Meeting 2024-11-12

• Team calendar

  • Next on Necko triage: Kershaw

• New contributors / Friends of Necko THANK YOU!!

  • Mike Hommey [:glandium]

    • Bug: 1928309 Adjust the User-Agent string on iOS

  • John Schanck [:jschanck]

    • Bug: 1926127 Enable mlkem768x25519 in h3 in early beta and earlier

  •  Oskar Mansfeld

    •  Bug: 1761248 Expose API to report Early hints information for devtools

  •  edwyn.zhou

    • Bug: 1926989 Preference name misspelled: "dom.security.https_first_add_exception_on_failiure

  • Cosmin Sabou [:CosminS]

    • Bug: 1925999 Perma [tier 2] netwerk/test/unit/test_http3.js | xpcshell return code: 0 | testOnStartRequest - [testOnStartRequest : 155] 0 == 2152398919

• Round Table

  • – enter topics –

• Active Projects roadmap Fx132 rel Oct 29, Fx133 rel Nov 26, Fx134 code freeze Nov 21 

  • Firefox Essential Internal Services Toughening[FEISTY] (in progress, TBD)

    • – enter updates –  

  • DoH Enablement - Fenix and Focus (in progress, TBD)

    • [Greg] Pinged Amedyne Moya to request dev support, TBD

    • – enter updates –

  • 3rd Party Cookie Deprecation (in progress, target Fx134)

    • [Ed] CHIPS re-rollout backed out, likely another attempt this week.

    • Tentative 3PCD/CHIPS rollout:

      • 134: 

        • 3PCD experiment

        • re-enable chips, ride the trains to release (Jan 7)

          • Note: CHIPS-related breakage

      • >134

        • (pending experiment): 3PCD in all windows in release

          • Note: 3pcd-related  breakage

      • >136

        • (pending all-windows): “Migration”/deletion of TCP cookies (non-chips partitioned)

  • Fetch keepalive: (riding the trains, Target Fx133) 

    • – enter updates – 

  • Firefox/Chrome benchmark test suite (in progress, Fx132)

    • First cut of download results: https://lookerstudio.google.com/reporting/f5fbaacd-a473-4daf-bf22-faa3d3c8391f/page/DO9EE

      • Update TC scripts 

      • Ship this week or next 

    • [kershaw] 

      • Bug 1930638 When running upload/download test, make sure a clean connection is used in each run

        • R?

      • Latest result of download test

unthrottled

fast5G

busy5G

slow3G

Firefox

953.96

220.44

41.4

0.92

Chrome

585.92

175.36

29.1

0.92

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] Wrote patch for https://bugzilla.mozilla.org/show_bug.cgi?id=1922028 

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • – enter updates –

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • milestone 1 - basic end-to-end integration

    • Report of quinn-udp breaking HTTP3/QUIC on Windows on ARM

      • Narrowed down the bug to the UDP segmentation system.

      • In touch with Microsoft (Nick Banks, MSQUIC author).

      • Trying to get my hands on a Windows on ARM machine to speed up debugging cycles. See Slack conversation.

  • milestone 3 - Benchmark result for download and upload

    • Use simulated time in qlogs (i.e. now), thus allowing us to easily plot congestion control and flow control behavior using various network scenarios. See mozilla/neqo#2211.

    • Add QUIC frame Glean metrics.

      • See bugzilla#1929818.

      • E.g. answers questions like: how many migrations do we see? Have we improved receive window auto-tuning (flow-control), …

• Implement WebTransport based on HTTP/2

  • [kershaw]

    • Just started

• Performance work: 

  • [acreskey]

    • FYI, proposal on solution to Fenix slow pageload progress bar

      • Bug 1906726 - Fenix pageload progress bar can take multiple times longer to complete than Chrome (e.g. 50 seconds vs 10 seconds)

  • Response protocol version and HTTP/3:

    • query - protocol version from root document request - about 14-18%

    • Other sources report ~30% of requests:

      • https://httparchive.org/reports/state-of-the-web#h3

      • https://w3techs.com/technologies/details/ce-http3

    • Useful to have a query based on migrated probe:

      • https://glam.telemetry.mozilla.org/fog/probe/networking_http_response_version/explore?aggKey=http_3&aggType=sum&app_id=beta&ping_type=*

Meeting 2024-11-05

• Team calendar

  • Next on Necko triage: Valentin

• New contributors / Friends of Necko THANK YOU!!

  • Henri Sivonen (:hsivonen)

    • Bug: 1887898 Improper URL parsing in servo/rust-url

  • John Schanck [:jschanck]

    • Bug: 1926127 Enable mlkem768x25519 in h3 in early beta and earlier

  • Robert Longson [:longsonr]

    • Bug: 1928694 Replace mozilla::clamped with std::clamp

  • Andrea Marchesini [:baku]

    • Bug: 1928123 Crash [@ IsCurrentThread]

  • Cosmin Sabou [:CosminS]

    • Bug: 1925999 Perma [tier 2] netwerk/test/unit/test_http3.js | xpcshell return code: 0 | testOnStartRequest - [testOnStartRequest : 155] 0 == 2152398919

• Round Table

  • [Greg] Globs reported networking issue NS_ERROR_CONNECTION_REFUSE hitting server on local LAN, Sunil looking at logs 

• Active Projects roadmap Fx132 rel Oct 29, Fx133 rel Nov 26, Fx134 code freeze Nov 21 

  • Firefox Essential Internal Services Toughening[FEISTY] (in progress, TBD)

    • Approved - ready to go

    • Have prototype 

  • DoH Enablement - Fenix and Focus (in progress, TBD)

    • Approved - ready to go

  • 3rd Party Cookie Deprecation (in progress, target Fx134)

    • [Ed] Due to CHIPS incident and 3PCD experiment, we are pushing back the rollout plan, probably won’t make 134:

      • 134: 

        • 3PCD experiment

        • re-enable chips, ride the trains to release (Jan 7)

          • Note: CHIPS-related breakage

      • >134

        • (pending experiment): 3PCD in all windows in release

          • Note: 3pcd-related  breakage

      • >136

        • (pending all-windows): “Migration”/deletion of TCP cookies (non-chips partitioned)

  • Fetch keepalive: (riding the trains, Target Fx133) 

    • – enter updates – 

  • Firefox/Chrome benchmark test suite (in progress, Fx132)

    • First cut of download results: https://lookerstudio.google.com/reporting/f5fbaacd-a473-4daf-bf22-faa3d3c8391f/page/DO9EE

      • Update TC scripts 

      • Ship this week or next 

    • Needs validation, maybe hand testing 

    • Related: Chromium’s loadline framework

      • https://chromium.googlesource.com/crossbench/+/refs/heads/main/config/benchmark/loadline

    • (greg) Perf Infra Hardware review Performance Hardware Review

      • How long test take to run and how many tests are we running 

    • [kershaw]

      • Landed a change to make the test more reliable.

• Interop 2024 URL (in progress, target H2 2024)

  • No progress

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • – enter updates –

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • – enter updates –

• Implement WebTransport based on HTTP/2

  • – enter updates –

• Performance work: 

  • Fenix pref-flip experiment live and actually working

    • https://experimenter.services.mozilla.com/nimbus/fetch-priority-fenix-nightly/summary

    • Related bug on about:studies

      • https://bugzilla.mozilla.org/show_bug.cgi?id=1928658

  • In-progress experiments:

    • https://mozilla-hub.atlassian.net/wiki/spaces/FIREFOX/pages/11047744/Experiments+Necko#Schedule

  • Working on:

    • Bug 1906726 - Fenix pageload progress bar can take multiple times longer to complete than Chrome (e.g. 50 seconds vs 10 seconds)

  • Pageload Event properties by TRR domain

    • https://sql.telemetry.mozilla.org/dashboard/pageload-event-by-trr_domain?edit&p_channel=nightly&p_cutoff_value_seconds=60&p_days=20&p_group_by=submission_date&p_minimum_entries_group=10000&p_product=firefox_desktop

      • Note: only counts non-zero dns lookup times

        • https://searchfox.org/mozilla-central/source/dom/base/Document.cpp#2142-2147

  • [sunil]

    • OnDataFinished and OnStopRequest probes indicate we still experience delays in between dispatching and running OnDataFinished in the background thread. 

    • Modified code to dispatch directly to the background thread skipping the queue. Testing ongoing.

Meeting 2024-10-29

• Team calendar

  • Next on Necko triage: Randell 

• New contributors / Friends of Necko THANK YOU!!

  • None in the last week. 

• Round Table

  • – enter public topics –  

• Active Projects roadmap Fx132 rel Oct 29, Fx133 rel Nov 26, Fx134 code freeze Nov 21 

  • Firefox Essential Internal Services Toughening[FEISTY]

    • [Greg] Passed internal review, now in strategic review

  • DoH Enablement - Fenix and Focus

    • [Greg] Passed internal review, now in strategic review

  • 3rd Party Cookie Deprecation (in progress, target Fx134)

    • 3PCD timeline

      • 131: Enable in nightly all windows(dry run)

        • Check back once 131 hits release and see the telemetry

      • 132: Enable in PBMode and ETP Strict to release 

        • Tracking breakage 

      • 133: Release telemetry on partition limit will start, review it (we are here)

        • – enter updates –

      • 134: All windows in release (targeting January)

      • 136: “Migration”/deletion of TCP cookies (non-chips partitioned)

    • [Ed] Ongoing investigations and discussions with privacy team for google-auth followup

  • Fetch keepalive: (riding the trains, Target Fx133) 

    • – enter updates – 

  • HTTP Priorities Experiments (riding the trains, target Fx 132)

    • [Greg] Released today closing!

    • [ACreskey] First experiment that selects only pages that don’t make use of fetchpriority (images). Promising results

      • https://protosaur.dev/perf-reports/fetchpriority-nightly_only_docs_with_fetchpriority_images.html#All-lcp_time

      • 2.6% improvement in mean lcp, up to 7% at P80

      • Following up on Fenix (nightly) and Desktop (beta)

    • [skim working on:]

      • 1926407 - Consider modifying the fetchpriority adjustments for so that priority="high" is given a higher priority than auto

  • Firefox/Chrome benchmark test suite (in progress, Fx132)

    • First cut of download results: https://lookerstudio.google.com/reporting/f5fbaacd-a473-4daf-bf22-faa3d3c8391f/page/DO9EE

    • Needs validation, maybe hand testing 

    • Related: Chromium’s loadline framework

      • https://chromium.googlesource.com/crossbench/+/refs/heads/main/config/benchmark/loadline

    • (greg) Perf Infra Hardware review Performance Hardware Review

    • – enter updates –

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] Landed Add defaultUri bypass and Enable defaultURI 

  • URL interop score moves to 99.3% of our commitments (96.8% if we consider all of URL test scores)

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • – enter updates –

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • milestone 1 - basic end-to-end integration

    • On Android x86 Focus and Fenix CI startup tests have failed due to a seccomp panic after a `recvmmsg` call.

      • Upgrade mozilla-central to quinn-udp v0.5.6 ready for review.

        • Upgrade was previously blocked due to version conflict in transient depedency, namely windows-sys v0.52.

        • After larger discussion, we found a solution not involving upgrading many crates in the Rust ecosystem.

  • milestone 3 - Benchmark result for download and upload

    • Add 100 MiB upload benchmark to Neqo CI.

    • Optimize neqo-client send path for more accurate benchmark results.

    • Simulating gbit link up- and download via Neqo network simulator.

  • milestone 4 - multi-packet IO

    • Lars' quinn-udp pull request merged, adding experimental multi-message support on Apple platforms.

    • Adjusted quinn-udp benchmark to measure GSO, GRO and recvmmsg performance.

  • milestone 5 - multi-packet state machines

    • Pull request stack, removing major allocations in Neqo UDP receive path is merged.

    • Mozilla-central upgrading to latest release (v0.10.0) containing the optimization happening in phabricator#D226823.

    • Minor follow-up for the Neqo test server is pending.

  • Identify where bottlenecks are is it GSO or other

• Implement WebTransport based on HTTP/2

  • – enter updates –

• Performance work: 

  • [acreskey]

    • Expecting to start these four experiments tomorrow

      • https://mozilla-hub.atlassian.net/wiki/spaces/FIREFOX/pages/11047744/Experiments+Necko#Schedule

        • https://experimenter.services.mozilla.com/nimbus/fetch-priority-fenix-nightly/summary (Fenix nightly)

        • https://experimenter.services.mozilla.com/nimbus/fetchpriority-beta-pageload-event/summary (Desktop beta)

        • https://experimenter.services.mozilla.com/nimbus/performance-metric-calibration-with-lcp-nightly/summary (A/A/A/A on Desktop nightly)

        • https://experimenter.services.mozilla.com/nimbus/fenix-performance-metric-calibration-beta-followup/summary (A/A/A on Fenix beta)

    • Based on Randell’s profiling of main thread, ran this test and logged this one:

      • Bug 1927538 -  URL Classification can delay AsyncOpen-to-transaction-pending by 26-25%, visual metrics by 3-10% in raptor pageload test

    • Great help from Sean Kim:

      • 1926407 - Consider modifying the fetchpriority adjustments for so that priority="high" is given a higher priority than auto

      • Network Marker Profiler upgrades

        • Bug 1915717 -  [network-markers] Expose http protocol version for a request in the profiler network tab

        • Bug 1919148 -  [network-markers] Expose class of service for each request in the profiler network tab

  • [sunil]

    • OnDataFinished and OnStopRequest probes indicate we still experience delays in between dispatching and running OnDataFinished in the background thread. 

    • Did local browser tests to confirm this. ChannelEventQueue design seems to be a possible bottleneck. Will investigate further this week.

https://docs.google.com/document/d/1RI237uPR3Yl-dXmmYfeCFJl75T6FUxJ5GsbF3dkEgX8/edit?tab=t.0

Meeting 2024-10-22

• Team calendar

  • Next on Necko triage: Sunil

• New contributors / Friends of Necko THANK YOU!!

  • Oskar Mansfeld

    • Bug: 1873874 Crash [@ mozilla::net::Predictor::Action::OnCacheEntryAvailable] via IPC method `NeckoParent::RecvPredPredict` 

    • Bug: 1922606 Remove or update metrics expiring in Firefox 134: networking.cookie_c* before Gecko version gets increased to 134 on 2024-10-28

  •  Maxime Thiebaut

    •  Bug: 1915982 isLocalIPv4 should consider 0.0.0.0/8 a local address per RFC1122 

• Round Table

  • No public topics this week  

• Active Projects roadmap Fx132 rel Oct 29, Fx133 freeze Oct 24 

  • 3rd Party Cookie Deprecation (in progress, target Fx134)

    • 3PCD timeline

      • 131: Enable in nightly all windows(dry run)

        • Check back once 131 hits release and see the telemetry

      • 132: Enable in PBMode and ETP Strict to release 

        • Tracking breakage 

      • 133: Release telemetry on partition limit will start, review it (we are here)

        • – enter updates –

      • 134: All windows in release (targeting January)

      • 136: “Migration”/deletion of TCP cookies (non-chips partitioned)

    • [Ed] Handling, investigating, and code review for google auth incident (CHIPS issue)

  • Fetch keepalive: (riding the trains, Target Fx133) 

    • [Sunil] Sent intent to ship. No objections so far. 

• HTTP Priorities Experiments (riding the trains, target Fx 132)

  • [acreskey] Code to run FetchPriorities experiment on Android in review

    • https://phabricator.services.mozilla.com/D226053

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • First cut of download results: https://lookerstudio.google.com/reporting/f5fbaacd-a473-4daf-bf22-faa3d3c8391f/page/DO9EE

  • Needs validation, maybe hand testing 

  • Related: Chromium’s loadline framework

    • https://chromium.googlesource.com/crossbench/+/refs/heads/main/config/benchmark/loadline

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] code clean up and try testing on defaultURI

    • added pref flip patch

    • Trying to land this week

    • Expectation is we’ll get to ~95%

      • Detailed update (slack)

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • – enter updates –

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • Presented progress at MIR^3

    • See slides.

  • milestone 5 - multi-packet state machines

    • Don't allocate in UDP receive path close to merge, see mozilla/neqo#2184.

  • Apple platform

    • Support for sendmsg_x and recvmsg_x in quinn-udp making progress, see quinn-rs/quinn#1993 and quinn-rs/quinn#2010.

• Implement WebTransport based on HTTP/2

  • No update

• Performance work: 

  • [acreskey] thanks to Sean Kim for adding navStart to request start to pageload event

    • https://bugzilla.mozilla.org/show_bug.cgi?id=1923894

Meeting 2024-10-15

• Team calendar

  • Next on Necko triage: Ed

• New contributors / Friends of Necko THANK YOU!!

  • Andrea Marchesini [:baku]

    • Bug: 1922422 Assertion failure: IsOnBackgroundThread(), at /builds/worker/checkouts/gecko/ipc/glue/BackgroundImpl.cpp:171

    • Bug: 1923872 Enable server-time to compute the max-age 

  •  John Schanck [:jschanck]

    • Bug: 1918532 Replace xyber768d00 with mlkem768x25519 in neqo_glue 

  • Oskar Mansfeld

    • Bug: 1923031 Can we avoid recording networking.http3_enabled on content processes? 

    • Bug: 1306801 Remove Necko IPC security 

  • Tim Huang[:timhuang]

    • Bug: 1922193 Sending multiple (unknown) cookies with same name 

• Round Table

  • [valentin] Related to 1910991 above, I tried using wireshark to see the decoded requests.

    • ECH is enabled on Cloudlare, which means Wireshark is unable to decode the packets. Yaroslav Rosomakho from zscaler has patches for NSS and Wireshark - will submit them next week. Until then network.dns.native_https_query:false and network.trr.mode:5 should avoid ECH

• Active Projects roadmap Fx132 rel Oct 29, Fx133 freeze Oct 24 

  • 3rd Party Cookie Deprecation (in progress, target Fx134)

    • 3PCD timeline

      • 131: Enable in nightly all windows(dry run)

        • Check back once 131 hits release and see the telemetry

      • 132: Enable in PBMode and ETP Strict to release 

        • Tracking breakage 

      • 133: Release telemetry on partition limit will start, review it (we are here)

        • – enter updates –

      • 134: All windows in release (targeting January)

      • 136: “Migration”/deletion of TCP cookies (non-chips partitioned)

  • Fetch keepalive: (in progress, Target Fx133) 

    • [Sunil] Pref to enable by default landed. 

    • [Sunil] Landed patch to increase the keepalive request quota as we saw request drops due to limits.

• HTTP Priorities Experiments (riding the trains, target Fx 132)

  • Experiments running

  • [acreskey] Results from Nightly Experiment (where we can compare pages that use fetchpriority=high on images) look good but we will need a calibration and beta experiment to confirm

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • – enter updates –

• Interop 2024 URL (in progress, target H2 2024)

  • (sekim) Landed Scheme Bug 1883033 - ws/wss/ssh url query-encoding wpt failures

  • (sekim) Setting non-spec:/.pathname = '/.//p' (rust-url) ready for review

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • – enter updates –

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • – enter updates –

• Implement WebTransport based on HTTP/2

  • – enter updates –

• Performance work: 

  • [acreskey] Exposing nimbus features to geckoview - DoH, Kyber, #sockets

    • If there any you need, let me know

  • FYI, CRLite experiment

    • https://protosaur.dev/perf-reports/crlite-filter-channels.html

    • https://experimenter.services.mozilla.com/nimbus/crlite-filter-channels/results

  • FYI, added some useful queries to our experiments page

    • https://mozilla-hub.atlassian.net/wiki/spaces/FIREFOX/pages/11047744/Experiments+Necko#Queries

Meeting 2024-10-08

• Team calendar

  • Next on Necko triage: Kershaw

• New contributors / Friends of Necko THANK YOU!!

  • Andrea Marchesini [:baku] 

    • Bug: 1922422 Assertion failure: IsOnBackgroundThread(), at /builds/worker/checkouts/gecko/ipc/glue/BackgroundImpl.cpp:171

    • Bug: 1921495 Enable schemeful cookie in nightly

    • Bug: 1484026 Add IPC Checks to Get/SetCookie methods   

• Round Table

  • [acreskey] HTTP/3 issue, appears to only be on cloudflare servers

    • https://bugzilla.mozilla.org/show_bug.cgi?id=1910991#c6

• Active Projects roadmap Fx132 rel Oct 29, Fx133 freeze Oct 24 

  • 3rd Party Cookie Deprecation (in progress, target Fx134, Fx133?)

    • 3PCD timeline

      • 131: Enable in nightly all windows(dry run)

        • Check back once 131 hits release and see the telemetry

      • 132: Enable in PBMode and ETP Strict to release 

        • Tracking breakage 

      • 133: Release telemetry on partition limit will start, review it (we are here)

        • Minimal user impact, proceeding with disabling dry-run

      • 134: All windows in release (targeting January)

      • 136: “Migration”/deletion of TCP cookies (non-chips partitioned)

  • Fetch keepalive: (in progress, Target Fx133) 

    • [Sunil] Pref to enable by default landed. 

    • Telemetry data indicates the feature is being used widely.

    • Added relnote-firefox flag and dev-doc-needed

    • Anything else left before this is shipped? Intent to ship email?

      • https://mozilla-hub.atlassian.net/wiki/spaces/FIREFOX/pages/853344520/Process

• HTTP Priorities Experiments (riding the trains, target Fx 132)

  • Experiments running

    • Writing queries to compare pages where fetchPriority is used

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • [kershaw]

    • Bug 1895361 

      • R?

      • Should be landing soon

• Interop 2024 URL (in progress, target H2 2024)

  • DefaultURI bypass patch: comments addressed another review cycle

  • Created Meta for file URL’s with hostnames sec mitigations

    • If you have any cycles and would like to help out with interop 2024, might be a good place to start

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • (sekim) Filed few bugs for future P1 glean probe migration (moved to next, marked them as P2 N/A)

  • (sekim) Added alias for Bug 1901798 (necko-ifndef-removal)

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • milestone 5 - multi-packet state machines

    • Don't allocate in UDP send & receive path back to draft github#2093. Will be split up into smaller atomic patches.

• Implement WebTransport based on HTTP/2

  • (sekim) Currently trying to test if FAILS component supports WebTransport over H2

• Performance work: 

  • Focusing on Android issues with connection migration

Meeting 2024-10-01

• Team calendar

  • Next on Necko triage: Randell

• New contributors / Friends of Necko THANK YOU!!

