I found cause - we were double-counting the number of crash reports we were gathering (details). Handed off to jimm because it's touchy code that he's familiar with.
• Perf P1 bug 1193838 - Make it possible to gather profiles from processes that have exited
  • This landed on inbound today
• Perf P1 bug 1195295 - content-sessionStore.js sends a sync message to the parent in SyncHandler.init
  • Patch written and reviewed. Waiting on bug 1177310 to land and stick.
• Perf P1 bug 1221846 - Get Task Tracer working on desktop
  • Almost got this building - just dealing with a linker error.
  • I've got local modifications in the Gecko Profiler add-on to let us toggle this feature.
  • Talked to cervantes in #perf, and we're going to have a chat in Orlando - seems that the UI for interpreting Task Tracer logs isn't amazing for analysis, and they want to try building it into Cleopatra instead.
• bug 1225921 - Async tab and window flushing may make it possible to accidentally save tabs we want to forget
  • Tests and potential patch written. Waiting on review feedback.
• bug 1226333 - Add tests for async window flushing
  • Been doing lots of try pushes to get the intermittents figured out. Seems to be related to the increasing ID that we send from the child so that the parent doesn't accidentally overwrite synchronously flushed data with asynchronously flushed data... interestingly, the intermittents go away with bug 1195295 landed, but I'd still like to understand why this is happening.

bobowen

• bug 1156742 - print to xps with low integrity sandbox - three patches left needing review, two with changes that I've discussed with Bas already and one that I'm not sure if he has looked at yet.
• bug 1173371 - Web page is not shown when launching Firefox from network drive on Windows - because of issues with a full update (bug 1228604) of the chromium code, it looks like I'll have to take the fix I need for this in isolation. Then I can look into the other changes it will require.
• bug 1198368 - [e10s] [ADBE 4044716] Playing Amazon Instant Video using Flash in NPAPI sandbox hangs the chrome process - cpeterson only saw this on Amazon and they have stopped using flash, so he closed it.
• bug 1189846 - Print Edit 15.10 add-on no longer works with e10s and sandboxing enabled - need to look into adding some way of initiating the print from the parent with settings that we can add a RemotePrintJob into.
• bug 1217185 - e10s video performance is very bad - looks like this is being caused by USER_INTERACTIVE of JOB_INTERACTIVE, investigating further.

vladan

• Beta 43 experiment scrapped, to be run on Beta 44 bug 1229104 instead
• We'll try to get the blockers of the bug in time for beta1, but these bugs will not block or delay the Beta 44 experiment
  • bug 1198209: Measuring overall memory usage
  • bug 1228437: BHR reporting isn't working in the e10s child process
  • bug 1224374: BHR pseudo-stack coverage needs to be expanded
  • bug 1228147: Add telemetry probe for synchronous scroll performance
  • bug 1225074: Derived dataset for e10s experiments
• bug 1230196: Release channel should support e10s prefs

jimm

• bug 1227312 (M8, Crash report telemetry data is off for content process crashes) - under review
• bug 1222662 (M8, Mouse cursor disappears after moving it outside the Flash video area) - patch landed but reopened, new fix under review.
• bug 1198459 (M8, Disable a11y in content when we detect e10s is enabled and disable e10s on startup if a11y recently loaded) - under review
• bug 1227515 (M8, e10s testing notification triggers in browser chrome test runs, causing failures) - under review
• bug 1096804 (M8, Support windows taskbar previews and thumbnails) - under review
• bug 1226040 (M8, dragging images sometimes works, sometimes doesn't) - probably greasemonkey related, not sure what to do here.
• bug 1190364 (M8, plugincheck often fails to find any plugins) - haven't been back to look at this.

gw280

• bug 1109714 - rss bug. got a patch up for review last week, got a LOT of review comments back. addressed the first wave of comments, flagged new patch for review, got another wave of review comments back. will deal with those today.
• bug 1227373 - regression from my patch landed with bug 1098131. actively investigating.
• bug 1186585 - tab switch times regression. identified the bugs that caused the discrete jumps in switch time. the biggest one is not actually a regression - it was a fix that benwa landed to ensure we were measuring the right thing :(
• general perf work - mostly spent time getting familiar with the latest hotness in perf tools.

elan

• \o/ Status: https://wiki.mozilla.org/E10s/Status/Nov27
• APZ go/no-go - will ride trains to beta
• New product person: Jeff Griffiths
• Criteria doc round after one more pass with perf team
• bug 1218484 - (e10s-beta) [Tracker] Happenings for e10s Beta
• bug 1218484 - (e10s-beta) [meta] Rollout of e10s to beta 45 users (no addons, no a11y)
• Filed bug 1229093 - [e10s] Identify Selection Process for Users with Compatible Add-On
• bug 1229947 - [meta] e10s + APZ Gradual Roll Out to GA

mrbkap

• Landed patch for bug 1171215 (m8, websocket request doesn't send cookies if 3rd party cookies are disabled).
• Attached patch for bug 1116478 (m8, cmd+click on mailto: link opens new blank tab in addition to tab with web handler). Under review.
• Last of my m8s!
• I have a bunch of needinfos to go through.