1865394 - Verify dev tools display of priorities; urgency / incremental

Status: RESOLVED FIXED

(Core :: Networking: HTTP, task, P3)

Description

[Andrew Creskey [:acreskey]]

In the excellent talk, Resource Loading at the Cutting Edge, Robin Marx shows Firefox dev tools displaying incorrect priorities relative to the request urgencies.

The bug is to verify if this is the case and fix.

Related, dev tools does not appear to show the urgency/incremental headers.

Comment 1

[Andrew Creskey [:acreskey]]

We can move this to dev tools, but I'd like to take a look first and verify.

Comment 2

[Andrew Creskey [:acreskey]]

I believe we are using the PRIORITY_UPDATE frame for HTTP/3 rather than the request header
(although this may change in Bug 1865040)

But we should verify that the class of service flags for the channels are correctly reflected in devTools and also propagated to neqo.

Comment 3

[Andrew Creskey [:acreskey]]

Valentin and DevTools team shed some light on this on:

DevTools is displaying the priority from nsISupportsPriority
https://searchfox.org/mozilla-central/rev/1a63fee6c2a8ea6846f6f16b85edeb6d19fc9f23/devtools/shared/network-observer/NetworkUtils.sys.mjs#196

But Necko also has a notion of 'ClassOfService'
https://searchfox.org/mozilla-central/rev/1a63fee6c2a8ea6846f6f16b85edeb6d19fc9f23/netwerk/base/nsIClassOfService.idl#23

To resolve this, Necko will provide a singular API to DevTools that unifies these concepts.

Comment 4

[Ed Guloien [:edgul]]

Depending on what we want to do exactly, this could be a dupe, noting as see also for now.

Comment 5

[Dylan]

Hey Valentin, based on how robin marx shows that the devtools priority is not aligned with the urgency (u), would a possible approach be to map the mPriorityUrgency value, set here , to a priority value highest,high...lowest, then expose that to nsISupportsPriority.idl?

Comment 6

[Valentin Gosu [:valentin] (he/him)]

(In reply to Andrew Creskey [:acreskey] from comment #0)

Related, dev tools does not appear to show the urgency/incremental headers.

That's because we don't appear to set them.

(In reply to Dylan from comment #5)

Hey Valentin, based on how robin marx shows that the devtools priority is not aligned with the urgency (u), would a possible approach be to map the mPriorityUrgency value, set here , to a priority value highest,high...lowest, then expose that to nsISupportsPriority.idl?

I also think a problem here is that we don't yet fully implement fetchpriority (bug 1797715) - so the priorities won't match what we expect.

Comment 7

[Valentin Gosu [:valentin] (he/him)]

would a possible approach be to map the mPriorityUrgency value, set here , to a priority value highest,high...lowest

We already set the priority to match the CoS sometimes here and other times it's completely separate from it
https://searchfox.org/mozilla-central/search?q=nsCOMPtr%3CnsISupportsPriority%3E&path=&case=false&regexp=false

Comment 8

[Dylan]

Attachment: Bug 1865394 - added urgency and incremental to request header in netmonitor. r=valentin

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/d1d9d57a9895
added urgency and incremental to request header in netmonitor. r=necko-reviewers,extension-reviewers,devtools-reviewers,valentin,robwu,bomsy

Comment 10

[Atila Butkovits]

Backed out for causing failures at browser_net_har_copy_all_as_har.js.

Backout link: https://hg.mozilla.org/integration/autoland/rev/d5bc130229070c4e63e755762fcafc37d4ecfcf6

Push with failures: https://treeherder.mozilla.org/jobs?repo=autoland&selectedTaskRun=JHOhLBVxSgqf7JGducjF7g.0&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&revision=d1d9d57a98952e87c1c6bb15f17edc476be040de

Failure log: https://treeherder.mozilla.org/logviewer?job_id=450647075&repo=autoland&lineNumber=4136

Comment 11

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/73f0a43c3ec3
added urgency and incremental to request header in netmonitor. r=necko-reviewers,extension-reviewers,devtools-reviewers,valentin,robwu,bomsy,webdriver-reviewers,whimboo

Comment 12

[Sandor Molnar[:smolnar]]

https://hg.mozilla.org/mozilla-central/rev/73f0a43c3ec3

Comment 13

[Dylan]

Still more work to do