Notices / Schedule
- 3.6.18 is going to the beta audience today. If Christian contacts you about an issue found in beta, please jump all over it
- Firefox 5 is in the end game
- We built beta 6 yesterday for a fix on 10.7 (thanks RelEng!)
- We are pushing the build out to the beta audience right now (thanks QA and RelEng!)
- We will officially determine if our current Firefox 5 build is the RC today @ 2:00 pm PDT (Warp Core, x8605)
- Next source migration is July 5th. That means we're 1/2 way through the Aurora and Nightly periods for Firefox 6 and 7 respectively
- Looking at instrumentation of browser to dynamically assess cross over points between content/chrome, as a next step to inform e10s work
- Bill McCloskey has a working version of incremental GC.
- We're currently seeing 3% slowdown on benchmarks due to write barrier. Bill's working to track down where that's from and improve it.
- Pause times can be dialed all the way down below 1 ms without further slowdown.
- Bidi performance improvements have landed; performance loading Tinderbox logs should have improved.
- Work on bug 90268 (plugins owned by content) has been progressing well. Basically working on Mac OS X and Linux now, hopefully Windows by the end of the day. Have reserved the "holly" branch, if people can help test once the branch is producing builds that would be great.
- The DOM's JS_ClearScope() usage is going away bug 637099 (peterv)
- Drew needs help from gal/mrbkap/bz about dynamic instrumentation of chrome/content interaction, but has had no response to emails.
- dherman beginning the matching static analysis
- felipe is working on a build-time switch which will let Firefox start up with content processes (with some features disabled)
- multiple content processes has been identified as a primary goal, so that CPU-hog or misbehaving content will be somewhat isolated from other content. bsmedberg will be going through the remaining platform work required for this to make sure we understand the implications.
- DOWNTIME 4-8am PT Thursday announcement
- Some talos numbers expected to change. We will trigger extra jobs to get a new baseline.
- Trunk pretty stable - ~ 49K ADUs, 1.91 crashes per 100 ADU.
- Several regressions have surfaced over the past few days.
- Bug 663916 - Showed up in low volume for 4.0.1 and 5.0 but seems to be much higher on the trunk (131).
- Bug 663924 - Only a handful on 4.0.1 but rising number on the trunk since June 8th.
- nsHTMLTextAreaElement::AddRef() - not super high volume and some dups but new signature appearing on the 12th.
- Crash team is trying to focus on some of the harder top crashes appearing on 4.0.1 and get the tools/reports in place for better diagnosis.
- Tracking what happened since we turned off JS PGO - Bug 605033