[meta] Implement 103 Early Hints
Categories
(Core :: Networking: HTTP, enhancement, P3)
Tracking
()
Tracking | Status | |
---|---|---|
firefox58 | --- | affected |
People
(Reporter: dragana, Unassigned)
References
(Depends on 14 open bugs)
Details
(Keywords: meta, Whiteboard: [necko-triaged])
Attachments
(3 files)
20.80 KB,
patch
|
Details | Diff | Splinter Review | |
10.98 KB,
patch
|
Details | Diff | Splinter Review | |
26.10 KB,
patch
|
Details | Diff | Splinter Review |
Updated•6 years ago
|
Reporter | ||
Comment 1•6 years ago
|
||
I have patches for necko part of this bug. We should work on the dom part of the implementation in bug 1409290.
Reporter | ||
Comment 2•6 years ago
|
||
Change to httpd.js test server and a test for the 103 response.
Reporter | ||
Comment 3•6 years ago
|
||
The change to nsHttpTransaction.
Reporter | ||
Comment 4•6 years ago
|
||
Propagate early hint response from parent to the child. I have implemented a new interface nsIEarlyHintObserver instead of extended nsIRequestObserver. It is possible to extend nsIRequestObserver as well, this only depends on how many observer we will really have and that depends on how we implement bug 1409290. Maybe we will need to propagate early hint responses through the listener chain(in that case we should extend the nsIRequestObserver interface).
Comment hidden (me-too) |
Reporter | ||
Updated•5 years ago
|
Comment 6•4 years ago
|
||
Any news on this?
I expect that website owners will only start to use early hints once a major browser supports it, but server-side support has been added in several servers/frameworks, for example:
- Apache HTTPD: https://httpd.apache.org/docs/2.4/howto/http2.html#earlyhints
- Puma & Rails (Ruby): https://eileencodes.com/posts/http2-early-hints/
Comment 9•2 years ago
|
||
Some movement here on the chrome side, intent to prototype:
https://groups.google.com/a/chromium.org/g/blink-dev/c/DAgWIczGtG0/m/gSXvjYn-AwAJ
From this design:
https://docs.google.com/document/d/1gCh_CnfrJq_VL7aGoq6skc7sn4yn5pKsM0gkHe5B9go/edit#heading=h.yh1rfx1p5ocs
Dragana, not quite sure if the proposed use of 103 reponse for a chrome trial is a great idea. Any thoughts?
Reporter | ||
Comment 10•2 years ago
|
||
(In reply to Benjamin De Kosnik [:bdekoz] from comment #9)
Some movement here on the chrome side, intent to prototype:
https://groups.google.com/a/chromium.org/g/blink-dev/c/DAgWIczGtG0/m/gSXvjYn-AwAJFrom this design:
https://docs.google.com/document/d/1gCh_CnfrJq_VL7aGoq6skc7sn4yn5pKsM0gkHe5B9go/edit#heading=h.yh1rfx1p5ocsDragana, not quite sure if the proposed use of 103 reponse for a chrome trial is a great idea. Any thoughts?
This is on my radar already. The document shows interesting possible performance improvement from the telemetry they have collected (it sounds also a bit high). There is not a good spec for this feature, e.g. integration with fetch spec is missing. In one of the docs they proposed 2 possible implementation, but I would prefer to have some spec or at lease a draft that we can discuss. Anne had similar comment.
I was thinking about adding telemetry to estimate possible performance improvement in Firefox (if you are interested to help :) ).
Updated•2 years ago
|
Comment 11•2 years ago
|
||
Is this picking up steam now here due to Cloudflare enabling beta Early Hints on accounts?
Reporter | ||
Updated•1 year ago
|
Updated•10 months ago
|
Comment 12•10 months ago
|
||
This is live in Cloudflare and out of beta, I just toggled it on for my website.
Updated•8 months ago
|
Description
•