meeting-2018-11-08 - Weekly Necko Meeting Notes

Thurs Nov 8 2018

Round Table:

Sync and debugging: https://bugzilla.mozilla.org/show_bug.cgi?id=1409570#c28
- related idea: we’re dumping lots of information wrt H2, stop dumping the table for every compression operation (nick to file a bug)
- sync -- by default in about networking, so that we get the data with crashes
- Future work: think about our logging situation and propose a way forward... someday -- or ask for 6 aka "turn it up to 11"
Unassigned p1: https://bugzilla.mozilla.org/show_bug.cgi?id=1418134
- moved to overholt
Unassigned p2s - selena is working through it (see below)
[selena] First party isolation for cache
- dragana / ehsan discussing
- step 1: telemetry on penalty for doing it; nick to pick it up in a few weeks
  - bug? next step: open a bug -- dragana please :)
triage issues:
- https://bugzilla.mozilla.org/show_bug.cgi?id=1505366 - assigned to daniel
- https://bugzilla.mozilla.org/show_bug.cgi?id=1358043 - reproducing crashes on Android?
- https://bugzilla.mozilla.org/show_bug.cgi?id=1451293
  - honza - open a bug for the audit
- https://bugzilla.mozilla.org/show_bug.cgi?id=1321466#c27 how much work is this?
- https://bugzilla.mozilla.org/show_bug.cgi?id=1340200
  - Chrome implementation https://bugs.chromium.org/p/chromium/issues/detail?id=435547
- https://bugzilla.mozilla.org/show_bug.cgi?id=1342484 - service workers
- https://bugzilla.mozilla.org/show_bug.cgi?id=1356686 - brotli decompression, can we devise a test / maybe with facebook?
  - honza close these out if they are fixed :)
[honza] many P2 -> P3 for inactivity (by Sylvester)
- surprised! 1 year of inactivity
- selena: get a list from sylvestre of what changes; schedule a q3 triage pass
[honza] stricter tracker annotation list on Nightly -> affects tailing
- https://bugzilla.mozilla.org/show_bug.cgi?id=1501461
  - Fonts!
  - need tests to show that tailing is helping / not helping -- ran a study on beta; pref flip - zero difference
- The flip pref bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1434379
- Some graphs not linked in the bug:
- https://moz-experiments-viewer.herokuapp.com/?ds=pref-flip-tracker-request-beta-1434379&metrics=ALL&next=%2F&pop=ALL&scale=linear&showOutliers=false
https://bugzilla.mozilla.org/show_bug.cgi?id=1503393 - proxy conf + UI, how to proceed with the UI patch while the bug is a sec bug?
- Target landing 67 ? dolske
- selena to figure out how to get UX support
Meeting invites

Nick

1502678 - Referer not sent with rel="noopener” (not us - docshell doesn’t even get referrer)
1503578 - Check h2 logs for why ctrl+f5 not reloading all resources (seems 0rtt related)
1312057 - Prefetch use telemetry (in prep for re-enabling on prerelease channels)
Intermittent Raptor slowness (server related?)
1504337 - Consider using http/1.1 on cell connections (interesting paper linked)
1501357 - Funky protocol field in devtools
Landed h2ws
1505388 - h2 proxy crash (fallout from h2ws)
1498661 - Conn racing - need to fix dns lookup on secondary conn

Michal

1503578 Ctrl + F5 doesn't load all the resources of the web
1455723 Firefox59 does not properly honor cache size set in autoconfig files
1500533 Ensure that TLS session tickets are only consumed if storage checks on the channel pass

Honza

BT,

1418134 - sec bug (nav spoof?),

1482084 - sec bug (stream data crash),

Bug 1505493 - Retarget nsHttpChannel handled by HttpChannelParent to a background thread, avoid main thread loop on critical path

Bug 1482599 - Landsend online purchase payment error (let users easily run logging locally)

Bug 1303314 - network.automatic-ntlm-auth.allow-non-fqdn does not work (need to look into the code)

Daniel

(can’t attend today)

TRR study #5 work. Will measure transfer times using Akamai servers

Bug 1504109 - archive.is doesn’t work when DoH is used

Bug 1503517 - whitespace cleaning for X-Content-Type-Options