Notices / Schedule
- Firefox 5 shipped today (mobile and desktop, rock!)
- Next source migration is July 5th
- We'll have a post-mortem for the rapid release process next week, time TBD
- We'll have a meeting about 3.6 EOL next week, time TBD
- See above, Firefox 5 shipped today!
- No dev update this week, it's been a bit of a distracting day.
- Spinning up a firefox team project branch, follow along at home! bug 663128
- There are blog posts forthcoming, but we have recently changed the font rasterization method on Windows Vista/7 when using DirectWrite to make fonts "less fuzzy."
- This change is now on mozilla-central, and we're hoping to land it in Aurora before too long.
- Technical details: We're now using the GDI_CLASSIC DirectWrite rasterization method for the most common web fonts at smaller sizes. We believe this should resolve the most common problems without pessimizing downloadable fonts unjustly.
- For more information, see bug 661471.
- Brian Hackett plans to land the type inference branch to TM shortly before the next mc->aurora merge.
- JS changes that want to go out for Fx7 would land directly to mc during the time between the TI->TM merge and the TM->mc merge.
- TM would merge to mc just after the mc->aurora merge for max bake time on nightly.
- He is currently seeing a ~1.6x improvement on Kraken and 1.1-1.3x on V8-V6 (vs trunk).
- Jonathan Kew's font inspector API landed. You can now write extensions that will tell you exactly what font(s) are being used to render any given DOM range --- including the entire document, or a single character. For downloaded fonts you can get metadata including the license.
- [josh] Bug 90268: Content ownership of plugin instances is mostly working on all platforms. Patch is on the holly branch. Need to fix remaining test failures and some to-do items before reviews can start. Can probably have that done and reviews started by the end of the week.
- AddEventListenerByIID() n' friends is going away (sicking)
- [hurley] NeckoNet now available. Standalone test suite for networking. Can record and replay test pages with different builds against custom-configured networking conditions. Will be adding features and improving usability quickly.
- Status http://benjamin.smedbergs.us/weekly-updates.fcgi/project/electrolysis
- Drew's work to instrument chrome/content crossovers continues (status page)
- Most of Felipe's patches to stand up the browser on e10s are reviewed or landed. Once they're all done, you will be able to build e10s mode desktop firefox with a single mozconfig switch (and not have it immediately fall over)
- Tree closure Wednesday June 22nd, 8-11am PDT announcement in dev.planning
- We have been running side by side tp4 and tp5
- We would like to disable it next week on June 30th
- comments/questions? bug 664831 - disable tp4 (except older release branches)
- The overall crash numbers look good but the trunk has been a bit more variable than for FF5 and even FF6.
- Weird spike yesterday due to a new signature - @0x12e117. All the reports appear to be dups so for now we are going to ignore this. It drove up the crash rate for a day (see graph).
- 2 signatures that are not exactly new but have been at the top of the list for the past week or so...
- New signature filed for 4.0.1 but present on other branches:
- MemShrink effort up n' running, next meeting today at 1pm PDT. (jst)
- Land aurora patches with approval and mark as appropriate
- mozilla-inbound is not the new try server, it is not OK to land untested patches there! :) (ehsan)