Electrolysis/Meetings/2016-05-19

« previous week | index | next week »

chutten

beta47-week3 analysis is available

• bug 1260769 - SLOW_SCRIPT_PAGE_COUNT difference is up to 0.11 from week2's 0.09
• bug 1272403 - hangs over 100ms difference is down to 0.30 from week2's 0.37

mconley

• M9 bug 1266575 - [E10s] click event is not dispatched to <select> when clicking on opened popup
  • Fixed and landed. Will request uplift today (all the way to 47?)
• bug 1166351 - XBL parent bindings not properly attached - tabs suddenly do not switch or close properly
  • Finally figured this out. Landed on inbound last night. Check out the last few comments in that bug if you want the gory details.
  • From what I can tell, this is probably a bug that got introduced in 2014. Dunno why it started rearing its head more aggressively this year.
• bug 1261842 - Make initial browser remote sooner if we're defaulting to using remote tabs
  • My refactor is done (locally), and now I'm doing test clean-up, because making the initial browser remote seems to change some assumptions in various components.
• My Outreachy intern Rutuja Serve starts on Monday!
  • She's Rutuja in IRC.
  • She's working on bug 1255843 - Develop a tool to measure resource usage of each subprocess
    • Got support from njn and erahm on this. Starting our work based on some stuff :past did for lightweight measuring of per tab memory.
• The usual slew of reviews and feedbacks and needinfos
  • Did a bunch of review clean-out this week. If I haven't gotten to you yet, I will today.

gabor

• bug 1262661 - (OOM - Many large SessionStore:update messages) - Looked into some tests, waiting for review.
• bug 1258317 - (crash - crash in mozilla::ipc::TransferHandleToProcess) - Tried to do some research and had some talks, in the end Bob Owen came up with a workaround for the problem.
• bug 1259480 - (crash - FatalError crash in PContentChild::Read from PContentChild::SendReadPrefsArray) - looked into the pref parser to find something fishy that can cause the message corruption. Did not find anything that could cause the problem.

cervantes

• bug 1268900 NEW:: Add opt-in way to immediately deserialize a message after serializing it
  • has a working prototype
  • working on the code generation part
• bug 1258317 REOPENED:: crash in mozilla::ipc::TransferHandleToProcess (PBackground related)
  • updated patch pt 2. Requesting review.
  • Working on pt3 patch
• bug 1264566 NEW:: [e10s] UAC error message "Acces Denied" when doing action on file/folder after upload in Firefox with form submit
  • Helped investigating

kanru

• bug 1264642 ASSIGNED:: Reduce the contiguous address space needed for StructuredClone serialization
  • Start to rebase patches on top of BufferList<T>

ting

• bug 1259512 NEW:: [e10s] significantly higher rates of OOM crashes in the content process of Firefox with e10s than in the main process of non-e10s
  • read minidumps to check why there're so many tiny (<1M) blocks in virutal space
  • tried to break at VirtualAlloc* to find the piece of code allocating PAGE_WRITECOMBINE, but no luck
• bug 1267329 ASSIGNED:: CreateMinidumpsAndPair doesn't record memory information
  • separated the patch to two parts, one for google breakpad and one for nsExceptionHandler
• bug 1110596 NEW:: All data passing through HttpChannelParent::OnDataAvailable() is copied
  • tried to make a patch based on :mccr8's proposal, but found difficulties

wcpan

• bug 1264566 NEW:: [e10s] UAC error message "Acces Denied" when doing action on file/folder after upload in Firefox with form submit
  • Investigating
  • Patch ready for feedback

felipe

• bug 1239552 - Marionette-e10s tests
• bug 1227230 - Remove all of the e10s prompting code
• bug 1272751 - Remove block of GTK+ 3.20+ from e10s on beta
• Beta 47/GoFaster talks
• bug 1273630 - about:tabcrashed should hide extra fields by default and let it be opened with a toggle
• bug 1274088 - Remove release=0% in the e10s rollout add-on for 47
• Reviews and feedbacks

bobowen

• bug 1035125 - On Windows, plugin-container.exe is linked against the sandbox_s library twice - landed
• bug 1271601 - Handling IPC transport failures more gracefully - reviewed, waiting on try

mrbkap

• bug 1271244 - about: shims don't work anymore - debugged, patch up.
• bug 1267844 - m9 - infinite throbber when dragging tabs between windows - fixed.
• bug 1244533 - dragend event doesn't pass screen coordinates - patch attached.
• bug 1259517 - dropping an image into a contenteditable doesn't do anything - rebased my patch and should have it working today.
• A few reviews and needinfos.

RyanVM

• All OSX 10.10 e10s test suites enabled on 48+ now
• PSA: Win7 tests on AWS

haik

• bug 1165945 - Better handle a failure to start a content child process - Posted patch with telemetry but needs more work on basic handling of the failure

aklotz

• First testing of a11y+e10s local build
• Possible thanks to patches for bug 1272146 and bug 1273635.
• Working on adding logging for incoming a11y calls
• Plan to start distributing try builds to initial testers (a11y team, client vendors, other interested parties) next week

elan

• We are clear for the entire 47 Beta cycle \o/ Thank you for everyone's hard work, seriously.
• Next up: release plan for 48, roadmapping for Add-ons, A11y, and sandboxing

ksteuber

• Further revisions of bug 1263774. It should be just about ready; just waiting for review.
• Work continuing on fixing tests broken by the proposed patch for bug 1206133.
• Will be on vacation for the next 2 weeks (May 23 - June 3).

mccr8

• bug 1271102 - Looking at large (>128MB) IPC messages.
• bug 1275707 - Filter junky message manager size telemetry.
• bug 1272423 - Message manager should reject large messages.
• bug 1274706 - Reduce message size for PExternalHelperAppChild::SendOnDataAvailable. Major source of 128MB+ messages.
• filed: bug 1273685 - PopupBlocking:UpdateBlockedPopups messages can be very large.
• filed: bug 1275398 - Use shared memory to send images in IPCDataTransfer.