Firefox/Channels/Meetings/2016-10-11

« previous week | index | next week »

Channel Meeting Details

• Tuesdays and Thursdays at 10am Pacific Time
• irc.mozilla.org #planning for backchannel
• Past meeting notes: https://wiki.mozilla.org/Firefox/Channels/Meetings

Video/Teleconference Details - NEW

• 650-903-0800 or 650-215-1282 x92 Conf# 99951 (US/INTL)
• 1-800-707-2533 (pin 369) Conf# 99951 (US)
• Vidyo Room: ReleaseCoordination
• Vidyo Guest URL

REMEMBER

These notes are read by people who weren't able to attend the meeting. Please make sure to include links and context so they can be understood.

Attendees

Johnny, ritu, milan, lizzard, marcia, erin, rail, martin, marco, blassey, shell, jcristau, kbrosnan, dveditz

Schedule Update

• Fx50
  • 50.0b6 (desktop) push to Beta channel blocked by bug 1309252
  • 50.0b6 Fennec signed off

• (liz) will evaluate async rendering system addon today

Telemetry

[marco] not sure where to put this Telemetry alert http://alerts.telemetry.mozilla.org/index.html#/detectors/1/metrics/1719/alerts/?from=2016-09-23&to=2016-09-23

            looks like more people are disabling e10s than before

Stability

Aurora

[marcia]

   NotifyInitializeSpies signature is new, will have to be filed.

Beta

overall rate: 1.23 (browser: 0.99, content: 0.24)

https://bugzilla.mozilla.org/show_bug.cgi?id=1292311 - Crash in nvwgf2um.dll | NDXGI::CDevice::DestroyDriverInstance https://bugzilla.mozilla.org/show_bug.cgi?id=1308863 - Crash in std::list<T>::clear() from D2D1::EndDraw() https://bugzilla.mozilla.org/show_bug.cgi?id=1299520 - Crash in NDXGI::CDevice::SignalSynchronizationObjectCB https://bugzilla.mozilla.org/show_bug.cgi?id=1307543 - Regression in D3D11DeviceContext crashes with 50.0b

• • This may have helped a little bit. (https://bugzilla.mozilla.org/show_bug.cgi?id=1291084 improved things for Intel devices)

https://bugzilla.mozilla.org/show_bug.cgi?id=1294748 - Crash in nvwgf2umx.dll & nvd3dumx.dll

• • https://bugzilla.mozilla.org/show_bug.cgi?id=1309225 should improve clustering of these crashes ^

Bug 1290403 - Crash in chtbrkg.dll@0x2e1f0

• Brad will help coordinate a fix for this in pre-beta channels.

AsyncShutdownTimeout | profile-before-change | MediaStreamGraph shutdown: blocking on msg thread was fixed in 50.0b5 (https://bugzilla.mozilla.org/show_bug.cgi?id=1255737 )

Release

overall rate: 0.93 (browser: 0.76, content: 0.17)

Nightly

[marcia]

• New bugs that have emerged:

    https://bugzilla.mozilla.org/show_bug.cgi?id=1308603 - Pretty high volume for Nightly -   Patch under review to add some debug assertions to try to figure out the crash.

   New crash in Core | Disability -> https://bugzilla.mozilla.org/show_bug.cgi?id=1309236

   webrtc::ViECaptureImpl::GetCaptureDevice is new and will need to filed, small volume

   Project Uptime crashes: https://wiki.mozilla.org/Platform/Uptime/NightlyCrashTriage/2016Q4

   Tab switch spinner 52 (mconley)

   mccr8 has a WIP patch to force painting during CC at https://bugzilla.mozilla.org/show_bug.cgi?id=1306728

   billm landed patches up to force painting during GC at https://bugzilla.mozilla.org/show_bug.cgi?id=1308039

   mconley suspects we’ve already got lots of data about these hangs, since BHR and (believe it or not) chromehangs works in the content process. There’s no way to differentiate these hangs from other hangs though, so currently looking at how we can use the annotation system to make them easier to spot and analyze.

   We’ll be able to get full native stacks on Windows on Nightly via chromehangs. Pseudostacks only for everybody else, but billm agrees that this is better than nothing.

   mconley is slowly putting together a dashboard to get a sense of spinner severity per client. See https://sql.telemetry.mozilla.org/queries/1355/source

Mobile

[marcia]

   Nothing new seen in brief time I looked at crash-stats

QE

• [desktop] [Cornel] WebRTC full duplex testing on Windows almost completed - need to wrap up the results and ask Maire for a feedback
• [desktop] [Cornel] We have reports that Firefox uninstalls after Windows 10 update (bug 1307418 and bug 1307413) - currently under investigation
  • initial Twitter report: https://twitter.com/ShaquilleRay/status/782499119596052480
  • SV was unable to repro this with their in-house testing

Roundtable

• websense : crash sig has stayed fairly low on 49.0.1. Still substantial ADI on 47.0.1 with websense unknown.
• other system addon possibility right now; graphics artifacts from d3d9 fallback+e10s (5% of users seeing this on release)
• facebook stereo sound (VR related)
  • Bug 1301518 - Opus 255 channel mapping family not playable
  • MBest will help from a product point of view.

• Plans/mechanics for 50.1 release?
  • What kinds of fixes will be planned for this dot release? Just security, stability.
  • ESR nomenclature: just another +.1 or +.0.1?