From MozillaWiki
Jump to: navigation, search

« previous week | index | next week »

Firefox 3/Gecko 1.9 Meeting Details

  • Tuesdays - Firefox 3 - 11:00am Pacific, 2:00pm Eastern, 19:00 UTC
  • Mozilla Building S - Central Area
  • 650-903-0800 or 650-215-1282 x91 Conf# 217 (US/INTL)
  • 1-800-707-2533 (pin 369) Conf# 217 (US)
  • #granparadiso for backchannel

Open Action Items

  • AI:chofmann Drive outreach to extension authors r.e. leak squashing
  • AI:chofmann File bugs for comment field analysis on bp server (done)
  • AI:peterv Help drive leak fixing
  • AI:Tomcat wikify/blog your steps to find extension leaks
  • AI:pav Drive list of footprint bugs for next perf meeting
  • AI:sicking Move content arenas patch to completion
  • AI:schrep need to find some help for threadmanager Ts issues
  • AI:dbaron Feedback for breakpad r.e. need for local vars (or equivalent deubgging aid)
  • AI:johnath Determine if we need a new NSS rev
  • AI:mconnor Figure out where list of 'done' new features can get help created

Status of Key Areas to wrap up for final ship

  • Visual
    • Icon Refresh
      • main window and preferences will be ready for beta 3
      • addons manager, download manager, page info & others likely not ready for beta 3
    • General Theme Polish
      • w32: main toolbar CSS changes will be ready for beta 3
      • all: awesomebar changes and star-dialog changes for beta 3
    • Mac Theme
      • proto will land in beta 3
  • Memory Usage
    • JEMalloc Work
      • Builds near working - #'s by end of weak
    • Leak hunting
      • Concerns over extension leaks. AI's above
    • General Footprint improvements
      • 10 or so bugs floating around yet - Pav to get list of bugs for tomorrow's perf meet
  • Performance
    • Ts/TXul
      • threadmanager - need help here
    • Tp
      • Need to investigate CJKT issues on windows
    • Graphics
      • Borders and backgrounds need some work. Vlad on it
    • Javascript
    • DOM bug 407216 - fast path top N DOM methods
    • FileURL Bug
    • XOW still needs some work
  • Awesome Bar
    • Global Frecency (bug 394038) - patch in post-review cleanup stage, should land today/tonight.
  • Places
    • Full tag backup and restore (bug 384370) - perf still not up to par with C++ bookmarks.html export, currently investigating bottlenecks.
  • Offline
    • Honza making fast work - much of backend done this week
    • DCamp has UI under control for this week
  • Malware
    • Waiting on new server impl
  • AMO Integration
  • XOW
  • xsite XMLHttpRequest
    • More spec changes in flux. Sicking to snap it into shape
  • General Website Compatibility
    • No update here. Need hendrix data and others
  • SW Update
    • Anti-fatigue stuff should be in this week. Then we are done
  • SVG
    • Filters - discussion of content-pref to turn them on/off
  • Printing
    • On GFX side we are better on everything but Linux - Vlad to fix
    • Dholbert to the rescue for layout issues
  • MathML
    • Visually-significant outstanding issues are:
      • Preferred width calculation bug 363240
        • A possible patch using reflow with unconstrained widths should be ready for review within a day or two.
      • Wrong glyphs selected when font subfamilies don't have the same glyph sets on MS Windows bug 382542 and Linux bug 401988
        • Shouldn't be major to fix these but may affect Ts on Windows. Unlikely to be done for beta 3.
    • With these issues resolved, rendering is respectable with STIX beta fonts.
    • Could make some small enhancements so that rendering with only default platform fonts is significantly improved.
      • Current state without downloading extra fonts is that some things are better than FF2, but some things are worse (particularly on Windows XP without Office 2007 or Powerpoint Viewer 2007).
  • Vista Breakpad
  • Crash Reporting server (Socorro) to the set of issues that we need some solution for.

    • Tracking for bugs in mozilla's instance of socorro has moved over to bugzilla and bug 411430, bug 411431, and bug 411432 are on file for the beta blocking issue related to comment reporting, access to raw stack dumps, and poor performance on access to analysis reports.
  • EV-SSL
    • working to get Draft 11 audited VeriSign EV root in for Beta 3
    • need to determine which root certs have been EVSSL 1.0 audited for final
  • NSS updates?
    • Jonath to ping
  • Web-based protocol handling
  • Microformats
    • Need review for bug 402871,bug 402877
    • Requested from sayre, no response
    • Note to mconnor for recommendation of another reviewer, no response
  • Help (SUMO etc)
  • Save Link As... filename fixes bug 299372
  • Plug-ins - concern over flash plug-in causing repaint