  • Eden Chuang[:edenchuang]

    • Bug: 1843001 GeckoView fails to return web requests after GeckoNetworkManager is marked as down

  • Andrea Marchesini [:baku]

    • Bug: 1920104 Improve WPT test coverage for CookieStore API

  • tannal2409

    •   Bug: 1917925 Remove network.http.strict_response_status_line_parsing pref 

• Round Table

  • No public topics this week

• Active Projects roadmap Fx131 rel Oct 01, Fx132 freeze Sep 26

  • 3rd Party Cookie Deprecation (in progress, target Fx134)

    • 3PCD shipped to nightly

    • Tracking breakage: https://bugzilla.mozilla.org/show_bug.cgi?id=1917788 

    • 3PCD timeline

      • 131: Enable in nightly all windows(dry run)

        • Check back once 131 hits release and see the telemetry

      • 132: Enable in PBMode and ETP Strict to release (we are here)

      • 133: Release telemetry on partition limit will start, review it

      • 134: All windows in release (targeting January)

      • 136: “Migration”/deletion of TCP cookies (non-chips partitioned)

  • Fetch keepalive: (in progress, Target Fx133) 

    • – enter updates –

  • HTTP Priorities Experiments (riding the trains, target Fx 132)

    • [acreskey]

      • Beta followup experiment launching today

      • Nightly followup launching today

        • 32-bit int in pageloadEvent to track pages that use FetchPriority on images. See https://phabricator.services.mozilla.com/D222991

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • [kershaw]

    • Bug 1895361 

      • Address review comments

      • Refine conditions used for network throttling

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] Made fixes to patch https://phabricator.services.mozilla.com/D207810

    • More review comments to address

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • – enter updates –

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • milestone 2 - ECN

    • Transform http_3_ecn_ce_ect0_ratio to custom_distribution to be displayable in Glean bugzilla#1921203

  • milestone 4 - multi-packet IO

    • Proof of concept GSO implementation, see mozilla/neqo#2138.

    • Proof of concept recvmmsg implementation, see mozilla/neqo#2138.

  • milestone 5 - multi-packet state machines

    • Don't allocate in UDP send & receive path ready for full review, see github#2093.

• Implement WebTransport based on HTTP/2

  • – enter updates –

• Performance work: 

  • [acreskey] Moving focus to handling of network transitions on Android, HTTP/3

Meeting 2024-09-17

• Team calendar

  • Next on Necko triage: Sunil

• New contributors / Friends of Necko THANK YOU!!

  • Kagami Rosylight [:saschanaz] (they/them)

    • Bug: 1907428 crash near null in [@ mozilla::dom::FetchChild::RecvOnResponseAvailableInternal]

  • Tannal2409

    • Bug: 1917924 Remove network.http.reject_NULs_in_response_header_values pref 

• Round Table

  • [greg] Manuel has raised this discussion regarding enabling DNS over SOCKS4 by default? 

    • SOCKSv4 does not support DNS SOCKSv4a does however we have no ability to determine version 

• Active Projects roadmap Fx131 rel Oct 01, Fx132 freeze Sep 26

  • 3rd Party Cookie Deprecation (in progress, target Fx134)

    • 3PCD shipped to nightly

    • Tracking breakage: https://bugzilla.mozilla.org/show_bug.cgi?id=1917788 

    • 3PCD timeline

      • 131: Enable in nightly all windows(dry run)

        • Check back once 131 hits release and see the telemetry

      • 132: Enable in PBMode and ETP Strict to release (we are here)

      • 133: Release telemetry on partition limit will start, review it

      • 134: All windows in release (targeting January)

      • 136: “Migration”/deletion of TCP cookies (non-chips partitioned)

  • Fetch keepalive: (in progress, Target Fx132 (check again on Friday if we still can make it to Fx132) 

    • Final Milestone

    • 2/4 patches r+d

    • [Sunil] Discussed pending issues  for worker keepalive with Eden Chuang

      • Incorporating the changes and testing

  • HTTP Priorities Experiments (in progress, target Fx 132)

    • [acreskey]

      • fetchpriority=high on images working very well, see 1797715 - [meta] Fetch Priority (was Priority Hints), propose we let ride the feature ride the trains

      • Working on documentation, feedback if you have any

        • WIP documentation:

          • https://github.com/acreskeyMoz/acreskeyMoz.github.io/blob/main/docs/prioritization.md

          • Bug 1915852 - Document network request scheduling and prioritization system

      • Possible bug:

        • Bug 1919158 -  Request urgency may ignore fetchPriority and supportsPriority in two scenarios

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • [kershaw]

    • Optimize server setup

      • H3 download: 838.66mbps -> 914.76mbps

      • Need ~8Gbps to verify improvement w UDP 

    • Working on TC scripts

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] Addressing review comments for defaultURI

    • Including getting data model created for remote settings server side

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • – enter updates –

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • milestone 1 - basic end-to-end integration

    • Report of network.http.http3.use_nspr_for_io false breaking HTTP3/QUIC on Windows on ARM

      • Still only reproducible on one Windows on ARM machine.

      • Debugging in progress.

  • milestone 2 - ECN

    • Landed in Firefox Nightly.

    • See implementation in Github#1678, Bugzilla#1902065.

    • See various metric sources in Glean Dictionary, Glam and Looker.

  • milestone 3 - Benchmarks and metrics

    • Adding UDP datagram (segment) size metrics (phabricator#D216034) landed. Should be available in Firefox Nightly soon.

    • Migrating Http3Session::CloseConnectionTelemetry to Glean.

      • Work in progress phabricator#D221615.

  • milestone 5 - multi-packet state machines

    • Don't allocate in UDP send & receive path (github#2093).

      • Ready for review.

      • "Early benchmarks are promising, showing e.g. a 10% improvement in the Download benchmark, and up to 40% improvement in the neqo-neqo-reno-pacing benchmark."

• Performance work: 

  • [acreskey] 

    • Paused the work on evaluating/tuning the ClassOfService tailing system until we ship HTTP Priorities

      • https://bugzilla.mozilla.org/show_bug.cgi?id=1915866

    • FYI: another profiler bug

      • Bug 1919148 -  [network-markers] Expose class of service for each request in the profiler network tab

Meeting 2024-09-10

• Team calendar

  • Next on Necko triage: Ed

• New contributors / Friends of Necko THANK YOU!!

  • amarchesini@mozilla.com

    • 1903403 - Do not reject cookies containing %x9

    • 1913161 - Implement a max-age cap for cookies

    • 1909980 - Duo 2fa hits "403 forbidden" if your system clock is wrong, only in Firefox

  • aborovova@mozilla.com

    • 1916105 - Perma /fetch/api/basic/request-upload.h2.any.html | Fetch with POST with empty ReadableStream - assert_equals: expected "" but got "[object ReadableStream]" when Gecko 131 merges to beta on 2024-09-02

  • tannal2409@gmail.com

    • 1916102 - Remove network.url.strict_protocol_setter pref

• Round Table

  • [kershaw] Bugs that might affect page load metrics

    • 1917528

      • Duplicate telemetry recording for mDefaultLoadRequest
      • Fixed

    • 1678312

      • TLS handshake timings are not correct
      • About to land

    • [Kershaw] Post in #perf #tls

• Active Projects roadmap Fx131 rel Oct 01, Fx132 freeze Sep 26

  • 3rd Party Cookie Deprecation (in progress, target Fx134)

    • No changes again (sept 10). 3PCD will proceed in a phased approach

      • 131: Enable in nightly all windows(dry run)

        • Check back once 131 hits release and see the telemetry

      • 132: Enable in PBMode and ETP Strict to release

      • 133: Release telemetry on partition limit will start, review it

      • 134: All windows in release (targeting January)

      • 136: “Migration”/deletion of TCP cookies (non-chips partitioned)

  • Fetch keepalive: (in progress, Target Fx132) 

[Sunil]

• Incorporated feedback from Eden, waiting for final patch approval

• HTTP Priorities Experiments (in progress, target Fx 132)

  • [acreskey]

    • Bug 1915490 -  privacy.annotate_channels.strict_list.enabled in Nightly and Beta leads to slower page load

    • Bug 1915186 -  UrlClassifierCommon::AnnotateChannel lowers priority of channels requested at high priority

    • Bug 1915866 -  Optimize Class of Service tailing and tracker request postponement

      • Experiment Brief incoming

    • Meta: 1797715 - [meta] Fetch Priority (was Priority Hints)

    • HTTP/2 extensible prioritization scheme would be helpful

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • [kershaw]

    • Chrome support implemented (patch not landed yet)

    • How to run the test:

      • ./mach try fuzzy –full

      • test-linux1804-64-qr/opt-browsertime-network-bench-firefox-h3-upload

        • H3 Download: Firefox 850.45 Mbps, Firefox NSPR 838.01 Mbps, Chrome 127 620.6 Mbps
        • H3 Upload: Firefox 349.24 Mbps, Firefox NSPR NA, Chrome 127 547.85 Mbps

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] Addressing more feedback on defaultURI phab/D207810  

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • (sekim) Migrated few other probes using glean’s new labeled timing distribution

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • Milestone 1 - end-to-end integration

    • On Android x86 Focus and Fenix CI startup tests have failed due to a seccomp panic after a `recvmmsg` call.

      •  Fix blocked on mozilla-central moving to `windows` crate `v0.59.0`. See Matrix discussion.

    • Report of network.http.http3.use_nspr_for_io set to false breaking HTTP3/QUIC on _Windows on ARM.

      • Thus far not reproducible on other _Windows on ARM_ machines. Both native and as VM.

      • See further discussion in Slack thread.

  •  Milestone 2 - ECN

    • Neqo v0.9.0 released, exposing ECN metrics in neqo_transport::Stats.

      • Landed in mozilla-central to phabricator#D221019.

      • Update triggers build failure on tier 2 Windows 2012 platform which is now[fixed with  phabricator#D221284.

    • Update phabricator#D220062 to use Glean to expose ECN metrics.

  • Milestone 3 - Benchmarks and metrics

    • Intermediate Thread-local buffered metrics proposal phabricator#D216034 ready for review.

    • Long term fix in Glean is work-in-progress mozilla/glean#2948.

    • Draft for migrating Http3Session::CloseConnectionTelemetry to Glean bugzilla#1917778

  • Milestone 5 - multi-packet state machine

    • Draft removing allocations in UDP IO send and receive path mozilla/neqo#2093

• Performance work: 

  • [acreskey] 

    • Bug 1915700 - Expose socket process network markers with MarkerThreadId::MainThread()

    • Fyi: Bug 1915717 -  [network-markers] Expose http protocol version for a request in the profiler network tab

    • Helping Robin Marx get answers to priority questions

      • Bug 1915852 -  Document network request scheduling and prioritization system

      • Bug 1917649 -  Duplicate img cached requests appear in the network monitor

    • Performance Telemetry Reports no longer public

  • New experiments:

    • OMT Decompression for Beta complete next steps??

Meeting 2024-09-03

• Team calendar

  • Next on Necko triage: Kershaw

• New contributors / Friends of Necko THANK YOU!!

  • Andrea Marchesini [:baku]

    • Bug: 1903403 Reject cookies containing %x9 

    • Bug: 1913161 Implement a max-age cap for cookies 

  • Andreas Farre [:farre]

    • Bug: 1912551 Crash in [@ mozilla::net::HttpBaseChannel::ShouldFilterOpaqueResponse] - MOZ_DIAGNOSTIC_ASSERT(ShouldBlockOpaqueResponse()) 

  • Mark Banner (:standard8)

    • Bug: 1914620 Enable remaining ESLint rules on dom/websocket  

  • Eric Chen

    • Bug: 1914995 Remove an unnecessary #include in MicrosoftEntraSSOUtils.mm 

  • Tannal2409

    • Bug: 1915613 Remove test_http3_large_post_telemetry.js since the probes are expired. 

  • Ted Campbell [:tcampbell]

    • Bug: 1915668 Support PUT method in ObliviousHttpChannel 

  • Alexandra Borovova [:Sasha]

    • Bug: 1916105 Perma /fetch/api/basic/request-upload.h2.any.html | Fetch with POST with empty ReadableStream - assert_equals: expected "" but got "[object ReadableStream]" when Gecko 131 merges to beta on 2024-09-02 

• Round Table

  • [valentin] 1894109 - Not caching byte range request responses 

    • Do we want to change our range caching decision logic?

    • Next: Valentin Identify how many HTTPChannels are range requests assess impact and effort

    • May need to do this as part to Compression Dictionary Transport also 

  • [valentin] 1915982 - isLocalIPv4 should consider 0.0.0.0/8 a local address per RFC1122 

    • Yes, lets consider as local address

  • [valentin] 1915848 - Pref off HTTP2 push 

    • It’s become a webcompat risk. I’ll send the intent to unship email soon.

    • Chrome disabled 2yrs ago, Safari TBD

• Active Projects roadmap Fx130 rel Sep 03, Fx131 freeze Aug 29

  • 3rd Party Cookie Deprecation (in progress, target Fx133)

    • No changes. 3PCD will proceed in a phased approach

      • 131: Enable in nightly all windows

      • 132: Enable in PBMode and ETP Strict to release

      • 134: All windows in release (targeting January)

      • 136: “Migration”/deletion of TCP cookies (non-chips partitioned)

  • Fetch keepalive: (in progress, Target Fx132) 

    • – enter updates – 

  • HTTP Priorities Experiments (in progress, target Fx 132)

    • Need to ship and confirm if priorities get reset

  • Firefox/Chrome benchmark test suite (in progress, Fx132)

    • Lars/Max to sync with Kershaw next week 

    • – enter updates – 

    • [kershaw]

      • Bug 1895362 Writing browsertime scripts for testing download and upload

        • WIP

        • Initial result shows no obvious difference between NSPR an quinn-udp

      • Bug 1895363 Reuse our current alerting/perfherder system to track results

        • WIP

  • Interop 2024 URL (in progress, target H2 2024)

    • [Ed] Patch fixup for defaultURI re-enable   

  • necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

    • [Acreskey] discussion on probes to remove

      • Necko probe migration discussion

    • – enter updates –

  • Fast UDP I/O in Firefox (in progress, target H2 2024)

    • Milestone 1 - end-to-end integration

      • Android x86 seccomp failure on recvmmsg fixed upstream in quinn-udp v0.5.5.

        • Mozilla-central update

      • Fix minor regression introduced in phabricator#D212959, not tracking bytes sent and received on socket for `about:networking#sockets`. See phabricator#D220379.

    •  Milestone 2 - ECN

      • Early blackhole detection on path migration by Lars merged.

        • Thus mitigating the impact of Android API Level 25 failing to send ECN marked datagrams.

        • See https://github.com/quinn-rs/quinn/pull/1975 for details.

      • Pending Neqo v0.9.0 release to bring https://github.com/mozilla/neqo/pull/2072 and https://github.com/mozilla/neqo/pull/2081 into mozilla-central 

      • Thus unblocking ECN pref and metrics https://phabricator.services.mozilla.com/D220062

    • Milestone 5 - multi-packet state machine

      • Proof-of-concept, removing allocation in UDP receive path.

        • https://github.com/mozilla/neqo/pull/2076

        • Has a mere 2% performance improvement on Download benchmark.

        • Proof-of-concept for send path in progress.

    • Glean metrics http3_udp_datagram_size_send and http3_udp_datagram_size_received

      • Ported buffered metrics proposal (mozilla/glean#2945) into phabricator#D216034. Local testing no longer shows significant CPU usage by Glean on 10 GB download.

      • Discussed steps forward with Jan-Erik (Glean team). They are able to reproduce the performance issue seen in phabricator#D216034 and have a local patch fixing it in Glean directly.

  • Performance work: 

    • – enter updates –

    • New experiments:

      • OMT Decompression + html5Parser, cssLoader for Beta

        • Experiment Brief

        • Data ticket

        • [Sunil] Experiment will re run, check with Sunil

          • We had meetings at MozWeek

          • Improving one of the probes

          • Looking for sites that show significant wins/losses with a local patch

Meeting 2024-08-20

• Team calendar

  • Next on Necko triage: Randell 

• New contributors / Friends of Necko THANK YOU!!

  • Andrea Marchesini [:baku]

    • Bug: 1909673 Enable cookie "sameSite=none requires secure" feature

    •  Bug: 1910886 Reduce duplicated code between CookieService and CookieServiceChild (nsICookieService.setCookieStringFromDocument/nsICookieService.getCookieStringFromDocument) 

  • Gregory Pappas [:gregp]

    • Bug: 1912853 Remove network.IDN.* pref

  •  Henri Sivonen (:hsivonen)

    • Bug: 1912011 Align the UTS 46 ASCII deny list with the URL Standard for non-nsStandardURL callers 

  • Tannal2409

    • Bug: 1841061 Remove "network.compress.allow_truncated_empty_brotli" pref if no regression forces this config to be flipped for a year. 

    • Bug: 1911935 Remove network.IDN_show_punycode pref

• Round Table

  • No public topics this week 

• Active Projects roadmap 

  • Native HTTPS RR Implementation (riding the trains, target Fx129)

    • [kershaw] Bug 1910593 proxyDNS is not respected when using proxy.onRequest

      • Ready to land

  • 3rd Party Cookie Deprecation (in progress, target Fx133)

    • – enter updates – 

  • Fetch keepalive: (in progress, Target Fx131) 

    • – enter updates – 

  • HTTP Priorities Experiments (in progress, target Fx130 )

    • Experiment Perf Reports:

      • https://mozilla-hub.atlassian.net/wiki/spaces/FIREFOX/pages/11047744/Experiments+Necko

    • Next steps: 

      • verify impact with webpagetest.org

      • Consider adding specific telemetry (i.e. only count where used)

  • Firefox/Chrome benchmark test suite (in progress, Fx132)

    • [kershaw] Bug 1895359 Install the Node.js server to handle file uploads and the Caddy server to provide HTTP/3 support.

      • Address comments

  • Interop 2024 URL (in progress, target H2 2024)

    • – enter updates – 

  • necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

    • – enter updates –

  • Fast UDP I/O in Firefox (in progress, target H2 2024)

    • Milestone 1

      • On Android x86 Focus and Fenix CI startup tests have failed due to a seccomp panic after a `recvmmsg` call.

        • Upstream bug discussion https://github.com/quinn-rs/quinn/issues/1947

        • Accepted upstream fix https://github.com/quinn-rs/quinn/pull/1964

        • Refined upstream fix after conversation with other `quinn` users on minimum supported Android version https://github.com/quinn-rs/quinn/pull/1966

      • On Firefox Nightly Windows we have seen multiple crash reports where `neqo-udp` fails to handle empty datagrams.

        • Fix has landed in mozilla-central through Neqo v0.8.2 release. 

      • On Firefox Nightly Linux x86-64 we have seen multiple crash reports due to a seccomp panic after a `recvmmsg` call.

        • This is due to users running `network.http.network_access_on_socket_process.enabled`, i.e. running the socket thread in a separate sandbox process 

        • Patch accepted and landing https://phabricator.services.mozilla.com/D219110.

    • General

      • Upstream Rust libc has merged Petr Sumbera's fixes for Illumos/Solaris

        • This is needed to run quinn-udp on Solaris.

        • See also discussion with Petr  and quinn pull request.

      • Discussed minimum supported Android API level with quinn folks.

        • Adjusted CI tests accordingly 

        • Quinn-udp will support Android 4, Firefox supports Android 5

  • Performance work: 

    • [acreskey]

      • FYI: Kyber on HTTP/3 in fx 128 experiment results

        • acreskeymoz.github.io/perf_reports/post-quantum-key-exchange-for-tls-and-h3-in-firefox-128/

      • Extending pageloadEvent: connectionTime, tls handshake

        • Considering removing trrDomain (expensive)

        • Throughput on document resource?

      • Progress with Mobile Testing on Firebase Test Lab live site Fenix/Chrome pageload tests

    • New experiments:

      • OMT Decompression + html5Parser, cssLoader for Beta

        • Experiment Brief

        • Data ticket

        • [Sunil] Experiment is live

          • We had meetings at MozWeek

          • Improving one of the probes

          • Looking for sites that show significant wins/losses with a local patch

    • https://bugzilla.mozilla.org/show_bug.cgi?id=1664492 - review strategy IPV6

    • [jesup] We should actively start looking at the DNS pre-warmup for intent startups with the android team - Bug 1880491

Meeting 2024-08-06

• Team calendar

  • Next on Necko triage: Ed

• New contributors / Friends of Necko THANK YOU!!

  • Max Inden

    • Bug: 1910360 Use quinn-udp instead of NSPR for QUIC I/O on Nightly

    • Bug: 1911658 Crash in [@ core::slice::impl$0::chunks]

  • :gerard-majax

    •  Bug: 1910821 Use correct mutex for gConnRecvMutex in FuzzyLayer

• Round Table

  • No public topics this week.

• Active Projects roadmap 

  • Native HTTPS RR Implementation (riding the trains, target Fx129)

  • 3rd Party Cookie Deprecation (in progress, target Fx133)

    • [Ed]

      • Need to make fixes to this patch

      • Tim should be back from PTO need to discuss getting chips partition limit to ride the rails

  • Fetch keepalive: (in progress, Target Fx131) 

    • Pushing release to Fx131 to accommodate additional scope to support workers.

    • [Sunil] Milestone 2 Changes implementation review complete, will merge after completion of test review.

    • [Sunil] Started work on Milestone 3 and extending WPTs for workers.

• HTTP Priorities Experiments (in progress, target Fx130 )

  • Experiment Perf Reports:

    • https://mozilla-hub.atlassian.net/wiki/spaces/FIREFOX/pages/11047744/Experiments+Necko

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • [kershaw]

    • Bug 1895359 Install the Node.js server to handle file uploads and the Caddy server to provide HTTP/3 support.

      • R?

    • Working on traffic shaping scripts

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] Addressing code review suggestions for 1889988

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

  • – enter updates –

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • Milestone 1

    • Enabled by default on Firefox Nightly https://phabricator.services.mozilla.com/D217830

    • Seccomp failure on Android x86

      • https://github.com/quinn-rs/quinn/issues/1947

      • Not able to reproduce in emulator thus far.

      • Using NSPR for now.

    • Crash when reading empty UDP datagram on Windows

      • https://bugzilla.mozilla.org/show_bug.cgi?id=1911658

      • Will be fixed with upcoming neqo v0.8.2.

      • https://github.com/mozilla/neqo/pull/2044

      • Using NSPR for now.

  • Milestone 2

    • Improve ECN blackhole handling merged.

    • Will be uplifted by upcoming neqo v0.8.2 https://github.com/mozilla/neqo/issues/2040

  • Glean metrics

    • Andrew Creskey built a dashboard.

• PMTUD for QUIC

  • Still blocked on test failure https://bugzilla.mozilla.org/show_bug.cgi?id=1909910#c3

  • Prioritizing Fast UDP I/O in Firefox milestone 2

• Performance work: 

  • [acreskey] 

    • Experiment Perf Reports:

      • https://mozilla-hub.atlassian.net/wiki/spaces/FIREFOX/pages/11047744/Experiments+Necko

    • Update to Network Performance Pulse dashboards

      • Pageload metrics by protocol version

    • Fenix slow pageload bugs

      • Fenix slow, stalled pageload bugs

  • New experiments:

    • OMT Decompression + html5Parser, cssLoader for Beta

      • Experiment Brief

      • Data ticket

      • [Sunil] Experiment is live

        • – enter updates –

Meeting 2024-07-16

• Team calendar

  • Next on Necko triage: Sunil

• New contributors / Friends of Necko THANK YOU!!

