Platform/2011-09-27

« previous week | index | next week »

Goals

• Platform/2011-Q3-Goals - Update status
• Platform/2011-Q4-Goals

Firefox Development

• Dao has fixed one of the more annoying copy/paste issues with the URL bar protocol-stripping behaviour (URLs selected from the awesomebar should not have the prefix added correctly) (bug 668019)
• (not really firefox front-end, but…) Gian-Carlo Pascutto is making very good progress on switching the safebrowsing store from SQLite to a flat file, to make it more suitable for mobile devices (bug 673470)

Mobile

Automation

• bug 688762 - fix return codes for remote harnesses, this should reduce hanging tegras
• bug 687974 - adb support for talos
• bug 688604 - make talos-remote, coming soon!

Mobile Perf Program

• Can Joel or Clint talk to: bug 685632 - Please collect RSS (memory) data during Talos testing on Android
• Also digging into helping Joel prioritize & drive which automated tests need to be fixed and helping with getting the hw lab scalable.
• The focus this week is to create a perf metrics matrix so we can obtain baseline measurements. This is a grassroots effort so that we can create a snapshot of the areas we need to focus on optimizing on mozilla-central in time for the mobile work week. We are focusing on "Macro" or observed measurements and will likely use a 7D/HDMI camera to capture screens
• Still need to land on what the spectrum of reference HW; targeting by the end of the week
• A draft of the matrix which will be ready for circulation tomorrow (wed).

Android Native UI

• DougT's blog post summarizes.

Tree Management

• 80 New 10.6 test machines ("rev4 mac minis") incoming in the next few weeks. This is a larger pool than before and faster hardware, so they'll turn around results faster. {{bug
• Old 10.6 test machines will be recycled as linux and windows test machines, which should help throughput for those platforms too.
• PGO being turned off for on-change builds. bug 658313
  • Nightlies will still be PGO
  • PGO builds will be run every 4 hours

Stability Report

• Notes from yesterday's crashkill meeting - https://wiki.mozilla.org/CrashKill/2011-09-26.

Socorro

• Release deployed today
• Has a skip list for mobile that will help get us more meaningful signatures.

Desktop

• FF7 out today so too early for meaningful data.
• Problem with ADUs on Trunk and Aurora showing up as 0 - seems to be fixed. bug 688587
• All channels look relatively stable.
• Most of the issues we are tracking relate to external problems
  • bug 684748 - Comcast issue. White Sky deployed a fix last week and crashes are decreasing as people get the update.
  • bug 665775 - DataMgr.dll - Marcia working on checking the list of versions that need to be blocked.
  • bug 688637 - Mcafee script scan - monitoring volume.
  • bug 688895 - Babylon toolbar - monitoring volume.
  • bug 688895 - crash nsRuleNode::Sweep
  • bug 687991 - mail.alice.it. Not getting reports due to problem with the crash reporter - this was fixed. Verified it's fixed on the mac side on 09/23/11. Need to verify on the windows side.
  • Haven't seen any issues yet related to the Flash release last week but marcia keeping an eye out for plugin crashes.

Mobile

• Naoki's detailed report - https://wiki.mozilla.org/QA/Fennec/crash_rep_26Sept2011
• Crash reporter issues
  • bug 688692 , bug 686492 and bug 686800
  • bug 686973 - crash happening in Java layer so we don't get meaningful signatures - around 15% of crashes.
  • bug 689234 - Mobile is not reporting the graphics data in the app notes field. Probably not working because on the desktop we are only reporting this for Windows and not other platforms.
• Top problems we are tracking.
  • bug 687367 - We worked around this but are now seeing it in the wild. Need to make sure that Andreas knows that this is pretty urgent for Michael to work on.
  • bug 663494 - JP will ask Brad Lassey and his team to look at this bug.
  • bug 677548 - crash [@ nsBuiltinDecoderStateMachine::AdvanceFrame ][@ libc.so@0x15dd8]
  • bug 661158 - Crash on mobile SMP devices while sending PRNetAddr over IPDL [@ Pickle::ReadBytes]