Platform/2009-11-10
From MozillaWiki
< Platform
« previous week | index | next week »
Contents
Notices / Schedule
- shipped on Thursday
- thanks again to everyone who helped get that release out
Firefox 3.0.16 / Firefox 3.5.6
- Code freeze is tonight at 11:59pm!
- 3.5.6: 6 patches need to land, 11 need patches
- 3.0.16: 6 patches need to land, 4 need patches
Firefox 3.6 Beta
- Released on Friday, Oct 30th
- over 250,000 daily users
- beta 2 refresh shipping today:
- DLL blacklisting patch (bug 524904)
- syntax changes for CSS gradients (see bug 513395)
- cycle-collector patch to stop cycle collection instead of crashing (see bug 521750)
- over 190 other changes on mozilla-1.9.2 since we shipped first beta
- next refresh hoping to come soon with DLL component directory lockdown & whitelist (bug 519357
Firefox 3.6 Release Candidate
- proposed code freeze November 18th (1 weeks)
- proposed release November 26th (Happy Thanksgiving)
Blocker Report
See more Firefox 3.6 related blocker queries, or learn about the new status and blocker flags
- Release Blockers (flag: blocking1.9.2 or blocking-firefox3.6)
- 93 OPEN (-27 w/w)
- 39 FIXED but not yet fixed on mozilla-1.9.2 (-14 w/w)
- 52 nominations (+12 w/w)
- Handy charts: Blocker snapshots, Blocker and Noms trends
- Approvals
- 115 requests (+9 w/w)
- 11 approved but not yet fixed on mozilla-1.9.2 (-28 w/w)
- Charts: Nominations snapshots
Browser / Front End
(Progress reports every weekend on Planet Firefox)
- see our active projects and get involved / propose others
- Namoroka/mozilla-1.9.2 front end development:
- 21 blockers left, 7 are crashkill, 7 need patches
- not including crashkill and review-waiting, we're below 10 and will keep it there
- per tab prioritization for session restore bug 514490 landed
- windows 7 aero peek per-tab preview is now preffed-off as per bug 525475
GFX Update
- 13 blockers (-5 w/w)
- Long poles right now are imagelib bugs (needs review from someone who knows imagelib) and a couple of Linux bugs (which have no patches yet).
- 5 nominations
- All annotated with should block/should not block in whiteboard; need a driver to make the final call.
- Jeff is starting work on one of our quarterly goals, a scrolling interactivity benchmark. You can follow along in bug 527728.
- Bas has built upon his DirectWrite font backend, bug 517642, with a Direct2D cairo backend, bug 527707.
- Direct2D is a hardware-accelerated, quite Cairo-like API for Windows 7/Windows Vista, and it interoperates with Direct3D too.
- The Direct2D backend depends on DirectWrite for fonts because GDI can't write to hardware-accelerated contexts.
- This will be our second hardware-accelerated Cairo backend, after the OpenGL/DirectDraw backend used on the Tegra devices.
- Along with DirectWrite, Microsoft has backported Direct2D to Windows Vista.
- Rob's still working on the bug reports that have resulted in us turning Aero Peek off in 3.6.
- He needs UX input on a number of the bugs: bug 522035, bug 526618, bug 527015 (easy one), bug 527156, bug 527158.
- Jim's got an initial, prototype patch for cross-process windowless plugins on Windows. bug 522299
- Mark's WebGL implementation still coming along well.
- One issue is that WebGL arrays are different from JS arrays, and that can cause perf issues. Mark's got some ideas.
- Blockers, topcrashes, electrolysis, performance.
Layout Update
- Blocker report
- 1 nom (which will not block)
- 7 Blockers:
- 1 only needs branch landing (bug 501627)
- 1 needs trunk and branch landing (bug 527306)
- 1 needs review (bug 522787)
- 1 has a patch, but maybe needs a better fix/to be reassigned (bug 525856)
- 2 are frame poisoning improvements and probably should not block IMHO (bug 507294, bug 522088)
- 1 is a crashkill bug that is still being researched, probably should not block IMHO (bug 492675)
- N900 arrived!
- [roc] failed to fix httpd.js GC issues, still no GC happening
- VMWare record and replay much improved, still working on automation to help with orange analysis
Content Update
- 1.9.2
- 9 non-crashkill blockers
- 3 fixed or dups
- 6 bugs still being worked on
- 3 bugs still being worked on (bug 501934, bug 501934, bug 526500)
- Continuing the crashkill effort
- HTML5 parser update (hsivonen)
- async script loading for 1.9.2 (sicking)? (bug 503481)
Platform-specific Support Update
JS
Startup Performance
- Snapshot results for startup (and all other tests) are now available here.
- Data from Friday 11/6
- Once deployed to graphs.m.o (bug 527703), it'll be updated every 5 minutes.
- Latest weekly update is here.
- Taras and Joel have much more library IO data in bug 524202, and Joel is pushing forward with making the build more static in bug 525013.
- Ryan landed bug 507073, combining the two browser binary components that we ship into one.
- Peterv just landed bug 512645, removing the 10ms setTimeout wait in chrome.
- More details on the project page.
- Join us on IRC in #startup.
Security
feature | review date | who | interested |
---|---|---|---|
Windows TSF integration (1.9.2) | unscheduled | Jim Chen, roc | |
DNS Prefetching | unscheduled | Patrick McManus | Jesse, bz, reed(?), ctalbert |
New system metrics (and media queries) | unscheduled | ? | dbaron |
Electrolysis
- necko: We're close to having child process create HTTP channel in parent.
- windows deadlocks gone.
- landing plugins on trunk blocker tracker: bug 523094
Tree Management
- Talos downtime Wednesday 8am-11am PST to:
- bug 524169 Enable Talos on Electrolysis
- bug 506772, bug 524089 Enable V8, Tsvg tests for group opacity
- bug 523471 Stop running tiger talos on try builds that don't support tiger
- bug 526026 Update minidump_stackwalker
- ideas for reducing try talos wait times
- out of space in colo; local master-and-slaves, other colos
- stop running try-talos twice per job.
- bug 524047 Power maintenance: all mobile devices will be off weekend.
Roundtable
- (vlad) tinderbox+pushlog broken on Windows -- anyone know how to fix? Also, it's become a pretty important part of our infrastructure; can we get some dev time for improvements etc.?
- (beltzner) power's going out this weekend - tree policy?