  • Andrea Marchesini [:baku]

    • Bug: 1903400 Reject cookies containing %xa and %xd for cookie WPTs 

    • Bug: 1903408 No cookie sent for redirected path mismatch WPT failure 

    • Bug: 1906914 Fix the max-age attribute parsing 

  • Henri Sivonen (:hsivonen)

    • Bug: 1889536 Migrate IDNA handling to ICU4X 

  • Tooru Fujisawa [:arai]

    •  Bug: 1906393 Add nsIIncrementalStreamLoader.onStartRequest 

  • Max Inden

    • Bug: 1907287 typo http3 `max_data` `max_stream_data` pref doc 

• Round Table

  • No public topics this week

• Active Projects roadmap 

  • Native HTTPS RR Implementation (riding the trains, target Fx129)

  • 3rd Party Cookie Deprecation (in progress, target Fx133)

    • No data being generated from CHIPS partition limit telemetry, investigating

  • Fetch keepalive: (in progress, Target Fx130) 

    • [Sunil] Milestone 2 Work Ongoing. 

      • Testing the implementation

      • Will post for review once tests are complete

  • HTTP Priorities Experiments (in progress, target Fx130 )

    • Experiment live

      • Priority header already shipped in Fx126

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1865394

      • Fx 128 includes:

        • Implement Extensible Prioritization Scheme for HTTP/2

          • https://bugzilla.mozilla.org/show_bug.cgi?id=1865040

        • Incremental flag on images, dom

          • https://bugzilla.mozilla.org/show_bug.cgi?id=1784496

        • [acreskey] Did not see clear, reproducible signal of impact, Fenix 4G browsertime test (realism over noise)

  • Firefox/Chrome benchmark test suite (in progress, Fx132)

    • [kershaw]

      • Close to having a prototype on test machines capable of running upload tests.

  • Interop 2024 URL (in progress, target H2 2024)

    • [Ed] DefaultURI bypass list - Adding tests for remote settings sync

  • necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks tracking such blocks using Bug 1901798  

    • (sekim) Migrating more high priority probes (https://phabricator.services.mozilla.com/D215537, https://phabricator.services.mozilla.com/D216204, https://phabricator.services.mozilla.com/D216620)

  • Fast UDP I/O in Firefox (in progress, target H2 2024)

    • [kershaw]

      • Improve test coverage by turning on HTTP/3 tests on Android. Ready for review.

    • [mxinden]

      • Glean probes

        • http3_udp_datagram_size_send

        • Http3_udp_datagram_size_received

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1906853

        • Need to measure performance impact of measurement itself. See discussion.

      • Milestone 1

        • Blocked on false-positive cargo vet failure.

          • Needs audit for `tracing*` and `tokio`.

          • Latest summary

        • Add Firefox `about:config` pref. Default is UDP IO via NSPR. We can later switch default to quinn-udp and eventually remove pref. Allows for fast roll-back in beta or release channel without full-blown patch release. See comment for details.

        • Simplify C++ -> Rust FFI, always using `i64`, thus preventing all implicit type casts.

        • Fix `RefCell` double-mut-borrow regression in neqo `v0.8.0` refactoring the `RefCell` away, thus making this impossible. Blocker to propagate neqo v0.8.0 changes into mozilla-central. 

        • Manual testing on Linux 64bit, Windows 64bit VM and MacOS (M2 with and without Rosetta)

        • Petr making progress on Solaris support for quinn-udp

      • Milestone 2

        • Improve ECN blackhole handling (thanks Lars!).

  • Performance work: 

    • [acreskey] Bug 1889771 - Increase network.http.max-connections on Android nightly (currently 40 (900 on desktop))

      • Landed telemetry on socket pool exhaustion https://bugzilla.mozilla.org/show_bug.cgi?id=1901855

      • No improvement measured so far on local browsertime tests (high latency, low bandwidth)

      • CI picked up loadtime improvements but vismet regressions, recordings missing

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1904643

      • Patch up that increases to 128 sockets

    • New experiments:

      • OMT Decompression + html5Parser, cssLoader 

        • Experiment Brief

        • Data ticket

        • [Sunil] Experiment is live

          • – enter updates –

Meeting 2024-07-09

• Team calendar

  • Next on Necko triage: Sunil

• New contributors / Friends of Necko THANK YOU!!

  • bwc (Byron Campen)

    • Bug: 1903905 Intermittent Assertion failure: stream != (0xFFFF), at /builds/worker/checkouts/gecko/netwerk/sctp/datachannel/DataChannel.cpp:2472 

  • serfreeman1337

    • Bug: 1810362 Firefox does not seem to support IPv6 unless DNS-over-HTTPS is enabled 

• Round Table

  • [acreskey] Summary of Fenix pain points:

    • PlayStore review summaries:

      • (June 21 - June 27)

        • “Users are reporting slow page loading, pages not loading, slow browsing speed, slow app responsiveness, and slow media.”

      • (June 14 - June 20)

        • Users are reporting slow page loading, pages not loading, slow browsing speed, slow app responsiveness, and slow media.

      • (May 31 - June 6)

        • Users are reporting slow page loading, pages not loading, slow browsing speed, slow app responsiveness, and slow media.

      • “Well, dear firefox. I like your performance.. but you MUST improve page loading speeds to be on par with chromium based ones on Android. Firefox does well on Linux desktops, but that speed.... It's missing here. Hope you'll get it done.”

    • Bug 1879387 -  Fenix fails to gracefully handle network transition during pageload

    • Bug 1906726 - Fenix pageload progress bar can take multiple times longer to complete than Chrome (e.g. 50 seconds vs 10 seconds)

    • Bug 1906323 -  Fenix silently fails to load links without WAN connectivity, with no error notification

    • Bug 1906733 - Very long resource loads for https://www.girince.com

    • 20-30% slower than Chrome in vismets, 3mpbs, +100ms rtt (small sampling of sites): results

• Active Projects roadmap Fx128 rel July 09, Fx 129 soft freeze July 04 

  • Native HTTPS RR Implementation (riding the trains, target Fx129)

    • Shipped

  • 3rd Party Cookie Deprecation (in progress, target Fx133)

    • Phase 1 landed in Fx129

    • Phase 2 discussions with a review of incoming telemetry in 2 weeks

  • Fetch keepalive: (in progress, Target Fx130) 

    • [Sunil] Milestone 1 completed

    • [Sunil] Milestone 2 work underway, 1871382, 1871383,

• HTTP Priorities Experiments (in progress, target Fx130 )

  • Experiment live

    • Priority header already shipped in Fx126

      • https://bugzilla.mozilla.org/show_bug.cgi?id=1865394

    • Fx 128 includes:

      • Implement Extensible Prioritization Scheme for HTTP/2

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1865040

      • Incremental flag on images, dom

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1784496

  • Browsertime pageload tests coming

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • [kershaw]

  • Run browsertime-upload tests on new test machines

  • Bug 1895357 Modify Taskcluster to run performance tests on the specific pool

    • r?

• Interop 2024 URL (in progress, target H2 2024)

  • On Hold:

  • https://bugzilla.mozilla.org/show_bug.cgi?id=1874119 (Rust URL)

  • https://bugzilla.mozilla.org/show_bug.cgi?id=1873973

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks, tracking such blocks using Bug 1901798  

  • [acreskey] Bug 1906664 -  Telemetry collection on android can add up to 30ms main thread time during pageload (parent process)

    • Considering disabling some existing probes

  • (sekim) Added a few more migration bugs for planning

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • [kershaw]

    • Improve test coverage by turning on HTTP/3 tests on Android

  • [mxinden]

    • Summary of relevant OS APIs per Firefox tier 1 build target.

    • Glean probes

      • Propose the addition of two new Glean probes

      • https://bugzilla.mozilla.org/show_bug.cgi?id=1906853

      • http3_udp_datagram_size_send

      • http3_udp_datagram_size_received

      • This will enable us to measure the impact of:

• Implementation of Packetization Layer Path MTU Discovery for Datagram Transports (RFC 8899) in Neqo

• Fast UDP for Firefox

• Milestone 1

  • Blocked on false-positive cargo vet failure.

    • (Will follow-up tomorrow.)

  • Fix Windows 64bit support using u64 file handles

  • Manual testing on Windows 64bit VM and MacOS (M2 via Rosetta)

    • Seeing up to 10 datagrams read in single GRO read.

• Milestone 2

  • Additional testing on ECN blackhole path https://github.com/mozilla/neqo/pull/1964

• Performance work: 

  • [acreskey] Bug 1889771 - Increase network.http.max-connections on Android nightly (currently 40 (900 on desktop))

    • Landing telemetry on socket pool exhaustion: (data review) https://bugzilla.mozilla.org/show_bug.cgi?id=1901855

    • No improvement measured so far on local browsertime tests (high latency, low bandwidth)

    • CI picked up loadtime improvements but vismet regressions, yet to analyze

      • https://bugzilla.mozilla.org/show_bug.cgi?id=1904643

  • New experiments:

    • OMT Decompression + html5Parser, cssLoader 

      • Experiment Brief

      • Data ticket

      • [Sunil] Experiment is live

Meeting 2024-07-02

• Team calendar

  • Next on Necko triage: Kershaw

• New contributors / Friends of Necko THANK YOU!!

  • Andrea Marchesini [:baku] 

    • Bug: 1904557 Unify and improve cookie warning messages 

  • Julian Descottes [:jdescottes]

    • Bug: 1900375 Expose an interface to provide a fake response object 

  • Thomas Wisniewski [:twisniewski]

    • Bug: 697151 Synchronous XMLHttpRequest (XHR) does not block readyState events for async XHR

    • Bug: 1848036 Add an ability to determine if it's safe to call an nsIRequest's suspend method 

  • Byron Campen [:bwc]

    • Bug: 1903905 Intermittent Assertion failure: stream != (0xFFFF), at /builds/worker/checkouts/gecko/netwerk/sctp/datachannel/DataChannel.cpp:2472 

• Round Table

  • No public topics this week

• Active Projects roadmap Fx128 rel July 09, Fx 129 soft freeze July 04 

  • Native HTTPS RR Implementation (in progress, target Fx129)

    • [kershaw] another potential blocker:

      • Bug 1895741 Intermittent valgrind-test | 28 bytes in 1 blocks are definitely lost at malloc / __res_context_send / __res_context_query / __res_context_querydomain

        • [kerahaw]

          • r?

  • 3rd Party Cookie Deprecation (in progress, target Fx132)

    • [Ed] Move to 133 TBD

  • Fetch keepalive: (in progress, Target Fx130) 

    • 1871378 Addressed review comments, waiting on review. 

  • HTTP Priorities Experiments (in progress, target Fx130 )

    • [acreskey]

      • Experiment live

      • Priority header already shipped in Fx126

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1865394

      • Fx 128 includes:

        • Implement Extensible Prioritization Scheme for HTTP/2

          • https://bugzilla.mozilla.org/show_bug.cgi?id=1865040

        • Incremental flag on images, dom

          • https://bugzilla.mozilla.org/show_bug.cgi?id=1784496

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • [kershaw]

    • Working on taskcluster

• Interop 2024 URL (in progress, target H2 2024)

  • Fixing Bug 1904582 

• necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks, tracking such blocks using Bug 1901798  

  • Updated dashboard: https://sql.telemetry.mozilla.org/dashboard/channel-success-error?p_channel=nightly&p_days=10&p_include_200_ok=false

• Fast UDP I/O in Firefox (in progress, target H2 2024)

  • Milestone 1 - Integrate quinn_udp into Necko (single-packet)

    • Blocked on false-positive cargo vet failure.

    • Needs testing on Windows. Setting up a virtual machine next.

    • Needs testing on MacOS. Will get Mac Book from servicedesk

    • Needs a decision whether fallback to NSPR is required.

  • Milestone 2 - Hookup ECN

    • Need to investigate behavior on path dropping all ECN marked packets.

• Performance work: 

  • [acreskey] Bug 1889771 - Increase network.http.max-connections on Android nightly (currently 40 (900 on desktop))

    • Dashboards convert impact via backout

      • Fenix pulse

      • Landing telemetry on socket pool exhaustion: https://bugzilla.mozilla.org/show_bug.cgi?id=1901855

      • Then intention is to land increase in Nightly, ride the trains

  • New experiments:

    • OMT Decompression + html5Parser, cssLoader 

      • [Sunil] Experiment is live

      • Experiment Brief

      • Data ticket

  • [jesup] overhead per-request (See https://share.firefox.dev/4bkNrqS)

Meeting 2024-06-25

• Team calendar

  • Next on Necko triage: Kershaw

• New contributors / Friends of Necko THANK YOU!!

  • amarchesini@mozilla.com

    • Bug: 1903110 cookie with empty name and value should be rejected 

    • Bug: 1768656 Deprecate and remove thirdparty.*sessionOnly prefs 

    • Bug: 1903415 Long cookie attrbiutes should be rejected 

  • ckerschbaumer@mozilla.com

    • Bug: 1883954 Consider moving HTTPS upgrade telemetry code into one place 

  • jdescottes@mozilla.com

    • Bug: 1903212 Expose a nsIDataChannel interface to JS 

  • mail@max-inden.de

    • Bug: 1903621 allow getsock* syscall in Rust binary 

  • twisniewski@mozilla.com 

    • Bug: 1796785 Addition of response status code to Resource Timing API 

    • Bug: 1842970 Pass failing xhr WPTs related to aborting XMLHttpRequests on workers

    •  Bug: 697151 Synchronous XMLHttpRequest (XHR) does not block readyState events for async XHR 

• Round Table

  • – no public topics this week  –

• Active Projects roadmap Fx128 rel July 09, Fx 129 soft freeze July 04 

  • Native HTTPS RR Implementation (in progress, target Fx129)

    • [kershaw] another potential blocker:

      • Bug 1895741 Intermittent valgrind-test | 28 bytes in 1 blocks are definitely lost at malloc / __res_context_send / __res_context_query / __res_context_querydomain

        • Need review

  • 3rd Party Cookie Deprecation (in progress, target Fx132)

    • [Ed] CHIPS partition limit: Reviews are coming back, making necessary fixes

    • Working to address review comments, may move to 133 TBD

  • Fetch keepalive: (in progress, Target Fx130) 

    • (greg) Reminder to ping Jason Kratzer once ms 1 has landed and we have added the fuzzing pref

      • [Sunil] Will do

    • [Sunil] Bug 1871378 - Incorporating review comments. MS1 can be marked done after merging this. We pass all WPTs now 

    • [Sunil] Will start work on MS2 next week. 

  • HTTP Priorities Experiments (in progress, target Fx130 )

    • [acreskey]

      • Bringing to DS Wednesday, Beta 128 (start as soon as approved)

      • https://mozilla-hub.atlassian.net/wiki/spaces/FIREFOX/pages/11047744/Experiments+Necko

      • Will run broad local tests

      • priorityHeader currently riding the trains; probably want to stop it

  • Firefox/Chrome benchmark test suite (in progress, Fx132)

    • (greg) Can we use this environment to verify Fast UDP I/O(GSO)?

    • – enter updates –

    • [kershaw]

      • Test machines ready

      • Start working

  • Interop 2024 URL (in progress, target H2 2024)

    • (sekim) Landed %2e entered in URL bar not normalized leading to denormalized request, this bug may also fix Bug 1902331.

    • [Ed] Added some tests for adding remote settings defaultURI bypass list, made some fixes as well. 

  • necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks, tracking such blocks using Bug 1901798  

    • [acreskey] Continue to hold on Android until further notice

    • (sekim) Dashboard for tracking networking_http_channel_disposition and networking_http_channel_onstart_status

    • (sekim) Landed Migrate HTTP_RESPONSE_STATUS_CODE probe to glean

    • (sekim) Added HTTP_RESPONSE_STATUS_CODE to the above dashboard.

  • Fast UDP I/O in Firefox (in progress, target H2 2024)

    • Congrats Max Inden for getting first project approved!

    • Allow getsock* syscalls in Rust code unblocking milestone 1.

    • Do `cargo vet` `safe-to-deploy` _audit_ for `quinn-udp`. Needed to import `quinn-udp` into `neqo_glue`'s dependency tree.

    • Propose patch to remove `quinn-udp`'s dependency on `tracing-attributes`. One less dependency to audit for milestone 1.

    • Debug false positive CI check, requiring `safe-to-deploy` audit for `tokio` crate. See Phabricator comment.

    • Tested https://github.com/cloudflare/quiche/pull/1764 as an alternative to `quinn-udp`. We will proceed with `quinn-udp` given that it supports all the Firefox tier 1 platforms.

  • Performance work: 

    • [acreskey] Bug 1889771 - Increase network.http.max-connections on Android nightly (currently 40 (900 on desktop))

      • Via Performance Pulse dashboards, saw improvements of 5-20%, P50 onwards in probes where you could expect changes:

        • http channel page open to first_sent

        • http channel sub item open to first sent

        • transaction wait time

      • Backed out to double-check impact.

      • Proposal:

        • Land os-limits telemetry

        • Increase in nightly, 128 sockets (beta & earlier)

        • Uplift to beta to observe impact of single change

        • Consider Fenix release population experiment

    • New experiments:

      • Performance Calibration (A/A/A) on Fenix, live (release channel might be affected, some probes)

        • [acreskey] Completed and summarized:

          • Fenix Performance Calibration results

          • Top-level metrics show less variance than desktop, despite much smaller population sizes

      • OMT Decompression + html5Parser, cssLoader 

        • Experiment Brief

        • Data ticket

Meeting 2024-06-18

• Team calendar

  • Next on Necko triage: Randell

• New contributors / Friends of Necko THANK YOU!!

  • bugmail@asutherland.org 

    • Bug: 1900681 XMLHttpRequestMainThread::DisconnectFromOwner should not call Abort for `mForWorker` XHRs because it double-clocks the state machine 

  • fwang@igalia.com 

    • Bug: 1881040 Use the destination to refine the fetchpriority mapping of FetchDriver::HttpFetch 

    • Bug: 1900764 urgency of fonts in fetchpriority-urgency.h2.html are not affected by fetchpriority 

  • ctuns@mozilla.com

    • Bug: 1902560 Perma [tier 2] /fetch/fetch-later/new-window.tentative.https.window.html | A same-origin window[target=''][features='0'] can trigger fetchLater. - Test timed out 

• Round Table

  • (greg) Standards positions:

    • Bug: 1475599 Implement Cookie-Store API 

      • Web compatibility with Chrome and Safari

        • Safari recent, web compatibility risk is higher

    • Request standards position review?

    • Randell ping Simon in channel 

      • Mondays standards meeting

    • Zstd - standards position update? 

      • We’re positive already: https://github.com/mozilla/standards-positions/issues/775

      • The dashboard shows an old position, and needs to be updated

      • Will bring to standards meeting monday

• Active Projects roadmap Fx128 rel July 09, Fx 129 soft freeze July 04 

  • Native HTTPS RR Implementation (in progress, target Fx129)

    • Performance regression identified in Nightly https://bugzilla.mozilla.org/show_bug.cgi?id=1895908#c15 

    • [kershaw]

      • Bug 1898191 Improving HTTPS RR Handling to Mitigate Performance Regression

        • https://glam.telemetry.mozilla.org/fenix/probe/networking_http_channel_page_open_to_first_sent/explore?aggType=avg&currentPage=1&ref=2024022904&timeHorizon=ALL&visiblePercentiles=%5B99%2C95%2C75%2C50%2C25%2C5%5D 

        • (greg) Looks like we have fixed performance regression - confirmed

          • [Andrew] Consider future optimizations for DoH - HTTPS RR

        • [acreskey] FYI, full query

          • Cached performance pulse

          • Do we know perf impact on Page_open_to_first_sent with DoH?

            • Smaller scale desktop regression? dashboard

        • [kershaw] another potential blocker:

          • Bug 1895741 Intermittent valgrind-test | 28 bytes in 1 blocks are definitely lost at malloc / __res_context_send / __res_context_query / __res_context_querydomain

  • 3rd Party Cookie Deprecation (in progress, target Fx132)

    • [Ed] Received some feedback on patches, addressing this week

      • https://phabricator.services.mozilla.com/D201757

      • https://phabricator.services.mozilla.com/D203521

  • Fetch keepalive: (in progress, Target Fx129) 

    • [Sunil] Merged 1871376  and 1871377. Keepalive WPTs now pass except for resource limits tests.

    • [Sunil] Working on patch for Bug 1871378 

    • [Sunil] Won't be able to land in Fx129.

  • HTTP Priorities Experiments (in progress, target Fx130 )

    • Valentin to share this week 

  • Firefox/Chrome benchmark test suite (in progress, Fx132)

    • [kershaw]

      • Test machines should be ready soon

  • Interop 2024 URL (in progress, target H2 2024)

    • (sekim): Incorrect parsing: file:c:\\foo\\bar.html against file:///tmp/mock/path landed.

  • necko-glean: (sekim) Landing glean patches using #ifndef ANDROID blocks, tracking such blocks using Bug 1901798  

    • [acreskey] Please continue to #ifndef non-essential probes on Android, I am re-raising this

  • Fast UDP I/O in Firefox

    • Milestone 1 - quinn-udp in Firefox

      • Ready for review

      • https://phabricator.services.mozilla.com/D212959

      • Looking into telemetry probes to add.

      • Discussion on static check disallowing Rust network calls.

    • Milestone 2 - sending and receiving ECN

      • Blocked on https://github.com/mozilla/neqo/issues/1925

  • Performance work: 

    • [acreskey] Landed increase socket pool of 40 to 100, android

      • Bug 1889771 - Increase network.http.max-connections on Android nightly (currently 40 (900 on desktop))

      • Meta: https://bugzilla.mozilla.org/show_bug.cgi?id=1903115

        • Probes in https://bugzilla.mozilla.org/show_bug.cgi?id=1901855

    • New experiments:

      • Performance Calibration (A/A/A) on Fenix, live (release channel might be affected, some probes)

        • Experiments completed; need changes to Performance Telemetry Reports to compare glean timing distributions

      • OMT Decompression + html5Parser, cssLoader 

        • Experiment Brief

        • Data ticket

        • Potential conflict with HTTP Priorities, push OMT first

Meeting 2024-06-11

• Team calendar

  • Next on Necko triage: Valentin

• New contributors / Friends of Necko THANK YOU!!

  • nika@thelayzells.com fixed Bug: 1718673 text/event-stream mime type starts downloading instead of streaming content 

  • fwang@igalia.com fixed 

    • Bug: 1882084 Support fetchpriority in Early Hints preloader

    • Bug: 1892734 Add tests for checking the priority set in our HTTP requests 

    • Bug: 1901220 Perma TEST-UNEXPECTED-FAIL | /_mozilla/fetch/fetchpriority/fetchpriority-urgency.h2.html | Adjustment of urgency parameter for <...> when Gecko 128 merges to beta on 2024-06-10 

  • jschanck@mozilla.com fixed Bug: 1892528 Enable Xyber768 in Http/3 TLS handshakes under a pref 

  • bugmail@asutherland.org fixed Bug: 1900681 XMLHttpRequestMainThread::DisconnectFromOwner should not call Abort for `mForWorker` XHRs because it double-clocks the state machine 

• Round Table

  • Potential performance opportunity? https://www.bleepingcomputer.com/news/google/google-chrome-reduced-cookie-requests-to-improve-performance/ 

    • Relaying same cookie that has been broadcast multiple times, we could be duping sends - Ed to review and consider 

  • Standards positions:

    • Bug: 1475599 Implement Cookie-Store API 

      • Web compatibility with Chrome and Safari

        • Safari recent, web compatibility risk is higher

    • Request standards position review?