Firefox 3


Security Reviews

  • Design Review Proposal
  • Still needed
    • Microformats (mkaply) 1/22 after this meeting
    • Malware Protection (dcamp) 1/22 after this meeting
    • Distribution/Customization (thunder) 1/29 after this meeting
    • XMLHttpRequest follow up 1/29
  • Completed
    • Password Manager (done)
    • Offline Apps (done)
    • Web Content Handlers (done)
    • Add-Ons (rob_strong) (done)
    • Places (dietrich/sspitzer) (done)
    • Cross-domain XMLHttpRequest - Done.
    • ContentEditable (peterv) - Done.

QA Status/Topics

  • Top critical issues tracking for beta 3:
    • bug 411249 – Top crash [@ jsds_ScriptHookProc[@jsds_ScriptHookProc(JSDContext*, JSDScript*, int, void*)] on shutdown with Firebug 1.1.0b10 installed
    • bug 411369 - Crash while trying to print [@ @0x0 @0x1000 _cairo_meta_surface_create] [@ _moz_cairo_surface_finish ] - This is a major regression that is blocking any printing on Vista and XP.
    • bug 412375 - Older crashreports no longer visible ?. Unable to retrieve reports prior to Jan 10th. Webdev investigating. Update - data has been lost due to running out of space on db. unrecoverable
    • bug 390568 - vista fails to submit breakpad reports. This has been broken since alpha8, and we are losing more crash data as long as this remains unfixed.
    • bug 378528 – crash reporter should attempt to resubmit pending reports. Needs to be rolled up into a queue for later if breakpad is down. This causes backlog of missing crash data that takes days later to resurface. Takes away time from researching critical issues.
    • bug 405198 Cannot drag and drop folders / items in Places Organizer - this cause the problem that its not possible to move bookmarks from one folder to another folder oder to merge folder etc .... - Tomcat Why is this assigned to nobody?
    • bug 324857 – MathML all screwed up in Cairo builds. Blocking MathML usage. Lots of action in the bug, but no resolution yet?
  • B3 testplan is in the works. Regression Testing coverage will include:
    • focused testing
    • Smoketests, BFTs
    • bug verifications
    • l10n builds (if string frozen)
    • Top websites
    • Supported plugins
    • Major update run (incl partners if permitting)
    • Leaks
  • MU update testing of Fx2.0.0.11 -> Fx3b2. Testplan and results here.
  • Leaks testing - bug 402335: up to 49 tracking bugs.

Localization Topics

Ahead of schedule for roughly 50% of our locales via B1 and B2.

The other 50% are more or less on fx2 tree status, including new localizations.

L10n dashboard in discussion phase, see mockup.

Need to get at a pretty accurate string freeze date, and broadcast it widely. I think what we agreed to was that any bug that has possible string landings after the end of Jan. ought to get the late-l10n keyword applied. mconnor belived that string changes should be pretty much wrapped up by beta 3, but if you spot any bugs that have string landing impact past the end of Jan. we should start to flag those with the keyword.

General Discussion

Discussion about upcoming change of tinderbox-client to buildbot-client. This had several interwoven-but-separate issues.

  • buildbot client vs tinderbox client; helps get the nightly builds to use the same machines, and more similar processes to production releases. Should happen in next week/two. Will start working on 1.8 branch first, and work on trunk next.
  • build-on-checkin vs build-always; See John's blog for details. The comments are also interesting, and well worth reading. John to file bug, and add blog&comments to bug.
    • Initially, we will continue to run as build-always, but easy to change from one mode to another.
    • There are some (unclear) changes needed in talos/perf/testing infrastructure to handle this change from build-always to build-on-checkin. Possibly only graphserve, but being investigated by robcee and alice.
    • currently talos never has idle time, because build is always generating new builds... even if there is no code checkin. However, if we change to build-on-checkin, there could be fewer builds, and maybe even talos catches up, and becomes idle. Discussion on how to have talos rerun tests on existing pretested builds in this case.
    • after idle time, 1st checkin would trigger all platforms with *same* start time. Currently each platform build starts whenever they finish the previous build. A big+
    • after idle time, build-on-checkin would make 1st checkin turnaround time better. Debate whether 2nd checkin better/worse; simple analysis of changes over the last 30, 60 days, indicates better, but possible to construct cases where worse.
  • start with one slave per o.s., grow to pool of slaves soon after. This would resolve the 2nd checkin turnaround time discussion.
  • tree stability timer feature, new in BuildBot, not present in Tinderbox. big+

Round table