Platform/2009-11-03: Difference between revisions

 
(29 intermediate revisions by 12 users not shown)
Line 5: Line 5:


'''[[Releases/Firefox 3.0.16|Firefox 3.0.16]] / [[Releases/Firefox_3.5.6|Firefox 3.5.6]]'''
'''[[Releases/Firefox 3.0.16|Firefox 3.0.16]] / [[Releases/Firefox_3.5.6|Firefox 3.5.6]]'''
* [https://bugzilla.mozilla.org/buglist.cgi?quicksearch=ALL%20blocking1.9.1:.5%2B%20-status1.9.1:.5-fixed&order=map_assigned_to.login_name,bugs.bug_id 3.5.5 open blockers]
* Code freeze is November 10 at 11:59pm!
* [https://bugzilla.mozilla.org/buglist.cgi?quickseach=ALL%20-!fixed1.9.0.16,verified1.9.0.16%20flag:approval1.9.0.16%2B&order=map_assigned_to.login_name,bugs.bug_id 3.0.16 open blockers]
* '''3.5.5:''' [https://bugzilla.mozilla.org/buglist.cgi?quicksearch=ALL%20blocking1.9.1:.6%2B%20-status1.9.1:.6-fixed&order=map_assigned_to.login_name,bugs.bug_id 35 open blockers] ... [https://bugzilla.mozilla.org/buglist.cgi?quicksearch=ALL%20blocking1.9.1:.6%2B%20-status1.9.1:.6-fixed%20-flag:approval1.9.1.6%2B%20-flag:approval1.9.1.6%3F&order=map_assigned_to.login_name,bugs.bug_id 27 need 1.9.1 patches]
* '''3.0.16:''' [https://bugzilla.mozilla.org/buglist.cgi?quicksearch=ALL%20flag:blocking1.9.0.16%2B%20-!fixed1.9.0.16,verified1.9.0.16&order=map_assigned_to.login_name,bugs.bug_id 15 open blockers] ... [https://bugzilla.mozilla.org/buglist.cgi?quicksearch=ALL%20flag%3Ablocking1.9.0.16%2B%20-flag%3Aapproval1.9.0.16%2B%20-flag%3Aapproval1.9.0.16%3F 11 need 1.9.0 patches]


'''Firefox 3.6 Beta'''
'''Firefox 3.6 Beta'''
Line 13: Line 14:
* will refresh frequently via the Firefox 3.6 beta update channel, each update will be versioned "3.6b#" and presented as a "revision"
* will refresh frequently via the Firefox 3.6 beta update channel, each update will be versioned "3.6b#" and presented as a "revision"
* next update is aimed to be this Friday, minimally containing:
* next update is aimed to be this Friday, minimally containing:
** DLL component directory lockdown and blacklisting patch (see {{bug|524904}} and {{bug|519357}}
** DLL component directory lockdown and blacklisting patch (see {{bug|524904}} and {{bug|519357}} (blocklist policy discussion in {{bug|525103}})
** syntax changes for CSS gradients (see {{bug|513395}})
** syntax changes for CSS gradients (see {{bug|513395}})
** cycle-collector patch to stop cycle collection instead of crashing (see {{bug|521750}})
** cycle-collector patch to stop cycle collection instead of crashing (see {{bug|521750}})
Line 20: Line 21:
'''Firefox 3.6 Release Candidate'''
'''Firefox 3.6 Release Candidate'''
* proposed code freeze November 18th (2 weeks)
* proposed code freeze November 18th (2 weeks)
* proposed release November 19th (Happy Thanksgiving)
* proposed release November 26th (Happy Thanksgiving)
* need to check these schedules with build and QA, obviously
* need to check these schedules with build and QA, obviously


Line 45: Line 46:


= GFX Update =
= GFX Update =
* [https://bugzilla.mozilla.org/buglist.cgi?quicksearch=ALL%20comp:gfx,image,widget,graphics%20flag:blocking1.9.2%2B,blocking-firefox3.6%2B%20-status1.9.2:fixed,unaffected,wontfix 11 blockers]
* [https://bugzilla.mozilla.org/buglist.cgi?quicksearch=ALL%20comp:gfx,image,widget,graphics%20flag:blocking1.9.2%2B,blocking-firefox3.6%2B%20-status1.9.2:fixed,unaffected,wontfix 18 blockers]
** 9 are FIXED, several have patches.
** GFX team will be triaging these to see which absolutely need to block.
** Looks good for a November 18 freeze date.
* 3.5.4 crashes
** We discovered once 3.5.4 had been released that we had caused two topcrashes: {{bug|524462}} and {{bug|525326}}. These were fallout from security bug fixes.
** More testcases helps, and we're adding more. But the nature of graphics code (handling user-provided data) implies that it's very difficult to have fully complete test suites.
** It's very likely that these could have been caught in a beta period for 3.5.4. Since we already look at crash-stats, should we just extend beta periods? Should we make it part of the "new release" signoff that we haven't introduced new topcrashes?
* GFX team focusing on blockers, Electrolysis, crashes, and performance.