    • Randell ping Simon in channel 

    • Zstd - standards position update? Randell to ping Simon

• Active Projects roadmap Fx127 rel June 11, Fx128 rel July 09, Fx 129 soft freeze July 04 

  • Native HTTPS RR Implementation (in progress, target Fx129)

    • Performance regression identified in Nightly https://bugzilla.mozilla.org/show_bug.cgi?id=1895908#c15 

    • [kershaw]

      • Bug 1898191 Improving HTTPS RR Handling to Mitigate Performance Regression

        • https://glam.telemetry.mozilla.org/fenix/probe/networking_http_channel_page_open_to_first_sent/explore?aggType=avg&currentPage=1&ref=2024022904&timeHorizon=ALL&visiblePercentiles=%5B99%2C95%2C75%2C50%2C25%2C5%5D 

  • 3rd Party Cookie Deprecation (in progress, target Fx132)

    • [Ed] Added tests and put into review, two patches now in review:

      • https://phabricator.services.mozilla.com/D201757

      • https://phabricator.services.mozilla.com/D203521

  • Fetch keepalive: (in progress, Target Fx129) 

    • (greg) Jason Kratzer I see that this is behind a pref. Is this ready for fuzz testing? Not yet, landing MS1 next week. Could be tested after next week we have all WPT passing

      • [Sunil] Can test after 

    • Landing in progress for 1871377, 1871376 

    • Started working on 1871378.

• HTTP Priorities Experiments (in progress, target Fx130 )

  • (valentin) Working on experiment brief. Will sync up with Andrew to launch experiment.

• Firefox/Chrome benchmark test suite (in progress, Fx132)

  • – enter updates – 

• Interop 2024 URL (in progress, target H2 2024)

  • (sekim) Patch for Bug 1874118 - Fix handling of trailing characters in opaque paths in URL

  • (sekim) Currently debugging a Rust URL bug 

  • (sekim) Few other patches are waiting for review.

  • (sekim) This Patch works except for Windows, Testing URL on Windows?  

• necko-glean: (sekim) due to incidents, currently on hold.

  • – enter updates – 

• Fast UDP I/O in Firefox

  • https://bugzilla.mozilla.org/show_bug.cgi?id=1852924 

  • [mxinden] Fast UDP for Firefox Project Plan - Fast UDP for Firefox

    • Milestone 1

      • https://bugzilla.mozilla.org/show_bug.cgi?id=1901295

      • QUIC UDP I/O using quinn-udp

    • Milestone 2

      • Neqo sending and receiving ECN markings

    • Milestone 4

      • Neqo receiving multi-packet via GRO

    • Each still needs a lot of work, especially across platforms.

• Performance work: 

  • Glean on Android update

    • [acreskey] Please #ifndef Android new glean probes unless specifically needed, until glean performance is improved 

  • [acreskey] FYI: 

    • Fenix TLS connection times appear to have increased around March 19, 2024

    • https://bugzilla.mozilla.org/show_bug.cgi?id=1901604

  • New experiments:

    • Performance Calibration (A/A/A) on Fenix, live (release channel might be affected, some probes)

      • Finishing today

      • https://experimenter.services.mozilla.com/nimbus/fenix-performance-metric-calibration/edit/overview

    • OMT Decompression + html5Parser, cssLoader 

      • Experiment Brief

      • Data team on work-week. Will attend data office hours next week.

Meeting 2024-06-04

• Team calendar

  • Next on Necko triage: Sunil

• New contributors / Friends of Necko THANK YOU!!

  • Andrew McCreight Bug: 1899863 Extremely frequent PNecko::Msg_CancelHTMLDNSPrefetch log spam 

  • Julian Descottes

    • Bug: 1899291 Improve channelId generation code to avoid clashes between tabs with identical pID [necko-priority-next]

    • Bug: 1899563 Perform SW upgrade before http-on-before-connect  [necko-priority-next]

  •  Manuel Bucher

    • Bug: 1897050 Remove or update metrics expiring in Firefox 129: networking.proxy_info_type before Gecko version gets increased to 129 on 2024-06-10

    • Bug: 1454850 Mislabelled options relating to SOCKS4, SOCKS5 and server-side DNS resolution  

• Round Table

  • No public topics this week

• Active Projects roadmap Fx127 rel June 11, Fx128 soft freeze June 06

  • Native HTTPS RR Implementation (in progress, target Fx129)

    • Performance regression identified in Nightly https://bugzilla.mozilla.org/show_bug.cgi?id=1895908#c15 

    • [kershaw]

      • Bug 1898191 Improving HTTPS RR Handling to Mitigate Performance Regression

        • Monitor changes

  • 3rd Party Cookie Deprecation (in progress, target Fx129)

    • [Ed] Added telemetry and submitted for data approval

    • Just need to add a few more tests

    • Note: CHIPS will be enabled soon

      • Let’s keep an eye on cookie same-site/cross-origin related bugs

  • Fetch keepalive: (in progress, Target Fx129) 

    • (greg) Jason Kratzer I see that this is behind a pref. Is this ready for fuzz testing? Not yet, landing MS1 next week. Could be tested after next week we have all WPT passing

    • Service worker support missing in WPT. 

    • – enter updates –

  • HTTP Priorities Experiments (in progress, target Fx129 )

    • (valentin) 1900555 - Add fetchpriority prefs to the Feature Manifest planning to land before code freeze

    • Frederic from Igalia has been landing some tests.

    • Working on experiment brief.

  • Firefox/Chrome benchmark test suite (in progress, Fx132)

    • [kershaw]

      • Test machines will be ready by the end of this week

  • Interop 2024 URL (in progress, target H2 2024)

    • (sekim) A patch backed out due to regressions: https://phabricator.services.mozilla.com/D211898

    • (sekim) Two patches waiting for review/wip, will be complete soon.

      • https://phabricator.services.mozilla.com/D210800

      • https://phabricator.services.mozilla.com/D212331

    • (sekim) Although not part of Interop 2024 URL, fixed a related bug (can be landed after windows fix with ‘/’): https://phabricator.services.mozilla.com/D212193

  • necko-glean: (sekim) due to incidents, currently on hold.

  • Performance work: 

    • Glean on Android update

      • Probes disabled in Release: https://pastebin.mozilla.org/eJEZEmLo

      • Batched DB writes issues fixed in Nightly, being uplifted: 

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1892230

    • Unrelated: Glam for Fenix stuck at April 13 updates:

      • https://github.com/mozilla/glam/issues/2843

    • New experiments:

      • Performance Calibration (A/A/A) on Fenix, live (release channel might be affected, some probes)

        • https://experimenter.services.mozilla.com/nimbus/fenix-performance-metric-calibration/edit/overview

      • OMT Decompression + html5Parser, cssLoader 

        • Experiment Brief

    • FYI, updated impact of Cookie Banner Handling documented here. Android results coming.

      • Cookie Banner Blocking Performance Results

Meeting 2024-05-28

• Team calendar

  • Next on Necko triage: Ed

• New contributors / Friends of Necko THANK YOU!!

  • Bug: 1741375 Proxy DNS by default when using SOCKS v5 (manuel@mozilla.com)
  • Bug: 1895319 refactor http3server to use neqo-bin for I/O, timers, event-loop etc. ( mail@max-inden.de)
  • Bug: 1896475 Implement Request/Response .bytes() method (evilpies@gmail.com)
  • Bug: 1897959 Perma /client-hints/sec-ch-width.https.html | Sec-CH-Width should be set - assert_equals: expected 100 but got 1 (aborovova@mozilla.com)
  • Bug: 1898129 Perma [tier 2] /fetch/api/abort/serviceworker-intercepted.https.html | response.bytes() rejects if already aborted - promise_test: Unhandled rejection with value: object "TypeError: response[bodyMethod] is not a function" (aborovova@mozilla.com)

• Round Table

  • Revive 1676432 - Add a UDP DNS resolver  ?

    • Chrome has their own native resolver that uses TCP/UDP

      • https://source.chromium.org/chromium/chromium/src/+/main:net/dns/dns_transaction.cc;l=1357-1371;drc=c0265133106c7647e90f9aaa4377d28190b1a6a9;bpv=1;bpt=1 

    • Could help bypass perf issues

    • Add to performance opportunities, experiment WIP patches 

      • Current patch for not performing IPV6 when not available, if this does not resolve issue we will continue to explore this.

      • 1122907 - Slow DNS lookupconnection timings on 64 bit Linux. 

  • Bug: 1897050 Remove or update metrics expiring in Firefox 129: networking.proxy_info_type before Gecko version gets increased to 129 on 2024-06-10 

    • Manuel landed a patch.

  • Necko Bug Bash Day 

    • Total Bugs updated: 197(7 off team)

      • Closed: 41 (3 off team)!

        • 14 fixed

        • 6 invalid

        • 11 won’t fix

        • 8 works for me

        • 1 duplicate

        • 4 incomplete

        • By Severity:

          • S3: 24

          • S4: 6

          • – : 6

          • NA: 5

      • Priority/severity updated: 16

      • Blocks updated: 28

      • Necko Priority Queue: 3 added 

  • (Sunil) Simon Peter will present Contributing to Standards Talk on 2nd August in Let’s Decipher Necko Session.

• Active Projects roadmap Fx127 rel June 11, Fx128 soft freeze June 06

  • Native HTTPS RR Implementation (in progress, target Fx128)

    • Sent Intent To Ship: Resolving HTTPS records using platform DNS APIs 

    • Performance regression identified in Nightly https://bugzilla.mozilla.org/show_bug.cgi?id=1895908#c15 

    • [kershaw]

      • Bug 1898191 Improving HTTPS RR Handling to Mitigate Performance Regression

        • Patches r+ed.

  • 3rd Party Cookie Deprecation (in progress, target Fx128)

    • [Ed] Writing telemetry for partition limit purging dry run

    • [Ed] A few reviews for the privacy team

  • Fetch keepalive: (in progress, Target Fx129) 

    • [Sunil] 2 out of 3 patch stacks r+. Incorporating review comments on the final patch.

    • [Sunil] Need to complete https://bugzilla.mozilla.org/show_bug.cgi?id=1871378 for Milestone 1 (Implement keepalive as per Fetch standard)

• HTTP Priorities Experiments (in progress, target Fx 129 )

  • 1865040 - Implement Extensible Prioritization Scheme for HTTP2 landed

  • 1864392 - Honor `fetchpriority` attribute for HTTP3 requests was backed out for a test failure. Will land again.

• Firefox/Chrome benchmark test suite (in progress, Fx 129)

  • [kershaw]

    • Configuring the test environment in VM

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] No progress on DefaultURI bypass

  • [Sean] Landed null origin blob 

• Performance work: 

  • New experiments:

    • Performance Calibration (A/A/A) on Fenix, live

      • Acreskey: fyi, possibility of telemetry being disabled on Android

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1898810

          • https://bugzilla.mozilla.org/show_bug.cgi?id=1899169

      • https://experimenter.services.mozilla.com/nimbus/fenix-performance-metric-calibration/edit/overview

    • OMT Decompression + html5Parser, cssLoader 

      • [Sunil] https://bugzilla.mozilla.org/show_bug.cgi?id=1899233

      • Started writing experiment brief

Meeting 2024-05-21

• Team calendar

  • Next on Necko triage: Kershaw

• New contributors / Friends of Necko

  • freddy fixed 1897132 - make netwerktestbrowserbrowser_dns_prefetch_link_header.js work with https-first 

  • mconley fixed 1896929 - Allow nsInputStreamChannel to have an arbitrary content type set on it 

  • cpeterson fixed 1894429 - Spoof Android 10 OS version in Firefox Android’s User-Agent string on Android versions = 10

• Round Table

  • – enter topics –

• Active Projects roadmap Fx127 rel June 11, Fx128 soft freeze June 06

  • Native HTTPS RR Implementation (in progress, target Fx128)

    • Sent Intent To Ship: Resolving HTTPS records using platform DNS APIs 

    • Performance regression identified in Nightly https://bugzilla.mozilla.org/show_bug.cgi?id=1895908#c15 

  • 3rd Party Cookie Deprecation (in progress, target Fx128)

    • [Ed] Started implementation of telemetry for chips partition limit, still in progress 

  • Fetch keepalive: (in progress, Target Fx129) 

    • [Sunil] Follow-up bug for enabling PFetch for all main thread Fetch requests

    • [Sunil] Addressing final comments from Eden, mostly will land the patch this week for Milestone 1.

• HTTP Priorities Experiments (in progress, target Fx 128 )

  • 1784496 - The necko consumers should use the HTTP3 priority's incremental flag landed last week

  • 1865040 - Implement Extensible Prioritization Scheme for HTTP2 on autoland

  • 1864392 - Honor `fetchpriority` attribute for HTTP3 requests planned to land for later this week

• Firefox/Chrome benchmark test suite (in progress, target H1 2024)

  • [kershaw]

    • Learning taskcluster, puppet, …

    • Start with a VM nearly identically as the test hardware

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] Discussed default URI bypass list blocker, startup URLs with valentin: startup URLs shouldn’t need access to bypass list so we should be fine to continue

• Performance work: 

  • New experiments:

    • Performance Calibration (A/A/A) on Fenix, live

      • https://experimenter.services.mozilla.com/nimbus/fenix-performance-metric-calibration/edit/overview

    • OMT Decompression + html5Parser, cssLoader 

      • [Sunil] Started the experiment preparation. Finished Experimenter Onboarding Training and overview of Nimbus. 

      • Will start writing experiment brief this week

Meeting 2024-05-14

• Team calendar

  • Next on Necko triage: Valentin

• New contributors / Friends of Necko

  • None in the last week

• Round Table

  • (sunil) Enable redirects for auth retries in release

    • Start rollout for using Release Experiments

    • Do we need more telemetry probes to identify regression sooner?

      • Probes for channel status exist

        • Success rate of channel monitor changes in trend

        • How can we see exact regressions introduced?

          • If regression we will never resolve 200 after 401

          • [acreskey]: Feature Manifest

          • https://searchfox.org/mozilla-central/source/toolkit/components/nimbus/FeatureManifest.yaml

          • Leverage a gradual rollout

            • Can we target users with specific auth?

            • Ask data folks for guidance on % of population

  • (sunil) Currently we use PFetch for workers and keepalive. Proposal to use PFetch for all Fetch requests. 

    • Reduces main thread contention

    • HTTPChannelParent/Child, logging and cookies dependencies 

    • Needs investigation/scoping, needs decision before we implement Fetch Streaming Upload

• Active Projects roadmap Fx126 rel May 14, Fx127 soft freeze May 09

  • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

    • OMT OnStart/OnStop Request (content) threading/api model

    • new Project Plan

    • Split HTML5Parser OnStop - OMT Shipped in Fx122 

    • Split CSSLoader OnStop - OMT riding the trains in Fx126

    • OnStart consumer

    • – enter updates –

  • Optimize, Measure and add tests for Speculative Connections (riding the trains, target: Fx126) 

    • – enter updates –

  • zstd content-encoding: (riding the trains, Fx126)

    • – enter updates –

  • Native HTTPS RR Implementation (in progress, target Fx127)

    • Riding the trains in 127

    • Sent Intent To Ship: Resolving HTTPS records using platform DNS APIs 

  • 3rd Party Cookie Deprecation (in progress, target Fx128)

    • [Ed] Wrote test for current implementation so far.

    • [Ed] Next steps: add test for insecure purging first

  • Fetch keepalive: (in progress, Target Fx129) 

    • [Sunil] Addressing review comments from Eden. 

    • We pass all the WPTs for the implemented feature.

    • Plan to merge the changes this week.

• HTTP Priorities Experiments (in progress, target Fx 128 )

  • Will start landing patches this week.

• Firefox/Chrome benchmark test suite (in progress, target H1 2024)

  • Created a JIRA issue to request test machines.

• Interop 2024 URL (in progress, target H2 2024)

  • [Ed] Continuing work on default-uri bypass list

• Performance work: 

  • New experiments:

    • Performance Calibration (A/A/A) on Fenix, planned to start today, nightly, beta, release

      • https://experimenter.services.mozilla.com/nimbus/fenix-performance-metric-calibration/edit/overview

    • OMT Decompression + html5Parser, cssLoader (Thank you Sunil)

  • [acreskey] 

    • Investigating: First visual change and reported time to first byte 1 second slower on Fenix compared to Chrome, en.m.wikipedia.org, high latency network

      • https://bugzilla.mozilla.org/show_bug.cgi?id=1895908

  • FYI: 

    • Predictor christmas tree:

      • https://bugzilla.mozilla.org/show_bug.cgi?id=1890559#c4

      • (greg) Keep enabled?

        • Continue investigating and assess performance priority

  • [jesup] Working on a next-stage OMT proposal

• [jesup] WebSocketStream - Chrome is pushing it.  Mt is very slightly positive if we remove some random cruft from the PR. Kagami and I are skeptical; we could do it; it wouldn’t be hard - but the big question would be “why not WebTransport?”  (especially once WebTransport over H2 is available globally).

Meeting 2024-04-30

• Team calendar

  • Next on Necko triage: Sunil

• New contributors / Friends of Necko

  • Thomas Wisniewski fixed bugs 1889075, 1893257, 1893219

  • Robert Longson fixed 1892748

• Round Table

  • No public topics this week

• Active Projects roadmap 

  • Active Projects: Fx126 rel May 14, Fx127 soft freeze May 09

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

      • Split CSSLoader OnStop - OMT riding the trains in Fx126

        • – enter updates –

      • OnStart consumer

        • No response from other team on ImageLoader 

      • [acreskey] Will re-run this experiment, adding in OMT compression variants (up for volunteers as well)

      • Next:

        • Review other OnStop consumers largest opportunity wins

        • Write a short doc on strategy, current/future states and opportunities identified 

    • Optimize, Measure and add tests for Speculative Connections (riding the trains, target: Fx126) 

      • [acreskey] Will re-run local tests to see if we budge LCP, FCP

    • Native HTTPS RR Implementation (in progress, target Fx127)

      • New regressions: 

        • 1891423 - Hosts file no longer blocks some sites since native HTTPS resolver was enabled on Ubuntu

          • Should we check /etc/hosts before doing a HTTPS lookup?

        • Bug: 1893944 Resolving any .local domain takes at least 10 seconds 

          • [kershaw]

            • Patch in review

      • – enter updates –

    • Interop 2024 URL (in progress, target H2 2024)

      • – enter updates –

    • 3rd Party Cookie Deprecation (in progress, target Apr 9)

      • – enter updates – 

    • zstd content-encoding: (riding the trains, Fx126)

      • – enter updates –

      • [Dylan] 

        • Landed devtools https://bugzilla.mozilla.org/show_bug.cgi?id=1891610 

      • [jesup] Riding to release

      • Next

    • Fetch keepalive: (in progress, Target Fx128) 

      • – enter updates –

    • Performance work: 

      • [acreskey] Setting up Rasberry Pi throttling host to investigate Fenix connection time issues

      • [valentin] I found an issue with speculative connect having the wrong isolation key: 1894206 - Speculative connect for HTTPS upgraded page loads has the wrong originAttributes

Meeting 2024-04-16

• Team calendar

  • Next on Necko triage: Kershaw

• New contributors / Friends of Necko

  • New contributor Serfreeman1337 fixed 1810362 - Firefox does not seem to support IPv6 unless DNS-over-HTTPS is enabled 

  • Thomas Wisniewski fixed 1886757 - network.dns.localdomains does not apply to FQDN 

  • Chris Peterson fixed 1873273 - Freeze UA string and navigator.platform on 32-bit x86 Linux to match Chrome's frozen navigator.platform (Linux x86_64)

• Round Table

  • (valentin) Next ESR is 128 - code freeze date: June 6 - that means if we have big refactorings we should aim to get them done by then, to avoid conflicts when uplifting stuff later.

• Active Projects roadmap 

  • Active Projects: Fx125 rel Apr 16, Fx126 soft freeze Apr 11 

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

      • Split CSSLoader OnStop - OMT to track in Fx126

        • – enter updates –

      • OnStart consumer

        • – enter updates –

    • Optimize, Measure and add tests for Speculative Connections (in progress, target: Fx126) 

      • No updates

    • Native HTTPS RR Implementation (in progress, target Fx126)

      • New regression: 1891423 - Hosts file no longer blocks some sites since native HTTPS resolver was enabled on Ubuntu

        • Should we check /etc/hosts before doing a HTTPS lookup?

    • Interop 2024 URL (in progress, target H2 2024)

      • [Ed] Looking at how to use RemoteSettings to get our default bypass list into the program from the cloud

    • 3rd Party Cookie Deprecation (in progress, target Apr 9)

      • [Ed] Added priority purging of insecure cookies, also byte limit check on cookie overwrite

      • [Ed] Started writing the test for cookie partition limit purging 

    • zstd content-encoding: (in progress, Fx126)

      • [jesup] zstd landed and is in beta!  No known regressions

      • [jesup] Performance results from the OMT decompression patch were surprisingly good from GLAM telemetry

        • FCP/LCP/DOM Dontent Loaded/etc improved by 10% in many cases, 30% in some higher %iles

      • [Dylan] 

        • done devtools support

        • Writing zstd devtools test

    • Fetch keepalive: (in progress, Target Fx128) 

                                               [Sunil] Bug 1871377 , Bug 1871376

• Most of the functionality required for Milestone 1 is working and wpt verified.

• Will publish patches for review

Meeting 2024-04-09

• Team calendar

  • Next on Necko triage: Randell

• Round Table

  • none

• Active Projects roadmap 

  • Active Projects: Fx125 rel Apr 16, Fx126 soft freeze Apr 11 

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

      • Split CSSLoader OnStop - OMT to track in Fx126

        • Experiment ended

        • No obvious changes in the field

      • OnStart consumer

        • Still no response from tnikkel

        • Probably drop for now

    • Optimize, Measure and add tests for Speculative Connections (in progress, target: Fx126) 

      • [acreskey] Complete: landed in Fx 126,  summarized here:

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1816539#c9

    • Native HTTPS RR Implementation (in progress, target Fx126)

      • [valentin] Will ride the trains

      • Investigate increase in DNS queueing time with native HTTPS records was fixed

    • Interop 2024 URL (in progress, target H2 2024)

      • [Ed] Landed defaultURI bypass pref default URI bypass pref

      • [Ed] Starting remote settings to update pref  

      • [Dylan] landed https://bugzilla.mozilla.org/show_bug.cgi?id=1874117 

      • [Dylan] started https://bugzilla.mozilla.org/show_bug.cgi?id=1874119 

    • 3rd Party Cookie Deprecation (in progress, target Apr 9)

      • [Ed] Added FIFO (older) cookie purging at partition byte limit 

      • [Ed] Need to also add insecure cookie purging first (before older cookies)

    • zstd content-encoding: (in progress, Fx126)

      • [jesup] On autoland!

