1638814 - Allow fetch() and XHR in workers use a <link preload> response

Status: NEW

(Core :: DOM: Networking, enhancement, P3)

Comment 1

[Honza Bambas (:mayhemer)]

https://phabricator.services.mozilla.com/D74899#inline-439234

Comment 2

[Honza Bambas (:mayhemer)]

https://phabricator.services.mozilla.com/D75557#inline-439509

Comment 3

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

(In reply to Honza Bambas (:mayhemer) from comment #1)

https://phabricator.services.mozilla.com/D74899#inline-439234

Andrew, do you know if this is something we need, and how hard it's to implement?

Comment 4

[Andrew Sutherland [:asuth] (he/him)]

(In reply to Valentin Gosu [:valentin] (he/him) from comment #3)

(In reply to Honza Bambas (:mayhemer) from comment #1)

https://phabricator.services.mozilla.com/D74899#inline-439234

Andrew, do you know if this is something we need, and how hard it's to implement?

The anchor link doesn't seem to work for me, I don't know what the question is.

Comment 5

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

https://phabricator.services.mozilla.com/D74899#inline-439234

Comment here

baku:
If you do this it means that fetch() from workers cannot use the preload data.
Readying the spec, this should be supported. Maybe this can be done in a follow up because the fix is not trivial.
We need to store the Preload data in a shareable place between workers and window.

https://phabricator.services.mozilla.com/D75557#inline-439509
and here

baku:
Here we have exactly the same problem as in fetch() for workers. It should be fixed too

This is a follow-up for workers reusing preload data.

Comment 6

[Andrew Sutherland [:asuth] (he/him)]

Thank you for excerpting the questions; I know I had changed the default phabricator setting so that it only shows inline comments made explicitly on the current revision in the code listings because the various bots were ending posting N copies of inline comments... I guess the anchors must be to those despite the comments seeming to exist in the transaction log?

My understanding of the current state of things is that Preload is not (sufficiently) specified in terms of fetch, so it's not actually clear what preload should be doing as it relates to workers. (Well, per spec, I guess workers shouldn't interact with preload, but it seems like there was some intent that it should and maybe Blink does it at some level?) I expect if we were looking to implement this, we'd first want to help out with the spec work. The key issues as I understand them are:

• Specify the preload cache: preload spec issue, fetch spec issue
• "Make preload headers work for fetches created by workers" preload issue

Comment 7

[Anne (:annevk)]

Do we know that Chrome supports this? And if so, for what worker types? (When they are same-process perhaps?)

Comment 8

[Andrew Sutherland [:asuth] (he/him)]

We know not.

Comment 9

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

Thanks!