= Layout Update =
= Layout Update =


* CSS gradient syntax change ({{bug|513395}}) landed on m-c yesterday; backport to 1.9.2 being tested now; propose to land it Wednesday morning
* Blocker report:
** 18 blockers, but mostly (9) fixed on trunk and need 1.9.2 landing, or have patches, or we're not sure there's a bug
** Silverlight issue fixed after conversation with Microsoft --- we're doing a trivial workaround for them calling NPN_Invalidate for a windowed plugin
** Only 1 bug where we know there's a bug and we don't have a patch (519767)
** 1 untriaged layout nom
* CSS gradient syntax change ({{bug|513395}}) landed on m-c yesterday; backport to 1.9.2 is posted to the bug, the a1.9.2 queue, and the try server; would like to land it tonight or Wednesday morning assuming all is well
* Layers API developed: https://wiki.mozilla.org/Gecko:Layers. Implementation to proceed shortly. Send feedback.
** Framework for hardware acceleration, 3D transforms, retained buffers for elements, cross-process rendering, off-main-thread compositing, animation and video playback
* Patches ready to remove nsIScrollableView on trunk
* Jonathan Kew in Toronto office this week
* WOFF getting lots of press
* Rapturous reception at AtypI


= Content Update =
= Content Update =
* Excellent work by bz and smaug on a 1.9.1 stability regression
* Working on 1.9.2 blockers (26 total, 5 fixed/dup, 7 crashkill)
* crashkill work
* HTML5 parser work moving forward, getting through reviews.


= Platform-specific Support Update =
= Platform-specific Support Update =
* (smichaud) 1.9.2/JEP update -- No new bugs in last week.  No blockers.


= JS =
= JS =


= Startup Performance =
= Startup Performance =
* Almost done with a tool that shows numbers for all Talos tests across all branches for all platforms, summarizing median and mean per platform, including weekly difference, and difference from Firefox 3.5.
* Found a scenario for stable cold startup numbers on Windows. Next step is to get a Talos patch up, and work with Rel/Eng to get it deployed into testing so we can see numbers on real Talos boxes.
* Rob Strong split up the update service in bug 311965, has most reviews (affects all toolkit apps) and is about ready to land. This showed a significant win on WinCE startup.
* Taras and Joel are working on {{bug|524202}}, tracking down exactly how and when dynamic library code is loaded. See [https://bugzilla.mozilla.org/show_bug.cgi?id=524202#c3 this comment] for a good summary of the issue.
<!--
{{:Firefox/Projects/Startup_Time_Improvements}}
{{:Firefox/Projects/Startup_Time_Improvements}}
-->


Join us on IRC in [http://irc.mozilla.org/#startup #startup].
Join us on IRC in [http://irc.mozilla.org/#startup #startup].
Line 66: Line 99:


= Electrolysis =
= Electrolysis =
* [http://benjamin.smedbergs.us/blog/2009-10-30/multi-process-fennec/ Fennelectrolysis lives]: with only a medium amount of hackery of the Fennec frontend itself. Currently getting the platform-side patches landed.
* Working quickly towards getting IPC plugins landed on mozilla-central, preffed off.
** Multiple plugins work, ipcplugin tests work (2 orange because NPN_SetException isn't implemented)
** cjones working on leaks during ipcplugins tests (probably shutdown-only leaks)
** bent finishing work on Windows hangs


= Tree Management =
= Tree Management =
* Help with making debug tests green
* Help with making debug tests green. {{bug|523385}}
* Upcoming power outage at 650 castro
* Upcoming power outage at 650 castro
* More machines added to Try Talos
** November 14th
* New Talos suites
** {{bug|524047}}
** Mobile tests will be disabled for the duration
 
* 11 more machines on Try Talos {{bug|524849}}


Upcoming this week:
Upcoming this week:
* Electrolysis tests {{bug|515436}}, {{bug|523142}}
* Electrolysis tests {{bug|515436}}
* Stopping refcounting build on trunk on linux
* Stopping refcounting build on trunk on linux


= Roundtable =
= Roundtable =
* roc: Trying to make httpd.js do GC; xpcshell tests uncooperative
* robc: firebug - event listener service? yay!
** {{bug|448602}}
** {{bug|506961}}
** {{bug|521010}}
** {{bug|507448}}
** shaver asserted that this is valuable and the risk profile seems right, but if it bounces it will bounce hard and need to be removed
Confirmed users, Bureaucrats and Sysops emeriti
2,976

edits