      • Anna Weine will be adding cert decompression with zstd

      • Dylan is looking at devtools

    • Fetch keepalive: (in progress, Target Fx128) 

      • – enter updates –

    • Performance work: 

      • Look for more OMT

Meeting 2024-04-02

• Team calendar

  • Next on Necko triage: Valentin

• Round Table

  • Necko Performance Opportunities select priorities for H1

    • Performance priorities selected for H1-2024

      • HTTP3 Upload Speed

      • Enable DNS pre-fetch for https

      • Bug: 1709175 Switch from select() to poll for MacOS to reduce CPU usage

      • HTTP/2, HTTP/3 priorities, urgencies, incremental flag(meta bug)

      • Optimize socket pool constants - Android

  • [acreskey] FYI:

    • Fenix performance gap to Chrome through throttling VPN reproduced

    • Live pageload comparisons

    • https://share.firefox.dev/3TVlXT3

      • Problems with tls, general request times

• Active Projects roadmap 

  • Active Projects: Fx125 rel Apr 16, Fx126 soft freeze Apr 11 

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

      • Split CSSLoader OnStop - OMT to track in Fx126

        • Experiment started 

      • OnStart consumer

        • – enter updates –

    • Optimize, Measure and add tests for Speculative Connections (in progress, target: Fx126) 

      • [acreskey] Update:

        • Beta experiment live, ended Mar 19, 2024 - no measurable improvements via telemetry

        • Local browsertime tests show significant improvements to sub-resource connections times (up to 50%) on some sites

        • (quick demo)

        • We don’t hit socket pool exhaustion on Desktop:

          • https://bugzilla.mozilla.org/show_bug.cgi?id=1819556
            We removed the telemetry for desktop socket exhaustion

            • NETWORK_SESSION_AT_900FD probes.       Telemetry shows all users are able to poll up to 1000 sockets - see glam

        • Landing patch to set Desktop to 20 for Fx126

    • Native HTTPS RR Implementation (in progress, target Fx126)

      • [valentin] Enabled on Nightly and Early beta 125

      • Investigating issues:

        • 1885529 - Investigate increase in DNS queueing time with native HTTPS records (Windows)

          • Telemetry patch ready.

          • I can repro locally on win11 (investigating)

        • 1885556 - 4.62 - 4.32% fandom ContentfulSpeedIndex  fandom loadtime + 1 more (Linux) regression

          • No progress yet

    • Interop 2024 URL (in progress, target H2 2024)

      • [Ed] In progress: unknown scheme default URI bypass list 

      • [Dylan]

        • Landed https://bugzilla.mozilla.org/show_bug.cgi?id=1879317 

        • Submitted patch in rust-url https://bugzilla.mozilla.org/show_bug.cgi?id=1874117 

    • 3rd Party Cookie Deprecation (in progress, target Apr 9)

      • [Ed] reviewed telemetry for guidance on order of potential purging when reaching cookie partition byte count limit (FIFO vs FILO)

        • Currently it’s not even clear on if this should be enforced at all

        • No existing telemetry to indicate how often a limit would be reached and potentially interfere with web compat

        • We can probably proceed with multiple purging implementations behind feature flags and then decide based on some telemetry that ships ahead of it.

        • Any concerns/differing opinions?

    • zstd content-encoding: (in progress, target H1 2024)

      • Code r+’d to land

        • xz library exploit doesn’t affect our import since we don’t build zstd binaries

        • Likely will update the library anyways since a new release with performance improvements, etc was just made to coincide with Chrome enabling zstd in release.

      • Works well with facebook

      • In contact with decoder about fuzzing

      • Remaining work with sec team to TLS cert comp and dev tools

    • Fetch keepalive: (in progress, Target Fx128) 

      • Testing the initial implementation

    • Performance work: 

      • [jesup] Results are in from the OMT decompression landing: https://treeherder.mozilla.org/perfherder/alerts?id=41924 

        • Some red. Lots and lots of big green through

        • Filtering to speedindex looks even better. 100% green for FCP and FirstVisualChange. LCP is good, though not all green.

        • Pushed tries for a patch that leaves things on MainThread if the compressed size is known and is small (512 and 1024 bytes). This could avoid overhead in cases where decompression is almost certainly very short. 

          • 512 shows improvement of 1-7% on 4 sites (of 30)

          • 1024 shows improvement of 1-3.5% on 5 sites, regression of 1-3% on 2 sites.

          • Planning to land with 512; includes a pref for size

Meeting 2024-03-26

• Team calendar

  • Next on Necko triage: Ed

• Round Table

  •  Bug 1420777 - Http/2 connection reuse to non-origin server for new hostname on DNS overlap was finally resolved, which fixes an annoying corner case for many developers. 

    • Backed out, Valentin will resolve and reland

  • https://bugzilla.mozilla.org/show_bug.cgi?id=1741375 

    • Telemetry landed - reminder set for 03/18

  • https://bugzilla.mozilla.org/show_bug.cgi?id=1887660 

  • https://bugzilla.mozilla.org/show_bug.cgi?id=1887714 

    • May need to work with IT to get to bottom of it

  • [acreskey] FYI: tooling to extract and compare PerfStats from local browsertime comparison runs:

  • https://github.com/acreskeyMoz/perfstats_from_browsertime/blob/main/README.md

  • NEL, can impl on our end and let CF know once ready and review telemetry

    • No need to schedule and coordinate

    • Send JSON to server if something goes wrong loading the page

    • If we find privacy leaks we will not send the data, anything we send is better than nothing.

    • Target in H2 2024, be nice to get it in

  • 103 Early Hints can be used to break adblocking (thread)

    • https://github.com/Mechazawa/103-early-anti-adblock 

    • I don't see a good outcome here.  Chrome lets sites exfiltrate data with prefetch; we allow sites to detect blocking of exfiltration with prefetch.  We're going to have to change how prefetch works to manage this, I think.

      • POC, Kershaw will review and file a bug

• Active Projects roadmap 

  • Active Projects: Fx125 rel Apr 16, Fx126 soft freeze Apr 11 

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

      • Split CSSLoader OnStop - OMT to track in Fx126

        • Experiment started 

      • OnStart consumer

        • – enter updates –

    • Optimize, Measure and add tests for Speculative Connections (in progress, target: Fx126) 

      • [acreskey]

        • Beta experiment live, ended Mar 19, 2024

        • No clear signal of improvements to top-level metrics for increasing # of speculative sockets and similarly no regression detected in top-level metrics for disabling rel=preconnect

        • Experiments page updated: https://mozilla-hub.atlassian.net/wiki/spaces/FIREFOX/pages/11047744/Experiments+Necko#Conclusions

        • Worth considering that significant changes in top-level metrics may take numerous improvements to be measurable in the wild.

          • See first_paint after response start, 2023-> current

        • Would like to try to find signal in a local browsertime run (timeboxed, 2 days)

    • Native HTTPS RR Implementation (in progress, target Fx126)

      • [valentin] Enabled on Nightly and Early beta 125

      • Investigating issues:

        • 1885529 - Investigate increase in DNS queueing time with native HTTPS records (Windows)

        • 1885556 - 4.62 - 4.32% fandom ContentfulSpeedIndex  fandom loadtime + 1 more (Linux) regression

        • – enter updates –

    • Interop 2024 URL (in progress, target H2 2024)

      • [Ed] Not much progress on my side

      • [Dylan] 

        • Submitted patch https://bugzilla.mozilla.org/show_bug.cgi?id=1873944 

        • Landed 

          • https://bugzilla.mozilla.org/show_bug.cgi?id=1873940 

          • https://bugzilla.mozilla.org/show_bug.cgi?id=1873906 

    • 3rd Party Cookie Deprecation (in progress, target Apr 9)

      • [Ed] Cookie capping: Agreed upon on putting capacity limit rejection behind a feature flag until we know if safe (telem or google fully ships) 

      • [Ed] Some code review

    • zstd content-encoding: (in progress, target H1 2024)

      • approved

    • Fetch keepalive: (in progress, Target Fx128) 

      • [Sunil] Added Keepalive attribute support under pref

      • [Sunil] Currently adding PFetch Support to keepalive. 

    • Performance work: [jesup]

      • Preliminary results for Bug 1356686 - Do decompression off main thread in content

      • Bounced last night (1 failure on android and a non-unified-build issue); will try again hopefully today

Meeting 2024-03-19

• Team calendar

  • Next on Necko triage: Ed

• Round Table

  • (greg) https://mozilla-necko.github.io/newsletter/2024/03/15/community.html 

    • Awesome job, Manuel Bucher thank you for coordinating and publishing!

    • Tracking community contributions? Are we seeing an increase?

      • Manuel effort to monitor

      • Could use whiteboard tag when we land to easily filter out

        • When review patch and land add tag to track 

      • Review bugs landed each week off necko 

        • Can we use bugbot to do this for us?

          • Sunil to investigate 

    • Improve coordination, regular cadence?

      • Schedule every 2 retro’s (12 wks)

        • Prepare google doc, review w team

        • [Greg] Reminder calendar item

• Active Projects roadmap 

  • Active Projects: Fx124 rel Mar 19, Fx125 soft freeze Mar 14 

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

      • Split CSSLoader OnStop - OMT to track in Fx126

        • – enter updates –

      • OnStart consumer

        • – enter updates –

    • Optimize, Measure and add tests for Speculative Connections (in progress, target: Fx125) 

      • Beta experiment live, ends Mar 19, 2024

    • Native HTTPS RR Implementation (in progress, target Fx126)

      • [valentin] Enabled on Nightly and Early beta 125

      • Investigating issues:

        • 1885529 - Investigate increase in DNS queueing time with native HTTPS records (Windows)

        • 1885556 - 4.62 - 4.32% fandom ContentfulSpeedIndex  fandom loadtime + 1 more (Linux) regression

    • Interop 2024 URL (scheduled, target H2 2024)

      • Comments: reviewing comments from Brian and Thomas

      • [dylan] https://bugzilla.mozilla.org/show_bug.cgi?id=1873944 

        • Pass some tests, fixing try push

    • H3 Upload speed

      • Status of HTTP/3 upload speed

        • Project Plan - Advancing neqo Library Performance and Developing a Robust Testing Framework

        • https://lookerstudio.google.com/reporting/4b5fbec3-2ead-40c1-864b-26490b85b6f5/page/p_7d9166xidd 

        • Chrome Benchmark results https://lookerstudio.google.com/reporting/3dcd8af9-9d9d-4197-a326-58b54f73bb36 

        • [kershaw]

          • Project plan for build a network performance test framework

            • In review process

            • Waiting for aerickson’s feedback

            • Will schedule a follow up meeting

    • 3rd Party Cookie Deprecation (in progress, target TBD)

      • [Ed] bvandersloot pushing to help resolve cookie limit behaviour

      • [Ed] looked into how we can trigger child-side SetCookieStringFromHttp for more thorough testing

    • Fetch keepalive: (in progress, Target Fx128) 

      • [Sunil] Adding Keepalive attribute support. Will land these by tomorrow:

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1871373

        • https://bugzilla.mozilla.org/show_bug.cgi?id=1871374

• Performance work: [jesup]

  • Preliminary results for Bug 1356686 - Do decompression off main thread in content

  • Bounced last night (1 failure on android and a non-unified-build issue); will try again hopefully today

Meeting 2024-03-12

• Team calendar

  • Next on Necko triage: Kershaw

• Round Table

  • (greg) HTTPS upgrade tracking telemetry https://bugzilla.mozilla.org/show_bug.cgi?id=1883954 

    • Yeah, looking at the bug I see that consolidating seems to break continuity in the existing probes. I think it's fair to keep them separate for now. I think we'll still be able to draw meaningful conclusions from separate sources

    • Any concerns if we leave it as is?

  • (vg) WebSocketStream · Issue 970 · mozillastandards-positions 

    • Randell Jesup what do you think?

    • Randell to file for standards position, current assessment is negative as this functionality is best supported with WebTransport over H2

• Active Projects roadmap 

  • Active Projects: Fx124 rel Mar 19, Fx125 soft freeze Mar 14 

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

      • Split CSSLoader OnStop - OMT to track in Fx126

        • [Sunil] CSSLoader and telemetry probes merged

      • OnStart consumer

        • – enter updates –

    • Optimize, Measure and add tests for Speculative Connections (in progress, target: Fx125) 

      • Beta experiment live, ends Mar 19, 2024

    • Native HTTPS RR Implementation (in progress, target Fx125)

      • 1874464 - Turn on native HTTPS resolver on Nightly landed!

        • Currently disabled on OSX due to intermittent crashes in dns_res_send - see Bug 1882856 - Marionette crash in dns_res_send with network.dns.native_https_query on OSX

        • Platforms where it currently works: Windows 11, Linux, Android 10+

        • Platforms where it doesn’t work: OSX, Windows 10, Android 9 and lower

      • Should it ride the trains or stay in Nightly?

      • Will send intent to ship soon.

    • H3 Upload speed

      • Status of HTTP/3 upload speed

        • Project Plan - Advancing neqo Library Performance and Developing a Robust Testing Framework

        • https://lookerstudio.google.com/reporting/4b5fbec3-2ead-40c1-864b-26490b85b6f5/page/p_7d9166xidd 

        • Chrome Benchmark results https://lookerstudio.google.com/reporting/3dcd8af9-9d9d-4197-a326-58b54f73bb36 

        • [kershaw]

          • Investigating running upload test with the my own worker

          • Asked some questions on #Taskcluster, but no response yet

    • 3rd Party Cookie Deprecation (in progress, target TBD)

      • [Ed] cookie capping - cookie limit behaviour

        • still need browser community insight from https://github.com/privacycg/CHIPS/issues/74

        • Ben will handle when he gets back next week

      • cookie capping - session cookies

        • not clear on whether they need to be rejected/counted against the capacity

        • need to look into how we are currently counting cookie space and discuss with 3PCD team

      • Tests for setting cookie from http child-side not obvious

        • Leander and I will be collaborating to get coverage here

      • Some code reviews happening as well

    • Fetch keepalive: target H1 2024 - Scheduled

    • WebTransport over H2: Targeting 2024 - Scheduled

    • Performance work: [jesup]

      • Preliminary results for Bug 1356686 - Do decompression off main thread in content

      • – enter updates –

Meeting 2024-03-05

• Team calendar

  • Next on Necko triage: Randell

• Round Table

  • Thanks Dr. Marten Richter for implementing serverCertificateHashes correctly! (Bug 1873263)

  • [acreskey] FYI, Firefox for Android vs Chrome pageload results, live sites, Pixel 3

    • Live pageload comparisons

• Active Projects roadmap 

  • Active Projects: Fx124 rel Mar 19, Fx125 soft freeze Mar 14 

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

      • Split CSSLoader OnStop - OMT to track in Fx125

        • Perf results are here.

        • TSAN looks good.

        • Waiting for final approval from Emilio

      • OnStart consumer

        • Email sent to tnikkel

        • Lots of discussion of feasibility in OMT matrix channel

          • OnDataStarting might not be practical unless we can sufficiently set up the transfer to allow OnDataAvailable to happen before OnStartRequest runs on MT.

    • Optimize, Measure and add tests for Speculative Connections (in progress, target: Fx125) 

      • Beta experiment live, ends Mar 19, 2024

    • Native HTTPS RR Implementation (in progress, target Fx125)

      • 1874464 - Turn on native HTTPS resolver on Nightly currently on autoland. Hopefully it will stick this time.

    • H3 Upload speed

      • Status of HTTP/3 upload speed

        • Project Plan - Advancing neqo Library Performance and Developing a Robust Testing Framework

        • https://lookerstudio.google.com/reporting/4b5fbec3-2ead-40c1-864b-26490b85b6f5/page/p_7d9166xidd 

        • [kershaw]

          • Investigating how to use quinn-udp in gecko

          • Quinn-udp doesn’t provide polling, so we need to use libc::poll directly or use mio.

    • 3rd Party Cookie Deprecation (in progress, target TBD)

      • [Ed] Clarifying question made to browser implementers of per-partition cookie cap here 

      • Also work continues on implementation/test of cookie cap

    • Fetch keepalive: target H1 2024 - Scheduled

    • WebTransport over H2: Targeting 2024 - Scheduled

    • Performance work: [jesup]

      • Preliminary results for Bug 1356686 - Do decompression off main thread in content

      • Big wins on pageload, up to 30+% on ESPN; many 5-10% wins

      • Surprising that we get this big a win; investigating

Meeting 2024-02-27

• Team calendar

  • Next on Necko triage: Manuel

• Round Table

  • Moving team discussions to matrix to encourage and support external contributors

    • Let's use more matrix instead of slack ourself?

      • Slack #quicdev -> Public #neqo:mozilla.org

        • Move to matrix

        • Close quicdev

      • Slack #necko-team -> Internal #necko-team:mozilla.org

  • (greg) Necko Bug Bash Day review bug stats 🎉

    • Total Bugs updated: 210

      • Closed: 69!

        • S3: 48 (Total S3 reduced by 2.6%)

        • S4: 3

        • NA: 18

      • Priority/severity updated: 20

      • Blocks updated: 73

      • Necko Priority Queue: 2 added 

      • Necko Next: 68 (added 19)

      • Necko Priority Review: down to 16

    • How could we improve by 10%?

      • Focus on another area next time

      • Rotate on areas, divide and have more overlap

        • One person focus on one area, rest of team move on to next ones, increase parallelism 

      • Try to fix some bugs together outside bash days (1-2hrs per week, with another team member)

    • Meta bug owners: Necko Meta Owners

      • (Sunil) What should be our plan going forward for driving our meta bugs?

        • Should we try to push 1-2 bugs into priority queue?

        • Discuss them in bug review and take appropriate action?

      • (greg) Meta Owner responsibilities?

        • Project meta

        • Maintenance 

          • Promote priorities to queue

          • Clean up bugs and drive fixes

          • Raise larger issues/initiatives  in roadmap planning

      • (manuel) Displaying ownership (meta bug assignee) in the necko-triage-helper

  • https://mozilla.social/@necko/ posts updates on our meeting notes.

  • [acreskey] FYI, thumbs up from Denis Jackson dns-prefetch for https

    • https://bugzilla.mozilla.org/show_bug.cgi?id=1596935

  • [jesup] Planning for possible ‘book club’ around “The Software Engineer’s Guidebook”

    • By Gergely Orosz, the author of The Pragmatic Engineer Newsletter and Building Mobile Apps at Scale

    • Perhaps in place of some Deciphering Necko sessions

  • https://bugzilla.mozilla.org/show_bug.cgi?id=1805796 tracked in Fx124/125

  • [cam] https://www.amazon.com/Beyond-Code-Setting-Software-Engineer/dp/B0CV5YMJB1 

• Active Projects

  • Active Projects: Fx124 code freeze Feb 15th roadmap Fx123 rel Feb 20

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

        • Release experiment ends next week

      • Split CSSLoader OnStop - OMT to track in Fx125

        • Debugging TSAN issue and testing updated changes.

        • [jesup] let’s push a perf try run

      • OnStart consumer

        • Jrmuziel positive on idea

        • Tnikkel responded, opening email conversation with him

    • Optimize, Measure and add tests for Speculative Connections (in progress, target: Fx125) 

      • Beta experiment live, ends Mar 19, 2024

    • Native HTTPS RR Implementation (in progress, target Fx125)

      • Landed a bunch of fixes. Turning on again this week.

    • H3 Upload speed

      • Status of HTTP/3 upload speed

        • Project Plan - Advancing neqo Library Performance and Developing a Robust Testing Framework

        • https://lookerstudio.google.com/reporting/4b5fbec3-2ead-40c1-864b-26490b85b6f5/page/p_7d9166xidd 

        • [kershaw]

          • Bug 1881490 Avoid duplicate process_output calls

            • Small fix to save 0.2~0.5% CPU usage

            • [jesup] r+’d

          • Investigating the performance difference between chrome and firefox

          • Packet reordering issue

            • Manuel will work on a patch to handle multiple ACKs at the same time

        • [jesup] issue with qlog cpu use at high bandwidths, even without qlog enabled

    • 3rd Party Cookie Deprecation (in progress, target TBD)

      • [Ed]

        • WIP patch in progress for cookie capping

          • https://phabricator.services.mozilla.com/D201757

        • Will be reviewing this patch ASAP

    • Fetch keepalive: target H1 2024 - Scheduled

    • WebTransport over H2: Targeting 2024 - Scheduled

Meeting 2024-02-20

• Team calendar

  • Next on Necko triage: Valentin

  • Bug bash day Friday Feb 23

  • IETF 16-22 March in Brisbane

• Round Table

  • (jesup)https://www.bleepingcomputer.com/news/google/new-google-chrome-feature-blocks-attacks-against-home-networks/ 

    • Bug 354493

    • With Google(Shipping 123) doing it we can follow, add to future watch list, maybe 2025

    • Indicated we’d probably look at it in 2024 in https://bugzilla.mozilla.org/show_bug.cgi?id=1481298#c6

    • https://chromestatus.com/feature/4869685172764672

    • Design document

    • See also https://bugzilla.mozilla.org/show_bug.cgi?id=1481298

  • (greg) Bug 1880173 - Networking: Domain Lists have moved to graveyard

  • (greg) Bug 1880165 - Move Proxy bugs into new Core :: Networking: Proxy component

    • In progress

  • (acreskey) non-comprehensive list of FOSDEM 2024 web perf talks of interest

    • The importance of Web Performance to Information Equity

    • Better than loading fast… is loading instantly!

    • Insights from the RUM Archive

    • Web Performance at Mozilla and Wikimedia (lightning talk)

    • Fast JavaScript with Data-Oriented Design

    • Some other talks in here: Web Performance devroom - FOSDEM 2024

    • Debugging HTTP/3 upload speed in Firefox

    • Firefox power profiling: a powerful visualization of web sustainability

  • (manuel) Another good FOSDEM 2024 talk: So You Think You Know Git (on fosdem.org)

  • (manuel) 103 Early Hints - preload (Shipped Fx123!)

  • (greg) Internet security and attacks stats from Cloudflare and privacytests.org 

• Active Projects

  • Active Projects: Fx124 code freeze Feb 15th roadmap Fx123 rel Feb 20

    • OMT: Move OnStartRequest and OnStopRequest to the content process (paused, target: TBD)

      • OMT OnStart/OnStop Request (content) threading/api model

      • new Project Plan

      • Split HTML5Parser OnStop - OMT Shipped in Fx122 

        • Release verification experiment running, ends march 5th

        • Early results seem at odds with beta experiment

          • Not clear they’re significant

        • Normal glean telemetry shows the same wins we saw in nightly and beta

      • Split CSSLoader OnStop - OMT to track in Fx125

        • [Sunil] WIll resume landing after code freeze. 

      • OnStart consumer

        • — enter updates —

    • Optimize, Measure and add tests for Speculative Connections (in progress, target: Fx125) 

      • [acreskey] Beta experiment w/ rel=preconnect starting today

    • Native HTTPS RR Implementation (in progress, target Fx125)

      • — enter updates —

    • H3 Upload speed

      • Status of HTTP/3 upload speed

        • Project Plan - Advancing neqo Library Performance and Developing a Robust Testing Framework

        • https://lookerstudio.google.com/reporting/4b5fbec3-2ead-40c1-864b-26490b85b6f5/page/p_7d9166xidd 

        • [kershaw] (read only)

          • New neqo 0.7.1 released

          • Reduces CPU use at high bandwidth, should raise max bandwidth

    • 3rd Party Cookie Deprecation (in progress, target TBD)

      • — enter updates — 

    • Fetch keepalive: target H1 2024 - Scheduled

    • WebTransport over H2: Targeting 2024 - Scheduled

Meeting 2024-02-13

• [acreskey] If you’ve experienced Android problems under network change or poor networks, let me know or add to this bug. https://bugzilla.mozilla.org/show_bug.cgi?id=1879387 So far the biggest issue seems to be that “Pull to refresh” is disabled again in Firefox for Android

• [acreskey] FYI, Performance Team has an OKR about full pageload performance benchmark that runs the entire network stack

• (greg) Networking components have been updated: FTP removed, Proxy added.

