2018-09-06
NECKO WEEKLY MEETING NOTES
Thurs Aug 30, 2018
TRIAGE: see https://public.etherpad-mozilla.org/p/necko-triage-guide for how-to
Nick
Honza
Michal
Daniel
Valentin
Junior
Dragana
Kershaw <<<<<===== August 31 to September 6
Round Table:
- sec-high bugs
Nick
- h2ws still progressing, have on* firing (finally), still no data flowing
- a few minor h2 issues
- avoiding httponly cookie values in child processes (wpt failures)
Honza
ni? 1463729 (fast net, slow logging IO -> hang socket thread in an endless loop, h2)
smb proxy bypass - ni? valentin
Michal
- 1428901 Persist TLS session tickets across browsing sessions
Daniel
- TRR - disable in devtools
- ECS mistake
- URI "template"
Valentin
bug 1476996 - cross process redirect impl [landed]
bug 908413 - mime type parsing in data URIs
Dragana
hq - http for quic
Junior
Bug 1485353 - Create PHttpTransaction IPC protocol
- HttpTransactionParent/Child is split. Fixing the testing failure.
- TODO: (a) Implement HttpTransactionParent::Cancel/Suspend/Resume
(b) env variable off to opt-out
Kershaw
Bug 1357676 - Increase cookie count limits to match Chrome's
- landed
Socket Process
- Address Honza's comment
- Use full XPCOM in socket process
Gary
Bug 1373708 - Convert Jar channels to stop using main thread I/O
- landed to mozilla-inbound
Amy
Batch Eviction
-
Had confirmed the algorithm with Josh and Martin.
-
Martin had sent the mail to discuss the algorithm with Mike West. Bug 1425031 - Firefox Quantum blocks cookies when JavaScript updates them
-
Found the root cause.
-
Started to create a mochitest for this bug. Bug 1426471 - Cookies set via JS during anonymous screenshoting override real cookies
-
Can't reproduce.
SC
[Socket Process]
-
no progress this week on my part [Others]
-
Bug 1426751 - Intermittent eventsource-close.htm | application crashed [@ mozilla::net::ChannelEventQueue::FlushQueue] (Assertion failure: !mFlushing, at netwerk/ipc/ChannelEventQueue.cpp:46)
-
multiple threads trying to flush ChannelEventQueue at the same time
-
need to check and set flag at single critical section
-
r+’ed and landed on m-c
-
security bug
-
discussing my proposal with @mayhemer on bugzilla
-
trying @mayhemer's suggestion and upload my WIP, debugging test failure while running mochitest
Will Wang [Back Pressure]
Bug 1280629 - e10s breaks necko back pressure
*- discussion, tracing code *- going to discuss the experiment for min/max queue size
Cervantes
[Socket Process]
- patch rebase and breakage fixes
- working on pref system