Bug 174980 – [Fetch API] Implement abortable fetch

• Status: RESOLVED
• Resolution: FIXED
• Component: WebCore Misc.

[Jake Archibald]

This includes AbortController & AbortSignal https://dom.spec.whatwg.org/#aborting-ongoing-activities.

Fetch abort tests are at https://github.com/w3c/web-platform-tests/pull/6484.

Fetch abort spec is in progress https://github.com/whatwg/fetch/pull/523.

Mozilla intends to start implementing next month, using the tests rather than the spec.

[Antonio Torres]

Is there a roadmap for these features? I cannot find them in Webkit Status, and Firefox and Chrome already have support for aborts.

[molsson]

It's a strange that:

* caniuse.com claims that Safari supports AbortController
* window.AbortController is defined in Safari

...but a simple testcase like this:

http://temp.minimum.se/abort-native.html

Shows that Safari 12.0.1 sends the request even if controller.abort() has been called, i.e. only some "placeholder AbortController" has been added in Safari. It is not working at all.

[Alexandre C]

I confirm what molsson mentioned. 
The abort on fetch request on Safari 12.0.1 doesn't work at all.

[Sergey Rubanov]

This is really sad that only Webkit is missing Abortable Fetch since April 2018

[Radar WebKit Bug Importer]

<rdar://problem/46861402>

[youenn fablet]

Created attachment 357950 [details]
Patch

[youenn fablet]

Created attachment 357951 [details]
Patch

[youenn fablet]

Created attachment 357952 [details]
Patch

[youenn fablet]

Created attachment 357959 [details]
Patch

[EWS Watchlist]

Comment on attachment 357959 [details]
Patch

Attachment 357959 [details] did not pass ios-sim-ews (ios-simulator-wk2):
Output: https://webkit-queues.webkit.org/results/10509229

New failing tests:
imported/w3c/web-platform-tests/fetch/api/abort/general-serviceworker.https.html

[EWS Watchlist]

Created attachment 357966 [details]
Archive of layout-test-results from ews126 for ios-simulator-wk2

The attached test failures were seen while running run-webkit-tests on the ios-sim-ews.
Bot: ews126  Port: ios-simulator-wk2  Platform: Mac OS X 10.13.6

[youenn fablet]

Created attachment 358267 [details]
Patch

[EWS Watchlist]

Comment on attachment 358267 [details]
Patch

Attachment 358267 [details] did not pass ios-sim-ews (ios-simulator-wk2):
Output: https://webkit-queues.webkit.org/results/10621118

New failing tests:
imported/w3c/web-platform-tests/fetch/api/abort/general-serviceworker.https.html

[EWS Watchlist]

Created attachment 358291 [details]
Archive of layout-test-results from ews122 for ios-simulator-wk2

The attached test failures were seen while running run-webkit-tests on the ios-sim-ews.
Bot: ews122  Port: ios-simulator-wk2  Platform: Mac OS X 10.13.6

[youenn fablet]

Comment on attachment 358267 [details]
Patch

Failing test is due to SSL connection issue which seems to happen on iOS in some cases. This does not seem related with the changes but may make the above test flaky on iOS.

[Chris Dumez]

Comment on attachment 358267 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=358267&action=review

> Source/WebCore/Modules/fetch/FetchBodyOwner.h:130
> +    struct LoadingError {

Can you really have both a ResourceError and an exception at the same time. If not, could we use a Variant instead?

[youenn fablet]

(In reply to Chris Dumez from comment #16)
> Comment on attachment 358267 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=358267&action=review
> 
> > Source/WebCore/Modules/fetch/FetchBodyOwner.h:130
> > +    struct LoadingError {
> 
> Can you really have both a ResourceError and an exception at the same time.
> If not, could we use a Variant instead?

I hesitated on this one.
I think we can only have one.
I went with a struct instead of a Variant as I thought it makes the accessors to resource error/exception simpler and easier to use.
I'll do the Variant experiment.

[youenn fablet]

Created attachment 358340 [details]
Patch for landing

[youenn fablet]

Created attachment 358342 [details]
Patch for landing

[youenn fablet]

(In reply to youenn fablet from comment #19)
> Created attachment 358342 [details]
> Patch for landing

With Variant

[EWS Watchlist]

Comment on attachment 358342 [details]
Patch for landing

Attachment 358342 [details] did not pass ios-sim-ews (ios-simulator-wk2):
Output: https://webkit-queues.webkit.org/results/10631023

New failing tests:
imported/w3c/web-platform-tests/fetch/api/abort/general-serviceworker.https.html

[EWS Watchlist]

Created attachment 358354 [details]
Archive of layout-test-results from ews124 for ios-simulator-wk2

The attached test failures were seen while running run-webkit-tests on the ios-sim-ews.
Bot: ews124  Port: ios-simulator-wk2  Platform: Mac OS X 10.13.6

[youenn fablet]

Created attachment 358374 [details]
Patch for landing

[youenn fablet]

(In reply to youenn fablet from comment #23)
> Created attachment 358374 [details]
> Patch for landing

Marking imported/w3c/web-platform-tests/fetch/api/abort/general-serviceworker.https.html as flaky on iOS

[WebKit Commit Bot]

The commit-queue encountered the following flaky tests while processing attachment 358374 [details]:

fetch/fetch-worker-crash.html bug 187257 (author: [e-mail])
The commit-queue is continuing to process your patch.

[WebKit Commit Bot]

The commit-queue encountered the following flaky tests while processing attachment 358374 [details]:

http/wpt/css/css-animations/start-animation-001.html bug 190903 (authors: [e-mail], [e-mail], and [e-mail])
The commit-queue is continuing to process your patch.

[WebKit Commit Bot]

The commit-queue encountered the following flaky tests while processing attachment 358374 [details]:

http/wpt/css/css-animations/start-animation-001.html bug 190903 (authors: [e-mail], [e-mail], and [e-mail])
The commit-queue is continuing to process your patch.

[WebKit Commit Bot]

Comment on attachment 358374 [details]
Patch for landing

Clearing flags on attachment: 358374

Committed r239644: <https://trac.webkit.org/changeset/239644>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.