• H3 upload speed

  • [kershaw]

    • Firefox v.s. Chrome result

      • Only preliminary results. Will redo on another machine to see if the result is reproducible.

      • Observed packet loss on loopback interface when there is no traffic  shaper enabled.

• Zstandard [jesup] :

  • Chrome just sent Intent To Ship

  • Ready for ticket

  • Initial code to support a sandboxed utility process written

  • Unbitrotted OMT decompression for all decompression types patch (Bug 1356686)

    • Orange on mochitests on Try, debugging (all pass locally)

• [valentin] jschanck: Hi all, I plan to enable the X25519+Kyber768 key agreement mechanism for TLS 1.3 over TCP on the desktop Nightly channel on Monday. This is the first step in our migration to post-quantum cryptography, and I think it's (very!) important. However, we may see some negative performance impact on the tails due to Kyber's large public keys (1184 bytes sent by the client) and ciphertexts (1088 bytes sent by the server). Please reach out if you hear any reports of breakage or performance issues. The mechanism can be disabled by setting security.tls.enable_kyber to false. (dev-platform, Bug 1878725)

• [valentin] Bug 1824218 - Captive portal login tab does not get closed automatically

  • I happened to look through this report (even though I don’t have a lot of experience with the UI parts of captive-portal). I think this patch should reduce the number of reports from users who’s windows don’t close because of the race.

Meeting 2024-02-06

• Manuel Bucher gave a talk at FOSDEM this weekend “Debugging HTTP/3 upload speed in Firefox”. Recording should be available soon: https://video.fosdem.org/2024/h1302/ 

  • Slides available at https://fosdem.org/2024/schedule/event/fosdem-2024-1873-debugging-http-3-upload-speed-in-firefox/

• We’ve had to backout 1603699 - Enable DefaultURI use for unknown schemes due to breaking some protocol handlers.

  • Some older websites use a protocol format that don’t conform to the URL standard (example: https://en.wikipedia.org/wiki/Ed2k_URI_scheme#File_link_format )

  • We will try to enable this feature again, but add an allow list for known schemes that are broken.

• Contributor Max Inden has been making some great contributions to neqo lately.

Meeting 2024-01-30

• Valentin: Following discussions in the previous meetings we decided we want to start publishing public meeting notes again, to increase visibility of the team’s achievements.

  • The team’s meeting document will include a Public section. Add all of your updates there.

  • After the meeting one team member will upload this section to https://github.com/mozilla-necko/meeting-notes/ 

  • I’ve built a simple web-extension to upload clipboard contents to github.

    • Copy the contents of this section to the clipboard

    • Click webextension to upload to github

    • (optional) Update file title and contents.

    • DEMO

• Valentin: 1874137 - (necko-pref-flips) meta Tracking bug for necko pref flips 

  • We sometimes land fixed/features preffed-off and forget to turn them back on

  • We should start tracking these prefs and make sure to enable them as soon as possible to avoid issues. I’ve created a confluence page to define a process for tracking such prefs.

  • Could use timer additionally: Reminder Functionality in Bugzilla (via Autonag Bot), Example usage: https://bugzilla.mozilla.org/show_bug.cgi?id=1841061

• Manuel: 103 Early Hints preload (riding the trains, target: Fx123)

  • Sent out Intent to ship

  • Bug 1875268 - Fixed critical bug – uplifted to beta Fx123

  • Make sure added to release notes, mention to #web-performance-slack

• Bug 1868987 comment 20 HTTP/2 upload canceled due to ping timeout

  • Same problem as 12 years ago: Bug 790184 comment 11 and following. Workaround turned into the HTTP/2 upload problem we fixed last year.

  • Need help to figure out exactly what we want to do to handle 

• Fosdem in Feb 3-4 https://fosdem.org/2024/

  • Manuel is speaking in the Mozilla devroom Sunday 14:50: Debugging HTTP/3 Upload speed in Firefox 🎉

Meeting 2020-08-19

Round table:

• Bug 1659040 firefox dev edition not able to access internet intermittently

Next on triage duty:

Valentin

[in review]

Bug 1659132 - Make AddrInfo immutable

Bug 1645108 - Parse additional section of TRR response

[working on]

Bug 1525854 - TRR shouldn't fallback to DNS on DNSSEC error (extended error support)

Bug 1658277 - Record telemetry for the time between TRR mode change events.

Bug 1658278 - Figure out better confirmation process for TRR

Bug 1656884 - Issue parallel AAAA/A and SVCB queries for the name

Meeting 2020-08-13

Round table:

Next on triage duty: Dragana

Valentin

1651672 | Clear and remove DataStorageClass::TRRBlacklist

1656862 | Improve TRR_SKIP_REASON telemetry

1657582 | Merge asyncResolveByType and asyncResolveWithTRRServer into asyncResolve

1657897 | Convert TRR split horizon mitigations to a doh-rollout heuristic

Meeting 2020-08-06

Round table:

• Honza: 1564034 - Confusing error message is displayed when a webpage can't be loaded
  • who in mozilla is responsible for updating error pages with respect to UX?

Next on triage duty: Kershaw

mayhemer

• analyzing rel=preload regressions found by Andrew (time consuming)
• assigned to emilio 1657173 - [rel=preload] stylesheets that are only preloaded with must not block document's onload
• still needs more care, but deprioritized now: 1609410 - 421 response code from icloud.com not handled well with a proxy, NTLM involved
• diagnosed 1655567 - Not gracefully erroring an h2 stream with more summed data from DATA frames than announced by content-length
• next 1654585 - 1.92 - 2.33% Explicit Memory / Resident Memory (windows10-64-shippable-qr) regression (Mon July 20 2020)
  • likely caused by my cookie console warning patch
• logan added a small feature: clicking the new wallclock icon (icons appearing on the left when hovering a line) will set that line as 0-time and show a relative time on all other lines in ms. useful for perf diagnoses
• triage

Dragana

Bug 1655843 Call ProcessPendingQueue when a Http3 connection changes to state CONNECTED

Bug 1655782 Turn on the QPACK dynamic table

bug 1656870 - Crash in [@ neqo_http3::recv_message::RecvMessage::receive_internal]

PR 899 - We were entering same stream multiple time into the block streams list

PR 898 Send Qpack's StreamCancelled instruction

Server-side tests in network simulator:

PR 886 Set granularity of the server timer to 1ms

PR 888 Write datagrams into socket immediately

PR 906 In neqo-http3-server we should read from the sockets more often

Working on:

PR for issue 895: Add more tests with simulated network conditions

PR for issue 896: Have a way to measure throughput for the net simulation tests

Junior

• This week
  • Bug 1647950 - Intermittent ASSERTION: Did not receive all required callbacks!: 'NS_FAILED(mResult) || mExpectedCallbacks == 0', file /netwerk/base/nsAsyncRedirectVerifyHelper.cpp | assertion count 1 is more than expected 0 assertions
  • Bug 1655566 - HTTP3 probes in cpp side to detect regressions
    • time to reuse of idle connection
    • delay from HTTP3 timers
    • blocked by stream limits
    • blocked by flow control as a sender
  • Bug 1655636 - avoid reinitializing the timers in HTTP3
  • Bug 1655895 - bogus timeout passed to Http3Session::SetupTimer
• Next week
  • Bug 1657169 - HTTP3 probes for TLS handshake and transport handshake time
  • Bug 1657172 - probe HTTP3 layer performance

Tania

Worked on docs.

Added more benches.

Experimented with making serialization faster.

Made minor changes in wrapper.

Kershaw

Landed:

• Bug 1650773 - [socket process] improve sending data directly to child processes
• Bug 1600129 - [socket process] Send ODA data to HttpTransactionParent in chunks
• Reland websocket patches

Ongoing:

• Bug 1652655 - Using HTTPSSVC as an AltSvc
• Bug 1652668 - Create a domain exclusion list for SvcDomainName that have failed
• Bug 1652670 - Add echconfig and A/AAAA DNS record to nsHttpConnectionInfo

Michal

• 1656128 Remove probe NETWORK_PAC_URL_SCHEME in Firefox 82
• QNS: Server test failures (DC,M)
• neqo vs TCP testing
• Enable http3 xpcshell test on Asan/Tsan

Valentin

Bug 1652106 - Address httpsvc draft changes [landed]

Bug 1656862 - Improve TRR_SKIP_REASON telemetry [landed]

Doing some more investigations about weird probe data Bug 1657582 - Merge asyncResolveByType and asyncResolveWithTRRServer into asyncResolve

Andy

Neqo (6, 48, 62)(+2, +0, +4):

• [merged] qlog: Fix packet length calculation 900
• [merged] Add APIs to get odcid to Connection and Http3Client 890
• [merged] client: Fix client for http3 889
• [in-review] Don't cleanup bidi recv stream until send stream is done 871
• [needs-revisions] Fix panic on QNS when using 0.9 server 877
• Diagnosed issue against Cloudflare server 614 which they fixed
• 0.4.9 tagged and autolanded
• [done] Rename master branch to main 845

Next:

• Fix or explain all QNS issues
• Make suggested changes to Add network.http.http3.qlog_dir pref to Firefox

Meeting 2020-07-30

Round table:

Next on triage duty: Honza

Andy

Neqo (4, 48, 58)(-1, +0, +3):

• [merged] Use indexmap instead of HashMap for send streams 874
• [merged] Do not report blocked to peer until credits run out 872
• [merged] Do not disable PTO when packets outstanding == 0 for initial/handshake 864
• [merged] Add qlog::connection_state_updated 859
• [in-review] Don't cleanup bidi recv stream until send stream is done 871
• [in-review] Fix some congestion control bugs and improve cwnd qlogging 878
• quic-network-simulator: More info for packet disposition for drop/corrupt scenarios 76
• quic-interop-runner: Give more info about end time and estimated next end time 121
• [Aug 1] Rename master branch to main 845

Next:

• Complete my tasks for Neqo Milestone 1 and 1.5, esp. 799
• Make suggested changes to Add network.http.http3.qlog_dir pref to Firefox

Tania

Fixing issues in wrapper

Adding benchmarks for the crate

Michal

• triage
• netsim for CI
• QNS: Server test failures (DC, B)

Kershaw

Ongoing:

• Bug 1652655 - Using HTTPSSVC as an AltSvc Working on creating a test.

Meeting 2020-07-23

Round table:

Next on triage duty: Michal

Kershaw

Landed:

• Bug 1652762 - Intermittent SUMMARY: ThreadSanitizer: data race /builds/worker/checkouts/gecko/netwerk/dns/TRRServiceBase.cpp:71:22 in CheckURIPrefs

Ongoing:

• Bug 1652655 - Using HTTPSSVC as an AltSvc
• Bug 1497249 - [socket-proc] Connection stickiness handling for upgrade observers
• Bug 1654282 - websocket fails to connect
• Bug 1653642 - Crash in [@ mozilla::net::CallOnStop::~CallOnStop]

Tania

Made minor changes to both crate and bindings

Added more tests

Wrote documentation

Andy

Triage

Neqo (5, 48, 58)(+2, +2, +4):

• [in-progress] QNS/client: Fix issues with handshakecorruption (C1) and handshakeloss (L1) tests 799
• [in-progress] QNS: handshakeloss (L1) failure with aioquic 841
• In working on aioquic issues, discovered we are not counting PTO packets as congestion control bytes-in-flight and other issues, causing Neqo to not match the spec. Now fixed, but work remains to pass handshake tests with aioquic.
• [in-review] Do not allow initial packet to exceed cwnd in persistent congestion 829
• Adding Neqo in server role in QNS has caused an increase in reported issues from other implementers -- opened 846 and 847

DoH:

• Issue with my WiFI AP's DNS not handling doh.test turned out to be nothing.

Next:

• Complete my tasks for Neqo Milestone 1 and 1.5, esp. 799
• Make suggested changes to Add network.http.http3.qlog_dir pref to Firefox

mayhemer

• [postponed] 1554652 - Implement asynchronous opening of alternative output stream
• [landed] 1605895
• [landed] 1596741 - Emit warning to console when rejecting to set cookie (from cross-site/foreign domain)
• [landing soon] 1654556 - Log also "failed the samesite tests" cookie rejection reason.
• [landing soon] 1609410 - 421 response code from icloud.com not handled well with a proxy
• [pending r?] 1649813 - Test that browser.dns APIs don't work when the proxy should be handling DNS
• [landed] 1628713 - Add Tokenizer::CheckPhrase to quickly check for compound strings
• [in prog] 1652898 - The stream transport service's thread pool keeps starting and stopping threads during downloads
• [landed] 1652300 - ThreadSanitizer: data race [@ mozilla::net::CacheEntry::InvokeAvailableCallback] vs. [@ mozilla::net::CacheEntry::OnHandleClosed]
• r?, ni?

Michal

• Bug 1650363 Performance testing of HTTP3 in Firefox using only xpcshell tests
• Bug 1646025 Add a test for 421 response
• QNS: Server test failures (DC, B)

Junior

• This week
  • Bug 1652189 - have a way to pass cookie to child process for LOAD_DOCUMENT_NEEDS_COOKIE
  • Bug 1654614 - resume channel event queue for parent intercepted channel
  • Bug 1649873 - Enable stream filter test for remote only
• Next week
  • QUIC telemetry

Dragana

ECH: proposal for the API between necko and nss(psm)

HTTP3:

• Meeting about Network Throttling for performance testing
• Looking to define what we need to test before a study in Nightly
• Defining fro performance test
• Define HTTP3 telemetry that we want to collect

Meeting 2020-07-16

Round table:

• osclientcerts

Next on triage duty: Andy

Dragana

HTTP3:

Bug 1652573 - Neqo 0.4.8 in mozilla-central

Bug 1652104 - Telemetry about HTTP3 usage

Bug 1651814 - Do not use Http3 if be-conservative flag is set

Bug 1650992 - http3 POST request never gets an answer

DoH:

Looking at TRR fraction (does not look that bad on Nightly and Beta)

planning to look at TRR fraction on release and at TRR_SKIP_REASON_TRR_FIRST probe

ECH/HTTPSSVC(HTTPS DNS record)

Revise the project plan and open bugs

mayhemer

• landed 1651661 - HttpChannelChild::Redirect3Complete doesn't call the channel listener on failure
• fixed 1649813 - Test that browser.dns APIs don't work when the proxy should be handling DNS
• landed 1651068 - Include Task from TaskController in MOZ_LOG=events
• landed 1650613 - about:cache shows error message: Unrecognized storage name 'appcache' in about:cache URL
• r? 1596741 - Emit warning to console when rejecting to set cookie (from cross-site/foreign domain)
• r? 1649320 - null pointer passed as argument 2, which is declared to never be null in netwerk/cache/nsCacheMetaData.cpp
• ongoing 1554652 - Implement asynchronous opening of alternative output stream
• investigating preload regressions mainly on cnn
  • waiting for more results from acreskey
• still investigating 1605895
• investigating 1641696 - Firefox takes 50% more time than Microsoft Edge to display first paint of comments in reddit posts
• investigating 1652083 - Slower than expected load of cnn.com
  • seems like proxy resolution (the thread) is hogged
  • not related to preload
• few more I didn't make note of
• r?, ni?

Kershaw

Landed:

• Bug 1497249 - [socket-proc] Connection stickiness handling for upgrade observers
• Bug 1646358 - [socket process] Fix test failures regarding devtool tests

Ongoing:

• Wrap up socket process project in this doc.
• Bug 1650773 - [socket process] improve sending data directly to child processes
• Bug 1600129 - [socket process] Send ODA data to HttpTransactionParent in chunks
• Bug 1648996 - [socket process] devtools/client/netmonitor/test/browser_net_truncate-post-data.js failed with socket process enabled
• Bug 1648515 - AddressSanitizer: heap-use-after-free /builds/worker/checkouts/gecko/dom/fetch/Fetch.cpp:600:9 in mozilla::dom::MainThreadFetchResolver::OnResponseAvailableInternal(mozilla::dom::InternalResponse*)
• Bug 1652762 - Intermittent SUMMARY: ThreadSanitizer: data race /builds/worker/checkouts/gecko/netwerk/dns/TRRServiceBase.cpp:71:22 in CheckURIPrefs

Junior

• [land] Bug 1596904 - Intermittent TEST-UNEXPECTED-TIMEOUT | toolkit/components/extensions/test/xpcshell/test_ext_webRequest_filterResponseData.js | Test timed out
• [land] Bug 1647954 - Intermittent toolkit/components/extensions/test/xpcshell/test_ext_webRequest_filterResponseData.js | test_alternate_cached_data - [test_alternate_cached_data : 392] message queue is empty - "[\"onHeadersReceived\"]" == "[]" After xpcshell return code: 0
• Bug 1633342 - [meta] Crash in [mozilla::net::nsHttpConnectionMgr::Shutdown] and other net related places. Shutdown hang.
• Bug 1651973 - 3.57 - 15.01% raptor-tp6-slides-firefox-cold / raptor-tp6-slides-firefox-cold loadtime (windows10-64-shippable-qr) regression on push d59f0a83a7e9ec6fda8f7327f3154d67b266ec65 (Wed July 8 2020)

Tania

Worked on minor changes to structured headers crate

Finished writing tests for xpidl bindings

Andy

Neqo (3, 46, 54)(-1,-5,+5):

• [merged] Large transport bufs follow-on PR 810
• [merged] QNS H/3 & H/0.9 server tests 808 815
• [in-progress] QNS/client: Fix issues with handshakecorruption (C1) and handshakeloss (L1) tests 799
  • Reviewed/merged mt's 822 to keep PTO active after Initial keys dropped
  • 2 servers went green but 2 others still remaining as failing

DoH:

• Investigating if my Wifi AP's DNS forwarder is blocking doh.test

Next:

• Complete my tasks for Neqo Milestone 1 and 1.5, esp. 799
• Revise "OS based DoH" doc.
• Make suggested changes to Add network.http.http3.qlog_dir pref to Firefox

Michal

• 1621345 Support for nsIWebSocketEventListener.send() API
• 1645051 Calculated smart max cache size shouldn't overwrite browser.cache.disk.capacity pref
• 1646024 The Http3 timer sometimes fires too early
• Issue 819 QNS: server data-send tests failing

Meeting 2020-07-09

Round table:

• bug 1648635 - channel.setResponseHeader
• Sending ODA from socket process to child process directly will break RCWN.

Next on triage duty: Junior

Kershaw

Landed:

• Bug 1647759 - Intermittent SUMMARY: ThreadSanitizer: data race /builds/worker/checkouts/gecko/netwerk/base/nsIOService.cpp:1280:16 in mozilla::net::nsIOService::SetOffline(bool)
• Bug 1527256 - [socket process] make about:networking work
• Bug 1644834 - about:memory tests fail on my machines (macOS Catalina, Ubuntu 18.04)

Ongoing:

• Bug 1650773 - [socket process] improve sending data directly to child processes
• Bug 1648996 - [socket process] devtools/client/netmonitor/test/browser_net_truncate-post-data.js failed with socket process enabled

Tania

Writing tests for xpidl bindings

mayhemer

• progressing on 1596741 - Emit warning to console when rejecting to set cookie (from cross-site/foreign domain)
• progressing on one sec bug
• progressing on 1554652 - Implement asynchronous opening of alternative output stream rebase + try push
• progressing on 1651068 - Include Task from TaskController in MOZ_LOG=events to diagnose 1641696 - Firefox takes 50% more time than Microsoft Edge to display first paint of comments in reddit posts
• progressed and then given away 1605305 - Origin header field is not set to HTTP request in cases where it is required

preload: acreskey found regressions on cnn and booking.com via testing, will look at them soon.

Junior

• Landed
  • Bug 1646592 - Sending endpoint of stream filter should not block main thread IPC
  • Bug 1645901 - Send necessary cookie through pBackground before OnStartRequest
  • Bug 1649965 - only waiting for main thread OnStartRequestSent for document load
  • Bug 1647978 - Intermittent toolkit/components/extensions/test/xpcshell/test_ext_webRequest_filterResponseData.js | test_filter_content_fetch - [test_filter_content_fetch : 311] Expected response for http://example.com/redirect?redirect_uri=http://example.net/dummy -
• Next
  • Bug 1596576 - Upload speed is so slow in Firefox, all other browsers are fine (please read detailed explanation)
  • Bug 1649848 - [meta] make test_ext_webRequest_filterResponseData.js stable

Michal

• extending QNS interop runner
• neqo testing

Dragana

neqo:

Server Push landed

PR 800 - Document errors for neqo-transport::Connection::stream_recv

PR 801 - Add better handling of transport errors in send_request_body function

PR 802 - Fix error handling in read_response_data

PR 803 - Handle properly errors from set_resumption_token

PR 805 - Faster fetch

Bug 1651489 - update neqo to 0.4.6

Looking at DoH telemetry

Andy

Neqo (4, 51, 54)(+1,-1,+1):

• Large transport bufs PR 806
• [in-progress] QNS: neqo-server transfer test case
• [in-progress] QNS/client: Fix issues with handshakecorruption (C1) and handshakeloss (L1) tests
• [merged] qlog: Refactor
• Neqo 0.4.6

DoH:

• Wrote a draft doc "Response Options for ISP Blocking of Global Canary"

Next:

• Complete my tasks for Neqo Milestone 1
• Revise the above doc, and the "OS based DoH" doc.
• Make suggested changes to Add network.http.http3.qlog_dir pref to Firefox

Meeting 2020-07-02

Round table:

Next on triage duty: Dragana

Valentin

[landed]

1618271 - DNS leaks about browser.dns API

1641222 - Follow CNAME/AliasForm chains

1649401 - Minor TRR code cleanup

1649417 - Don't follow CNAME chains when not necessary

[working on]

1649127 - Make sure we only set AddrHostRecord::mTRRUsed = true when we actually send the request

1649143 - Record telemetry with reason we fell back to Do53

[triage]

PTO for 3 weeks - still working 20% with Tania and urgent issues.

