« previous week | index | next week »

OKR confidence

1. O: Make Firefox more stable [New stability dashboard] [current crash rates] [Crashes per 1000 usage hours]
2. O: Improve Google document performance on Firefox
   1. O: Understand Google document performance issues on Firefox
      1. KR: Implement proposed fixes and measure results
         • Current:

Previous Week

ting

• Previous
  • [x] 64K/1M experiment
  • [x] clang-cl ASAN & static analysis
• Still don't observe significant between 1m/64k for ExecutableAllocator
• Triaged 1030 nightly crashes
• bug 1307933 NEW:: Windows ASan build failure: intel-aes-x86-masm.obj : error LNK2026: module unsafe for SAFESEH image
  • bug 1307933 am trying to figure things out
  • bug 1307933 submitted a patch
• bug 1298905 UNCONFIRMED:: Unused memory does not get released to OS
  • bug 1298905 checked the stacks from wpa the reporter attached, found a culprit
  • bug 1298905 found a workaround for download xul.pdb, commented on the bug
  • bug 1298905 tried wpr/wpa for recording nightly, but failed translating the symbols even firefox symbol server url is set, double checked but still the same
  • Responded to bug 1298905 and the ni of bug 1315285
• bug 1251936 NEW:: Stand up Windows x86 debug static analysis builds
  • bug 1251936 trying to build updated clang for tooltool
• bug 1315285 NEW:: Frequent Windows IPC-related test hangs with jemalloc4 enabled
  • Responded to bug 1298905 and the ni of bug 1315285
• bug 1307561 NEW:: Windows ASan LIB path isn't getting picked up
  • bug 1307561 searched for suitable env variable for vc dir during configuration at both taskcluster and local, no luck
  • Submitted a patch to bug 1307561
• bug 965936 NEW:: [meta] Virtual address space fragmentation
  • bug 965936 still don't observe significant differences between 1m/64k for ExecutableAllocator
  • bug 965936 unfortnuately 1m executable pool size does not show siginficant benefit from 64k, commented on the bug

wcpan

• Previous
  • [ ] Coverity
  • [x] MOZ_MUST_USE in netwerk/protocol
• Adding moz_must_use to netwerk/protocol/res
• Sent group mail to dev-platform
• bug 1310127 NEW:: Use MOZ_MUST_USE in netwerk/protocol/http
  • bug 1310127 - got some unreproduceable xpcshell error, trying to figure out
  • bug 1310127 - still bunch of oranges ... can't reproduce on local, can't see stack on try

gregweng

• Previous
  • [x] Trace log websocket backend
  • [x] Discuss with devtools peers
• Created a very ugly copycat of LayerScopeWebSocketManager to handle TraceLogger and WebSocket
• Define a message protocol for TraceLogger data format
• Encounter dependencies hell when trying to pull the socket manager out...
• Now the socket manager can be enabled or disabled from a pref. The port is also in that
• Took a look at the binary TraceLogger tree format; need to port the data view in C++
• With help from wcpan , solved the issue and now I can have a standalone TraceLoggerWebSocketManager compilation unit

kanru

• Previous
  • [x] Plan hasal work week
  • [x] Ask digitharald about crash rates and Flow
  • [ ] Study breakpad
  • [ ] Automatic retrieval of uptime triage data
  • [ ] Study how to generate test coverage on try
• Done 10-31 nightly triage
• So the kusage_hours query url is listed on https://chutten.github.io/telemetry_crashes/, look at the bottom left
• Unable to run socorro locally
• bug 1314254 NEW:: Change IPDL message handlers return value to mozilla::Result or nsresult or just MOZ_CRASH
  • Filed bug 1314254
  • Study how to use LLVM to do source-to-source transform

cervantes

• Checking coverity failures
• Rerunning xpcshell-test with app verifier

This Week Priorities

gweng

• PTO

kanru

• Fix IPC handler error handling
• Study how to collect test coverage data
• Study new telemetry data

cervantes

• Coverity

thinker

• To land back-face visibility patch
• Task tracer
• Hasal discuss with ES team about test case

wcpan

• PTO

ting

• clang-cl - to enable ASan and S build

Incoming

• Nov: WebSummit/CodeSummit