Tania

Continue working on xpidl bindings for structured-headers

Junior

• Land
  • Bug 1633935 - Allow sending onStartRequest via pHttpBackgroundChannel
  • Bug 1648035 - Intermittent Assertion failure: mChannelChild (no channel child in RecvOnStatus), at /builds/worker/checkouts/gecko/netwerk/protocol/http/HttpBackgroundChannelChild.cpp:160
  • Bug 1648712 - Intermittent dom/base/test/chrome/test_bug884693.xhtml | Got expected message count - got 1, expected +0
  • Bug 1578230 - Crash [@ mozilla::net::SpdyConnectTransaction::MapStreamToHttpConnection] with HTTP2 Proxy
  • dup -Bug 1647096 - GeckoProfiler.Markers gtest crashes after changing tests to use socket process
• Ongoing
  • Bug 1649848 - [meta] make test_ext_webRequest_filterResponseData.js stable
  • other regressions
  • cookie and permission via pBg

mayhemer

• investigated 1567863 - Firefox startup slow - window loads quickly, but first network call delayed, works normal after that
  • lot of ~8 seconds delays here and there (I/O, script execution blocking main thread)
• took over and tuning 1605305 - Origin header field is not set to HTTP request in cases where it is required
  • some tests need care
• investigating two sec bugs
• fixed and waiting for field test 1609410 - 421 response code from icloud.com not handled well with a proxy
• fixed 1615121 - Intermittent GECKO(1119) | SUMMARY: ThreadSanitizer: data race /builds/worker/workspace/build/src/netwerk/cache2/CacheEntry.cpp:1436:33 in mozilla::net::CacheEntry::MetaDataReady()
• fixed 1632215 - Intermittent SUMMARY: ThreadSanitizer: data race /builds/worker/checkouts/gecko/netwerk/protocol/http/nsHttpChannel.cpp:8439:7 in mozilla::net::nsHttpChannel::OnDataAvailable(nsIRequest*, nsIInputStream*, unsigned long, unsigned int)
• fixed 1649003 - Log JS stack on more places in nsHttpChannel and Child
• fixed 1648781 - Log timer events
• fixed 1642602 - Some disk cache values in about:cache are incorrect.
• duplicated 1560090 - High refresh rate monitor (>200Hz) slows performance
  • an idle dispatch (with 250ms timeout) on the critical path for all resource loads when filtering extensions are involved
• blog post about preload
• preload for workers (fetch/xhr) seems not be a priority
• plan for better network markers in "Waiting for socket thread" span
  • don't expect anything soonish
• ni, r

Andy

Neqo (3, 52, 53)(-3,+3,-4):

• [merged] Implement more qlog traces: CC and Recovery state
• [merged] client: Make H9 code read all available data on RecvStreamReadable
• [in-review] Add network.http.http3.qlog_dir pref to Firefox
• [approved] qlog: Refactor
• [approved] qlog: more-unique filenames
• QUIC planning mtg: Realization of the scope of work for testing needed for rollout. Prioritization of work vs other projects like Fission. Contacts and coordination with QA. Introduction of Milestone 1.5. Followup mtg early next week.

DoH:

• Wrote a draft doc to start talking about how to interact with OSes rolling out DoH support

Next:

• Write a policy options doc for dealing with ISPs who block canary
• Fix Neqo interop issues
• Confirm DoH canary abuse and take follow-up actions

Kershaw

Landed:

• Bug 1648189 - [socket process] Perma Assertion failure: tqParent, at /builds/worker/checkouts/gecko/netwerk/protocol/http/HttpTransactionParent.cpp:173
• Bug 1650068 - [socket process] reenable profiler tests
• Bug 1648008 - Unable to connect to Google first time using it through DuckDuckGo bang after clearing all history

Ongoing:

• Bug 1527256 - [socket process] make about:networking work
• Bug 1646358 - [socket process] Fix test failures regarding devtool tests
• Bug 1648996 - [socket process] devtools/client/netmonitor/test/browser_net_truncate-post-data.js failed with socket process enabled

Michal

• 1627910 Enable test netwerk/test/unit/test_http3
• 1649410 [socket-process] test_version_fallback() in netwerk/test/unit/test_http3.js fails
• neqo testing

Dragana

Finishing up PUSH for neqo,

Looking into how errors are propagated in necko: there are 3(4) layers and looking which layer should handle which error.

Looking at goodput and crosstraffic test result -> flow control need to be improved.

PR 769 Small changes

PR 770 Small fix 2

Bug 1642559 SPDY_REQUEST_PER_CONN_2 looks broken

Bug 1595269 HTTP/2.0 still displayed even if HTTP/3.0 is used

1648553 Sometimes page loads start hanging in all tabs with http3

Meeting 2020-06-11

Round table:

• are VPN related bugs (1638542, 1647985, 1635935) higher priority now when DoH is also affected?

Next on triage duty: Valentin

mayhemer

READONLY (can't attend for personal reasons)

• 1648471 - [rel=preload] Enable fully on Beta 79 we don't have a macro for that
• landed 1631327 - Crash in [@ IPCError-browser | RecvPredLearn]
• landed 1487101 - CacheStorageService::PurgeOverMemoryLimit() doesn't work correctly under specific circumstances
• landed 1603542 - [rel=preload] Log a warning when preloaded resources are not used
• landed 1638925 - Extend events MOZ_LOG module, includes IPC, DOM promises, message queue, logan updated too, I will intro some other time
• moved on a bit with 1605895
• reduced my ni? queue greatly!
• next 1639914 - SEC_ERROR_OCSP_SERVER_ERROR with most https websites when security.OCSP.require is set to true and a https .pac file is used
• next 1632215 - Intermittent SUMMARY: ThreadSanitizer: data race /builds/worker/checkouts/gecko/netwerk/protocol/http/nsHttpChannel.cpp:8439:7 in mozilla::net::nsHttpChannel::OnDataAvailable(nsIRequest*, nsIInputStream*, unsigned long, unsigned int)
• next 1560090 - High refresh rate monitor (>200Hz) slows performance

Michal

readonly

• triage
• 1643866 Firefox fails to resolve specific domain: NS_ERROR_UNKNOWN_HOST
• 1645797 PR_CONNECT_RESET_ERROR when connecting to belastingdienst.nl
• 1644374 Assertion *countWritten > 0 in nsHttpTransaction::WritePipeSegment
• PR750 neqo-qns
• PR755 QNS qlog
• PR768 neqo-client: fix QNS compliance check
• PR765 - fix zerortt QNS test
• 1627910 Enable test netwerk/test/unit/test_http3

Kershaw

Landed:

• Bug 1644239 - Crash in [@ mozilla::net::Http2Session::RecvPushPromise]
• Bug 1644834 - about:memory tests fail on my machines (macOS Catalina, Ubuntu 18.04)
• Bug 1643515 - Crash in [@ mozilla::net::HttpTransactionParent::DoNotifyListener]
• Bug 1640612 - socket process gets killed due to sandboxing

Ongoing:

• triage
• Bug 1646358 - [socket process] Fix test failures regarding devtool tests
• Bug 1647096 - GeckoProfiler.Markers gtest crashes after changing tests to use socket process
• Bug 1646689 - 2.01 - 4.86% raptor-tp6m-allrecipes-geckoview-cold loadtime / raptor-tp6m-cnn-geckoview-cold confidence (android-hw-g5-7-0-arm7-api-16-shippable) regression on push 567513cee4895872b1e7fb53fbd2793593d11f97 (Fri June 5 2020)
• Bug 1648189 - [socket process] Perma Assertion failure: tqParent, at /builds/worker/checkouts/gecko/netwerk/protocol/http/HttpTransactionParent.cpp:173

Tania

Implemented headers serializer

Added specification tests from https://github.com/httpwg/structured-field-tests

Working on XPIDL bindings for structured-headers crate

Junior

• Landed
  • Bug 1633935 - Allow sending onStartRequest via pHttpBackgroundChannel
    • Depends on
      • Bug 1645941 - data race StreamFilterParent::mState in OnStartRequest and ODA
      • Bug 1645527 - Remove redundant PHttpChannel IPCs
    • Regressions
      • Bug 1648132 - Crash in [@ mozilla::extensions::StreamFilterParent::Init]
• Next
  • Bug 1645901 - Send necessary cookie through pBackground before OnStartRequest
  • Bug 1646592 - Sending endpoint of stream filter should not block main thread IPC
  • Bug 1648035 - Intermittent Assertion failure: mChannelChild (no channel child in RecvOnStatus), at /builds/worker/checkouts/gecko/netwerk/protocol/http/HttpBackgroundChannelChild.cpp:160
  • Bug 1578230 - Crash [@ mozilla::net::SpdyConnectTransaction::MapStreamToHttpConnection] with HTTP2 Proxy

Valentin

Bug 1647638 Add nsStandardURL assertions

Bug 1618271 - DNS leaks about browser.dns API

TRR documentation

rust-url fixes

DNS code refactoring

Andy

Neqo (6,49,57)(+0,+3,+1):

• Neqo 0.4.4 tagged and merged into m-c.
• PR 754 Properly handle receipt of StreamDataBlocked frame
• PR 746 Avoid borrowing FlowMgr twice when handling lost MaxStreamData token
• [in-review] PR 752 Implement more qlog traces: CC and Recovery state
• WIP: Add network.http.http3.qlog_dir pref to Firefox. Works, but needs as-yet-untagged Neqo changes before can be posted for review
• WIP: Set Qlog dir via about:networking (depends on above)
• PR Reviews
• Tried to get QNS Docker image autobuilding from GitHub commit without success. Some sort of permissions issue? Maybe moving neqo repo to mozilla-necko org would make easier.

Next:

• Fix Neqo issues
• Confirm DoH canary abuse and take follow-up actions

Dragana

Doc about HTTP3 Performance study

Requirements for turning Http3 in Nightly (a doc)

Looking on Http3 bugs in Firefox

Add support for draft28 and draft29 in m.-c.

PR 766 - Write the header block's prefix properly

Bug 1646318 - test_post and test_patch in netwerk/test/unit/test_http3.js time out

Meeting 2020-06-04

Round table:

Next on triage duty: Honza

Andy

Neqo (8, 39, 55)(+3,+0,+1):

• Neqo 0.4.0 tagged and merged into m-c
• Worked on draft 28 support while still keeping draft 27 support (#686) 27 needed for deployed QUIC sites (FB, Google, CF) with FF but interop/testing (e.g. QNS) quickly moved to 28
• merged: Persistent congestion interval start time
• PR Reviews
• News: Virtual interop June 10. Draft 29 in the works, may be the final draft release

PTO

Next:

• Finish draft 28 and dual-version PR
• Plan for another quick Neqo tag and m-c merge

Valentin

[landed]

Bug 1631748 - Fix COOP expected fails

Bug 1637512 - Loss of network connections when IPv6 and DoH enabled

Bug 1640867 - Record TRR steering telemetry separately

Bug 1641901 - MOZ_RELEASE_ASSERT(is()) when WebSocketChannel::SendBinaryStream() is called and nsWebSocket logging is enabled

Bug 1642318 - network.http.sanitize-headers-in-logs should be RelaxedAtomicBool

Bug 1643264 - Failed ResolveWithTRR requests should not be retried to get the TTL

[in review]

Bug 1636888 - Allow specifying entire rust modules in MOZ_LOG

[working on]

Bug 1640724 - Intermittent netwerk/test/unit/test_trr_case_sensitivity.js

Bug 1640091 - dnsResolve() failure in Proxy Auto-Configuration (PAC)

Kershaw

Landed:

• Bug 1641496 - [socket process] Investigate disabled xpcshell tests
• Bug 1524035 - Assertions in socket process do not output a stack trace

Ongoing:

• Bug 1640612 - socket process gets killed due to sandboxing
• Bug 1640105 - [socket process] Run mochitest with networking over socket process enabled
• Bug 1643382 - [socket process] make sure we don't leak HttpBackgroundChannelChild and BackgroundDataBridgeChild
• Bug 1642193 - 3 /xhr/ web-platform tests are expected TIMEOUT

Tania

Working on tests for structured-headers.

Michal

• 1621345 Support for nsIWebSocketEventListener.send() API
• FTP PAC telemetry data from release
• Neqo PR682 - neqo-client: fix multiplexing test
• 1635935 HTTP/2 issue when toggling VPN state
• 1569715 CORS preflight requests are cached when 'Disable cache' is checked
• Issue 676 - neqo-server: implement "transfer" test case

Dragana

PR 677 - Server Push: client side only

Reviewing quic docs and looking what is missing in http3 and qpack part

Meeting 2020-05-28

Round table:

Next on triage duty: Dragana

mayhemer

• preload:
  • enabled by default on Nightly (and rides to Early Beta only)
  • investigating a perf regressions: 1641532 - 2.9 - 17.82% raptor-tp6- (linux64, macosx1014-64-shippable, windows10) regression on push 2bccc3e80006f2ad27060453135b19fac2d43460 (Tue May 26 2020)
• reviews and ni?

Valentin

[landed]

Bug 1611990 - Rust log messages of less than WARN level are optimized out #645

Bug 1635566 - logincdn.msauth.net fails to resolve if network.trr.mode=3

Bug 1636891 - Move logging specifics from toolkit/library/rust/shared/lib.rs to the gecko_logger crate

Bug 1640872 - Add more TRR logging

Bug 1640879 - Fix COEP data.http.html test

[working on]

Bug 1641222 - Follow CNAME/AliasForm chains

Bug 1616252 - Maybe add a /etc/hosts parser

Bug 1636888 - Allow specifying entire rust modules in MOZ_LOG

Bug 1627800 - 404 http error isn't displayed in Firefox

Bug 1629862 - Improve error messages for COEP failures

Dragana

Working on improving neqo-http3 code and implementing ServerPush:

• Done: PR 657 - Use separate send_message/recv_message instead of transactions
• One more small refactor and ServerPush PRs almost ready

Working on HTTP3/QUIC telemetry doc

PR 669 - Drop packets with unknown versions

neqo reviews

Open some Http3 spec issues

Michal

• [landed] 1637947 network ID: link status isn't updated in NetlinkService::OnAddrMessage()
• [landed] Neqo PR667 - Add test chacha20
• [landed] Neqo PR663 - Add retry, resumption and zerortt tests
• Issue 674 - Handle Error::StreamLimitError returned from Connection::stream_create
• Issue 676 - neqo-server: implement "transfer" test case
• 1621345 Support for nsIWebSocketEventListener.send() API
• FTP PAC telemetry data from release

Tania

Continue working on structured http headers parser

Kershaw

Landed:

• Bug 1629457 - Intermittent netwerk/test/unit/test_trr.js | application crashed [@ __thiscall InitHash::InitHash + 0x12d3] | TRRServiceChannel using nsWeakReference OffMainThread
• Bug 1546358 - [socket process] cache storage service is used in Http2Session
• Bug 1602832 - [socket process] Enable xpcshell tests for socket process
• Bug 1641427 - startup Crash in [@ mozilla::SandboxBroker::LaunchApp]

Ongoing:

• Bug 1637648 - [socket process] Forward observer notifications in nsHttpHandler to socket process
• Bug 1641496 - [socket process] Investigate disabled xpcshell tests
• Bug 1640612 - socket process gets killed due to sandboxing
• Bug 1640105 - [socket process] Run mochitest with networking over socket process enabled
• Bug 1524035 - Assertions in socket process do not output a stack trace
• Bug 1638387 - nsOSXSystemProxySettings::Init() is failed in socket process when sandbox is enabled

Junior

Bug 1633935 - Allow sending onStartRequest via pHttpBackgroundChannel

• all r+, fixing oranges - cross-process redirection/onStart before permission received/graphic async test/socket process/...

[landed] Bug 1638172 - avoid name hiding for handling multipart channel in HttpChannelChild

[landed] Bug 1638313 - Cross-Origin request blocked is not visible in Network panel

Meeting 2020-05-21

Round table:

• [mayhemer]: 1628262 - any thoughts about how to move the crash investigation?
• Implement asynchronous opening of alternative output stream
• [grover]: 1636109 - 76.0 doesn't always resend the POST request body after Kerberos authentication
• Odd results testing ./mach browsertime with fission

Next on triage duty: Junior

Andy

Triage

Neqo (5, 39, 54)(-2,+7,+2):

• Targeting re-enabling in Nightly for FF 80
• Worked on neqo_glue changes for header being returned in an event. All other changes needed to land a new Neqo are working.
• in review/changes req'd: Persistent congestion interval start time
• in review/changes req'd: qpack encoder small refactor
• merged: usize and Res simplifications in recv_stream.rs
• merged: Add [self] to padding log
• PR Reviews

Next:

• Still have PR backlog
• Finish glue changes for next Neqo->Gecko merge
• PTO next week

Michal

• 1630000 Hitting Back sometimes shows old cached content
• 1637947 network ID: link status isn't updated in NetlinkService::OnAddrMessage()
• 1635935 HTTP/2 issue when toggling VPN state
• Neqo PR663 - Add retry, resumption and zerortt tests (https://github.com/mozilla/neqo/pull/663)
• review 1497249 - WebSocketConnection
• 1616979 Nightly - Firefox caching too much - Not respecting the "CTRL+ Shift +R" shortcut or the "Disable Cache" checkbox

Junior

[WIP] Bug 1633935 - Allow sending onStartRequest via pHttpBackgroundChannel

[r+] Bug 1638313 - Cross-Origin request blocked is not visible in Network panel

[r+] Bug 1638172 - avoid name hiding for handling multipart channel in HttpChannelChild

[landed] Bug 1633942 - introduce HttpChannelOnStartRequestArgs

mayhemer

• preload
  • ready to enable on nightly and early beta next week
  • polishing details
  • perf and PI scheduled
• r+-
• ni
• investigating 1560090 (defect/P1/critical): High refresh rate monitor (>200Hz) slows performance

Valentin

Bug 1638789 - Also log originAttributes in nsHostResolver::ResolveHost [landed]

Bug 1636891 - Move logging specifics from toolkit/library/rust/shared/lib.rs to the gecko_logger crate

Bug 1636888 - Allow specifying entire rust modules in MOZ_LOG

Bug 1616252 - Maybe add a /etc/hosts parser

Bug 1631748 - Fix COOP expected fails

Bug 1627800 - 404 http error isn't displayed in Firefox

Meeting 2020-05-14

Round table:

• what should we do with 404 response with no content ?
• 1604056 kernel panic on LineageOS
• HTTP/2 issue when toggling VPN state

Next on triage duty: Andy

Valentin

Triage / Outreachy

[Landed]

Bug 1572513 - Abort and BFCache handling for COEP headers

Bug 1614988 - COOP+COEP data URL tests timing out

Bug 1624090 - We should make possible to turn on neqo logging in about:networking

Bug 1635899 - We should cancel TRR::mTimeout in OnStopRequest.

Bug 1634462 - Check doh-rollout.mode in TRRService

Bug 1635566 - TRR: Perform a case-insensitive match for the host name

[Working on]

Bug 1636888 - Allow specifying entire rust modules in MOZ_LOG

#528 - Implement all qlog probes

Bug 1595848 - Intermittent COOP/COEP tests

Andy

Neqo (7, 32, 52)(-2,+1,+1):

• Project Plan 0.2
  • Q: What are conditions to re-enable in Nightly? A: Serious quic-network-sim (QNS) progress and no prio=high issues remaining
• #101 Can't xfer data with winquic - Reproduced with another QUIC impl. It's not us, it's them. (It's only HTTP 0.9 so not critical tbh.) Closed.
• Google interop: They report an issue when initial CID != 8 bytes. We do this. They estimate 3 weeks to fix & roll out. No temporary workaround on our side planned.
• Worked on neqo_glue changes for header being returned in an event.
• Reviewed many PRs from mt, dragana, & vonasek. Wow!

Next:

• Even more PRs to review!
• Work on glue changes for next Neqo->Gecko merge
• Maybe Persistent congestion interval start time

Michal

• 1630000 Hitting Back sometimes shows old cached content
• https://github.com/mozilla/neqo/pull/619
• https://github.com/mozilla/neqo/issues/621
• https://github.com/mozilla-necko/neqo-qns/pull/2
• Quic network-simulator
• 1635935 HTTP/2 issue when toggling VPN state
• 1637947 network ID: link status isn't updated in NetlinkService::OnAddrMessage()

mayhemer

• preload
  • landed: M1 and half of M2, still disabled
  • dogfood at will!
  • missing: fetch(), XHR, Link:
  • on track: perf and integrity testing
• landed: MOZ_LOG=events:1 and logan
• planned: MOZ_LOG=events to add IPC
• updated: our triage helper
• ni?, r?, here and there

Kershaw

Landed:

• Bug 1549323 - [socket-process] Check whether SSL_ClearSessionCache() is called on the socket process.
• Bug 1636999 - [socket process] handle the case that connection manager is used when socket process is not ready
• Bug 1636930 - [socket process] Make sure http response head can be serialized and deserialized properly

Ongoing:

• Bug 1539819 - [socket process] TRR is not working in socket process
• Bug 1637648 - [socket process] Forward observer notifications in nsHttpHandler to socket process
• Bug 1616014 - DNSListenerProxy::mListener should not be a main thread pointer holder

Dragana

Finishing HTTPSSVC design

Done

• PR 601 Close a connection if critical streams are reset or a stop_sending is received
• PR 598 - Improve table.rs
• PR 611 - Add an encoder's max_table_size.
• PR 612 - Remove total_num_of_inserts, use table.base() instead
• PR 614 - Refactor qpack errors
• PR 617 - Add a reference to a qpack dynamic table entry as soon as it is referred in a header block encoding
• PR 616 - The qpack encoder uses stream_send_atomic to send instructions

Next

• Issue 459 - qpack: Fix Header Block Prefix
• Issue 300 - Implement server push, client side
• Issue 417 - telemetry

Junior

• [land] Bug 1632045 - Content-Type and related headers not stripped for certain redirects when using XMLHttpRequest
• Bug 1636572 - [webvtt] The testcase fails if we allow sending onStartRequest via pHttpBackgroundChannel
• Bug 1633935 - Allow sending onStartRequest via pHttpBackgroundChannel
  • P2 let HttpChannelBackgroundChild know the sink events before OnStartRequest

Meeting 2020-05-07

Round table:

• New table of triage helper: Bugs awaiting for ni are shown in the new table.
  • [triage person] look at NIs older than 2 months and do something about them
  • [triage person] drop the necko-triaged tag, maybe starting in June
  • [Honza] fix problems with main table containing bugs that are NI'ed
  • [Honza] review the overlap with triage center and see if we need to maintain our own triage center
• 1569715 - caching preflight requests
• Necko bi-weekly updates

Next on triage duty: Valentin

Valentin

Bug 1623126 - Implement HTTPSSVC [landed]

Bug 1629882 - COEP requires a secure context [landed]

Bug 1635899 - We should cancel TRR::mTimeout in OnStopRequest [landed]

Bug 1624090 - We should make possible to turn on neqo logging in about:networking

Bug 1634462 - Check doh-rollout.mode in TRRService

Bug 1635566 - TRR cname resolution is case sensitive

Bug 1630439 - Issue opening url with underscore at the end on Android

• got to look at the chrome dns code

Outreachy intern starts on 19th of May

mayhemer

• preload:
  • just rebased all patches, pushed to try, landing when green, still disabled
  • the image patch has a leak, Edgar notified
  • PI request for 78 refiled
  • continuing the work, had a talk with Emilio about css coalescing, a little rework is on the way
  • left: SRI, XHR, fetch
• updated necko triage helper
  • do we still need [necko-triaged]?
• having a patch pending r for logging runnables in nsThread, pool, cache... with APIs we can add around our code to have more event tracking, it conveniently displays in logan (a distillation of backtrack in a well known tool)
• ni? f?

Kershaw

Landed:

Bug 1623380 - [socket process] Send data directly from the socket process to a content process

Ongoing:

• Triage
• Bug 1539819 - [socket process] TRR is not working in socket process
• Bug 1629457 - Intermittent netwerk/test/unit/test_trr.js | application crashed [@ __thiscall InitHash::InitHash + 0x12d3] | TRRServiceChannel using nsWeakReference OffMainThread
• Fix test failures for socket process

Michal

• 1631341 Random form body loss for concurrent HTTP/2.0 POST requests
• 1569715 CORS preflight requests are cached when 'Disable cache' is checked.
• 1621345 Support for nsIWebSocketEventListener.send() API
• https://github.com/mozilla/neqo/issues/595
• https://github.com/mozilla/neqo/pull/600
• https://github.com/mozilla/neqo/pull/604
• Quic network-simulator
• 1630000 Hitting Back sometimes shows old cached content

Junior

Bug 1633935 - Allow sending onStartRequest via pHttpBackgroundChannel

Bug 1633942 - introduce HttpChannelOnStartRequestArgs

[land] Bug 1582117 - test_prompt_http.html|test_prompt_async.html Fission failure: mCrossOrigin/Ci.nsIAuthInformation.CROSS_ORIGIN_SUB_RESOURCE isn't true

Dragana

• PR 576 - Write encoder instruction directly into a transport stream

• PR 598 - Improve table.rs

• PR 599 - Send DATA_BLOCKED and/or STREAM_DATA_BLOCKED if qpack cannot write complete instructions into a transport stream

• PR 601 - Close a connection if critical streams are reset or a stop_sending is received

• google issue with Firefox http3

• Some reviews (rust API for http request will use nsHttpChannel)

• Review HTTPSSVC draft

Andy

Neqo (9, 31, 53):

• Project Plan 0.2
  • Lean more heavily on GitHub for individual time estimations. Heavily using labels.
  • Update for current status (emphasis on interop, fixing bugs)
• #101 Can't xfer data with winquic - asked for assistance from MS devs
• #597 http://quic.aiortc.org/ works with neqo-client but not in the browser - cert issue
• Team is also working on Google interop issues
• #591 Prepare for sending less frequent ACKs [landed] - more work TBD in this area

Misc:

• Root-caused 1629982 SEC_ERROR_NO_TOKEN when using neqo-client (SSL_HkdfExtract) as distro-specific

Next:

• Review backlog of PRs
• Work on glue changes for next Neqo->Gecko merge
• Solicit initial feedback on revised Project Plan
• Resolve #101
• More ACK work

Meeting 2020-04-30

Round table:

• intermittent leaks (PSimpleChannelChild)
• dFPI bugs 1634080, 1634079
• change to triage process

Next on triage duty: Kershaw

Kershaw

Landed/Complete:

• Bug 1512478 - Have IPC for picking client certificate for authentication

• Project plan for proxy resolution in socket process https://docs.google.com/document/d/1BM-m1iUNSobiVu24BlRrLS35k6ZsiKHfCwnC5xVM2G8/edit?usp=sharing

Ongoing:

• Bug 1497249 - [socket-proc] Connection stickiness handling for upgrade observers
• Bug 1539819 - [socket process] TRR is not working in socket process
• Bug 1632881
• Bug 1607426 - ThreadSanitizer: data race [@ get] vs. [@ assign_assuming_AddRef] ([@ nsPACMan::ProcessPending] vs. [@ PACLoadComplete::Run])
• Bug 1629457 - Intermittent netwerk/test/unit/test_trr.js | application crashed [@ __thiscall InitHash::InitHash + 0x12d3] | TRRServiceChannel using nsWeakReference OffMainThread

Junior

OnStart/StopRequest OMT in child

Bug 1633942 - introduce HttpChannelOnStartRequestArgs

Bug 1633935 - Allow sending onStartRequest via pHttpBackgroundChannel

[landed] Bug 1632204 - Origin is null for an XHR request to the same origin when Referrer-Policy is no-referrer

Michal

• triage
• 1631341 Random form body loss for concurrent HTTP/2.0 POST requests
• Quic network-simulator
• https://github.com/mozilla/neqo/pull/588
• 1630000 Hitting Back sometimes shows old cached content

Andy

Neqo:

• Project (13, 25, 50): Triaged backlog (~20 issues). Now at 0 untriaged issues. All priority=high issues assigned owners. Please make sure to tag new issues into the "1H 2020 Triage" project.
• 0.3 Integration into Gecko: Found an issue in 0.3 leading to 0.3.1. Still seeing instability and variability in results on test sites. Holding off on integration until things stabilize.
• Working on sending fewer ack frames, implemented but not passing all tests yet
• Reviewed PRs
• Working on improving reporting/metrics on Neqo dev activity

DoH:

• Nothing this week

Misc:

• NSS 1629982 SEC_ERROR_NO_TOKEN when using neqo-client (SSL_HkdfExtract)
• 1 day PTO

Next:

• Revise Neqo Project Plan

mayhemer

preload:

• one patch still needs r+, postponing landing, too late now
• working on M2 bugs: different coalescing, SRI checking
• next: XHR, fetch

Valentin

Bug 1626057 - Implement TRR steering platform APIs [landed]

Bug 1633567 - Re-enabled AppCache [landed]

Bug 1572513 - Abort and BFCache handling for COEP headers

Bug 1623126 - Implement HTTPSSVC

Bug 1634462 - Check doh-rollout.mode in TRRService

Dragana

review of HTTPSSVC

PR 572 - Make HeaderReady hold headers

PR 579 - If there is no more data to read in read_response_data we should exit the loop

PR 573 - It should not be possible to create new streams in GoAway state.

PR 570 - A transaction should try to read the next frame when it is done reading a data frame

PR 567 - Remove DataReadable events when a stream is done

In progress:

PR 576 - Write encoder instruction directly into a transport stream

PR 587 - Enforce a correct stream id in a GOAWAY frame

Meeting 2020-04-23

Round table:

• necko-reviewers can be removed now
• Fetch from rust: https://phabricator.services.mozilla.com/D70257
• High refresh rate monitor (>200Hz) slows performance

Next on triage duty: Michal

mayhemer

possibly READONLY:

• preload:
  • need two simple test-change reviews from dragana
  • need two small reviews from baku
  • need one review of edgar's image loading patch
  • has a green try! --> ready to land
  • thinking: land now (disabled) and possibly open users with this flipped on to sec issues or wait for early 78? probably the latter

Dragana

Mostly neqo:

Issue 567 - Remove DataReadable events when a stream is done

Issue 563 - Use a tree as a helper structure for the huffman decoder.

Issue 547 - Turn on clippy pedantics for neqo-http3

Issue 570 - A transaction should try to read the next frame when it is done reading a data frame

Issue 569 - HeaderReady event should have headers as well

Andy

Neqo

• Bug triage and documenting things. New Neqo Eng Weekly meeting
• Preparing for Neqo 0.3.0 release: PR reviews; Tested Gecko with latest incl. API changes
• Finally landed qlog PR :tada:
• Some tiny PRs
• Google: Enabled H3 (for me) on YouTube, seeing it used there, but just a little

DoH

• Mtg with Markus the UX person and everyone. Selena wants more info on why users are choosing Disable when presented with doorhanger.
• Canary: Telemetry shows canary abuse by CenturyLink and now Cox, both large US ISPs. Peter is attempting to contact.

Next:

• Review & document: timers and recovery state machine
• Tag 0.3.0 and merge into m-c
• More interop/testing work

Kershaw

Landed/Complete:

• Bug 1626076 - Use DataStorage on socket process
• Bug 1485652 - Make nsNSSSocketInfo::IsAcceptableForHost work asynchronously on the socket process
• Bug 1624996 - Add a telemetry to see how often TRR request is redirected

Ongoing:

• Bug 1497249 - [socket-proc] Connection stickiness handling for upgrade observers
• Bug 1539819 - [socket process] TRR is not working in socket process
• Bug 1512478 - Have IPC for picking client certificate for authentication

Junior

[landed] Bug 1554218 - Handle 407 from a HTTP/2 proxy

[landed] Bug 1341689 - Missing Origin header when requests to the same-origin and crossorigin set

OnStart/StopRequest OMT in child

Bug 1632045 - Content-Type and related headers not stripped for certain redirects when using XMLHttpRequest

Valentin

reviews 🙂

Bug 1629882 - COEP requires a secure context

Bug 1631751 - Fix COEP expected fails

Bug 1631748 - Fix COOP expected fails

Bug 1572513 - Abort and BFCache handling for COEP headers

Bug 1623126 - Implement HTTPSSVC

Bug 1626057 - Implement TRR steering platform APIs

Bug 1624090 - We should make possible to turn on neqo logging in about:networking

Michal

• [landed] 1630899 Update NETWORK_PAC_URL_SCHEME probe to collect data on release
• [landed] 1624515 Fix data race on CacheStorageService::mShutdown
• 1630486 websocket binary message splitting
• 1630000 Hitting Back sometimes shows old cached content
• https://github.com/mozilla/neqo/issues/529
• 1631016 Find out why HTTP_SUB_COMPLETE_LOAD_V2 reports longer load times with every new release since beta 73

Meeting 2020-04-16

Round table:

• Moving proxy resolution to socket process. Can we detect proxy configuration changes?
• unassigned P1
• PAC scripts over FTP

Next on triage duty: Honza.

Andy

Triage

Neqo

• H3 disabled in Nightly due to Google* issues
• New Neqo Eng Weekly meeting
• interop container image issues
• qlog PR good to land (once reviewer comments addressed)
• Neqo breakage due to NSS changes
• Simplifying API work: eliminating process_http3(); eliminating process_timer()

DoH

• Canary: Telemetry shows canary abuse by CenturyLink. First response is to reach out and ask what's up.
• UI: We have a UI person, initial meeting with them on Tuesday

Next:

• Get more organized for more people on Neqo, more mgmt scrutiny
• Land qlog
• Resolve API issues
• More interop/testing work

Valentin

Bug 1595886 - [done]

Bug 1625749 - [done]

Bug 1625563 - Intermittent test_trr.js::test_resolve_not_confirmed [landed]

Bug 1619673 - Disable appcache in release [landed]

Bug 1623126 - Implement HTTPSSVC

Bug 1595848 - Intermittent COOP/COEP tests

Dragana

need-infos and reviews

neqo:

Issue 547 - Turn on clippy pedantics for neqo-http3

Issue 544 - Set the idle timeout to 30s

Issue 538 - Check if transaction is done(), do not check only fin before removing a transaction

Issue 537 - Fix debug assertion in encode_headr_block

Kershaw

Landed/Complete:

• rel=preload review

Ongoing:

• Bug 1497249 - [socket-proc] Connection stickiness handling for upgrade observers
• Bug 1626076 - Use DataStorage on socket process
• Bug 1539819 - [socket process] TRR is not working in socket process
• Bug 1485652 - Make nsNSSSocketInfo::IsAcceptableForHost work asynchronously on the socket process

Junior

Bug 1554218 - Handle 407 from a HTTP/2 proxy

OnStart/StopRequest OMT: image (good to OMT but not trivial), CSS (no)

Michal

• 1622335 Add telemetry for PAC scripts served over FTP
• Effect of cache partitioning
• Quic network-simulator

mayhemer

• preload
  • all M1 bugs but one patch r+
  • pushing to try
  • perf testing request when green (enough)
  • some bugs for M2 on the way (fetch, font: r?)
• Bug 1605895

Meeting 2020-04-09

Round table:

• OnStartRequest/OnStopRequest OMT
• Review queue
• DocumentChannel regression/failures
• Bug 1627407 - Support Save-Data: Request Header so users can save bandwidth
• Bug 1625262 - (from 800MB heap-unclassified) unreported 86MB from Http2Session::WriteSegmentsAgain

Next on triage duty: Andy

Valentin

Bug 1625213 - Refactor DNS code to remove OnLookupByTypeComplete [landed]

Bug 1623126 - Implement HTTPSSVC

Bug 1624090 - We should make possible to turn on neqo logging in about:networking

Bug 1572513 - Abort and BFCache handling for COEP headers

Bug 1595886 - ??

• Outreachy application time is over.
• #necko reviewer group bug 1628714
• AppCache plan - disable in 77. Keep pref in 78 (for ESR). Disable in 79

Kershaw

Landed:

• Bug 1568028 - [wfh] about:support doesn't report/show separate socket process in "Remote Processes" section
• Bug 1626743 - [socket process] Fix test test_content_sniffer.js
• Bug 1625151 - [socket process] Move DNS resolution to socket process
• Bug 1627654 - Padlock in URL bar is sometimes missing when a FPN session is active

Ongoing:

• Bug 1626076 - Use DataStorage on socket process
• Bug 1539819 - [socket process] TRR is not working in socket process
• External mentoring about socket process.

Michal

• [landed] 1627635 Websocket connects to DIRECT instead of Proxy-HTTP/HTTPS with SystemProxySettings on Linux
• [landed] 1622409 Put FTP code behind a pref
• [landed] https://github.com/mozilla/neqo/issues/533
• 1627910 Enable test netwerk/test/unit/test_http3
• https://github.com/mozilla/neqo/issues/505
• reviews (Outreachy)
• 1628660 netwerk/test/http3server fails to build on Tier3 platforms

Junior

triage

Bug 1554218 - Handle 407 from a HTTP/2 proxy

[land] Bug 1601671 - websocket connects to DIRECT instead of Proxy-HTTP/HTTPS with SystemProxySettings in Windows

next:

OnStart/StopRequest OMT

Bug 1341689 - Missing Origin header when requests to the same-origin and crossorigin set

mayhemer

• preload
  • milestone 1: 1 patch r?, 1 patch delegated, 1 simple patch in-progress, all other r+
  • we have automated tests: wpt and cover for the current implementation
  • answer from yoav
• reviews, ni?
• some mentoring
• Tokenizer::CheckString("foo bar")
• landed 1626809 - disk-storage memory reporter does not properly anonymize
• landed 1627572 - Intermittent tsan: ThreadSanitizer: data race /builds/worker/checkouts/gecko/xpcom/io/FilePreferences.cpp:272:7 in mozilla::FilePreferences::IsAllowedPath(nsTSubstring const&)
• investigating bug 1605895 with a life diagnostic patch, needs update

Andy

Neqo

• Posted new version of qlog PR for review, using new streaming log API. There will be further work to add probes in the code.
• PR reviews

Next:

• qlog: Address any additional review comments
• Some Neqo bugs need more investigation
• Blog post on H/3 in Nightly
• Triage

Dragana

• HTTP/3 is turned on, looks good
• QPACK is turned off for now
• Bug 1628008 - HTTP/0.9 Shoutcast stopped working in v76.0b1
• Bug 1618158

neqo:

• Issue 525 - We can still have RecvStreamReadable after a stream has been read completely
• Issue 520 - Turn on clippy:pedantic for neqo-qpack

Working on:

• Issue 159 - Fold process_timer() into process_output()
• Issue 276 - Do not increase congestion window if under utilized

Meeting 2020-04-02

Round table:

• Remove appcache (meta bug: 1584984)
  • performance bug?
  • Bug 1626026 - Appcache regression in Firefox 74
• bug 1586630
• FTP deprecation: addon support concern (see doc in bug 1626365)
• controlling TRR prefs

Next on triage duty: @JuniorHsu

Valentin

[Outreachy][reviews]

Bug 1623807 - FQDN DoH queries are encoded incorrectly [landed]

Bug 1626057 - steering heuristics

Bug 1625213 - Refactor DNS code to remove OnLookupByTypeComplete

Bug 1623126 - Implement HTTPSSVC

Bug 1572513 - Abort and BFCache handling for COEP headers

Michal

• [landed] 1625284 Increase receive buffer size of QUIC socket
• [landed] 1587353 Add tests for http3
• [landed] 1622335 Add telemetry for PAC scripts served over FTP
• 1622409 Put FTP code behind a pref
• 1601671 websocket connects to DIRECT instead of Proxy-HTTP/HTTPS with SystemProxySettings
• try to enable http3 tests
• measure QUIC throughput

Junior

Project Plan: OMT OnStartRequest/OnStopRequest on child

Bug 1554218 - Handle 407 from a HTTP/2 proxy

Dragana

Http3 testing

Bug 1626799 - Update Neqo to 0.2.3

Bug 1626359 - Crash in [@ mozilla::net::Http3Stream::WriteSegments]

Bug 1625841 - Crash in [@ mozilla::net::Http3Session::Authenticated]

Bug 1625198 - Number of bytes read reported n HttpConnectionUDP::OnSocketReadable is bogus

mayhemer

• preload
  • done css and image
  • image needs more care, given to Edgar Chan
  • giving font to Cameron
  • restarted comm with Yoav and Addy
  • running WPT tests for preload
• external and internal mentoring

Andy

Neqo

• Reviving qlog PR: Rebased & coalesced commits

DoH

• Minimal UI rework doc revised 2x

Next:

• qlog: Address review comments, move to new streaming API in qlog 0.2
• Fix more Neqo bugs
• More testing of H3 in Gecko

Meeting 2020-03-26

Round table:

• Can we turn on rust logging from about:networking? Bug 1624090
• AsyncOpen OMT

Next on triage duty: Dragana

Dragana

Http3/Quic bugs:

Issue 47 - Timers that end up being set in the past

Bug 1618158

Bug 1623501 - HTTP3 - Websocket - Firefox can’t establish a connection to the server at wss://web.whatsapp.com/ws - Whatsapp

Bug 1616323 - H3: Cached H3 resources show "HTTP/3.0" instead of "HTTP/3" in webtools Version field

Looking at AltSvc with HTTP/3

Looking into moving the DoH heuristics into platform

It is IETF virtual meeting week

Valentin

[Outreachy][reviews][triage]

steering heuristics

• impl more complicated than initially expected
• will pass back to Nihanth

Bug 1623807 - FQDN DoH queries are encoded incorrectly

Bug 1595886 - Use temp string

Bug 1625213 - Refactor DNS code to remove OnLookupByTypeComplete

Bug 1623126 - Implement HTTPSSVC

Bug 1614962 - Intermittent netwerk/test/unit/test_trr.js

Michal

• [landed] 1624105 We don't try to activate Http3Stream again if it failed once
• 1587353 Add tests for http3
• 1622335 Add telemetry for PAC scripts served over FTP
• Http3 tests intermittent failures
• 1622409 Put FTP code behind a pref

Kershaw

Landed:

• Bug 1607221 - ThreadSanitizer: data race [@ mozilla::net::nsSocketTransport::OnSocketEvent] vs. [@ SetTimeout]

Ongoing:

• Bug 1623380 - [socket process] Send data directly from the socket process to a content process
• Bug 1539819 - [socket process] TRR is not working in socket process
• Bug 1625151 - [socket process] Move DNS resolution to socket process
• Bug 1485652 - Make nsNSSSocketInfo::IsAcceptableForHost work asynchronously on the socket process

Junior

OMT OnStartRequest/OnStopRequest

Bug 1570889 - blob URLs and CSP sandbox'ed pages should inherit Cross-Origin-Opener-Policy

Bug 1601671 - websocket connects to DIRECT instead of Proxy-HTTP/HTTPS with SystemProxySettings in Windows

Andy

Neqo

• 1624800 panic when indexing into CryptoStreams fix in review
• Some Outreachy interactions
• PR/Phab reviews
• Our Docker image (repo here) for QUIC Network Sim (QNS) now working for http3 test in client role. TODO: Enable other tests, and server role.

DoH

• Minimal UI rework doc

Next:

• Fix more Neqo bugs
• Refine doc & reach consensus within our team on Minimal UI plan
• More testing of H3 in Gecko

mayhemer

• preload: continuation
• reviews

Meeting 2020-03-19

Round table:

• potential breaking changes in next few releases

Next on triage duty: Valentin

Kershaw

Landed:

• Bug 1619584 - TRR: Support redirect for TRR request channel
• Bug 1607257 - Enable SSLTokenCache on nightly
• Bug 1621601 - Assertion failure: PR_GetEnv("MOZ_SANDBOXED") != nullptr
• Bug 1527384 - [socket process] Make SpeculativeConnect work
• Bug 1620824 - Tabs do not load after restart with network.trr.fetch_off_main_thread=true and under specific condition

Ongoing:

• Triage
• Bug 1623380 - [socket process] Send data directly from the socket process to a content process

Valentin

[outreachy]

Bug 1617889 - Fix COOP comparison

Bug 1618042 - early DNS lookups fail with network.trr.mode=3 using network.trr.uri with hostname in it

Bug 1621877 - Add more unit tests for setting an empty host on URLTYPE_AUTHORITY URLs

Next:

Bug 1623126 - Implement draft-ietf-dnsop-svcb-httpssvc + DNS refactoring

Junior

Bug 1272302 - navigator.sendBeacon doesn't set Origin header for same-origin request

Bug 1570889 - blob URLs and CSP sandbox'ed pages should inherit Cross-Origin-Opener-Policy

Andy

Neqo

• Another serious bug fixed. Neqo 0.2.2 tagged and merged into m-c
• Trying to help some Outreachy people get started

next:

• Fix more Neqo bugs
• Review Dragana's QUIC/H3 project plan
• Review Valentin's Comcast DoH doc
• More testing of H3 in Gecko
• More quic-network-interop

Michal

• 1587353 Add tests for http3 next

• 1622335 Add telemetry for PAC scripts served over FTP

• 1622409 Put FTP code behind a pref

mayhemer

• 5 rel=preload bugs now have first patches
• reviews
• here and there

next:

• file sec review request for rel=preload
• schedule meeting with Olli for next steps in DOM code

Dragana

Bug 1623312 We should support slow read in Http3

Bug 1622861 Improve Http3 closing

Bug to be filed - Parsing ReponseHead from cache is wrong

Http3 testing in Firefox

Meeting 2020-03-12

Round table:

• 1608087 - Assertion failure
• valentin: If you know of any code that can use some light refactoring, please add it to the spreadsheet. I'll mentor it.
• 1620708 - cookies/samesite/iframe-reload.https.html.ini WPT FAIL with Fission
• 1620713 - offline/appcache/appcache-iframe.https.html.ini WPT FAIL or TIMEOUT with Fission
• 1620750 - ws-over-h2
• ftp resources, PAC scripts

Next on triage duty: kershaw

Valentin

ReadOnly:

outreachy mentoring

Bug 1611651 - TRR: Infinite loop when visiting https://1.1.1.1/help

Bug 1618042 - early DNS lookups fail with network.trr.mode=3 using network.trr.uri with hostname in it

Bug 1621877 - Add more unit tests for setting an empty host on URLTYPE_AUTHORITY URLs

Bug 1617889 - Fix COOP comparison

Michal

• [landed] 1470571 Sync with cache IO thread before checking alt-data flag in index
• [landed] https://github.com/mozilla/neqo/issues/475
• 1619520 Abrupt Websocket disconnect
• 1619618 sMmapAccessScope; error: initializer for thread-local variable must be a constant expression
• 1616979 Nightly - Firefox caching too much - Not respecting the "CTRL+ Shift +R" shortcut or the "Disable Cache" checkbox
• 1587353 Add tests for http3

mayhemer

• DONE preload plan
  • only left: file sec review req
• DONE 1616082 - Crash in [@ shutdownhang | nsThread::Shutdown | mozilla::net::WaitForThreadShutdown::Run | nsThread::Shutdown | ProfileResetCleanupResultTask::Run]
  • not a regression
• IN-PROG 1611081 (defect/P1/critical): Blank content area when reloading multipart/x-mixed-replace page with developer tools pane open
  • recent dev tools doc channel regression
• IN-PROG ni+
• ? Priority of h2 priories rewrite