1219085 - Import the fetch-request-redirect.https.html test from Blink

Status: RESOLVED FIXED

(Core :: DOM: Service Workers, defect)

Comment 1

[(no longer active)]

Attachment: Import the fetch-request-redirect.https.html test from Blink

Comment 2

[Josh Matthews [:jdm]]

Comment on attachment 8679761 [details] [diff] [review]
Import the fetch-request-redirect.https.html test from Blink

Review of attachment 8679761 [details] [diff] [review]:
-----------------------------------------------------------------

::: testing/web-platform/mozilla/tests/service-workers/service-worker/resources/fetch-rewrite-worker.js
@@ +16,5 @@
>    var init = {};
>    init['method'] = params['method'] || base['method'];
>    init['mode'] = params['mode'] || base['mode'];
>    init['credentials'] = params['credentials'] || base['credentials'];
> +  init['redirect'] = params['redirect-mode'] || base['redirect'];

This seems like we should be rewriting the test, rather than changing this.

Comment 3

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/71cb0cec8da0

Comment 4

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

This broke wpt(4): https://treeherder.mozilla.org/logviewer.html#?job_id=16417818&repo=mozilla-inbound

Backed out in https://hg.mozilla.org/integration/mozilla-inbound/rev/6b66dba1086c

Comment 5

[(no longer active)]

I can reproduce the timeout in an opt build but not on a debug build.  This test sets up 2 iframes and waits for 10 seconds to wait for an asynchronous operation.  RyanVM said on IRC that the default timeout for these tests is 20s, but eyeballing the test run in an opt build, the test times out in clearly less than 20 seconds.  Furthermore, if I run it with --timeout-multiplier 2, it passes, which seems to suggest to me that the default timeout values in debug vs opt runs is actually different!

James, what is the default timeout value?  And what's the equivalent of SimpleTest.requestLongerTimeout() in WPT?  Thanks!

Comment 6

[(no longer active)]

I also tried adding |<meta name="timeout" content="long">| to the test to no avail.

Comment 7

[James Graham [:jgraham]]

The default timeout is 10s in opt builds. Debug builds have a default timeout multiplier of three, chosen because tests seems to be mostly stable with that value, but not with 2.

<meta name="timeout" content="long"> before <script src=/resources/testharness.js> is the right way to update the timeout value, but because the harness also sets a timeout you need to make sure that the metadata is updated after you add this to the file; run with --manifest-update to do the update.

Comment 8

[(no longer active)]

Thanks a lot, James!

Comment 9

[Ben Kelly [:bkelly, not reviewing]]

Attachment: Bug-1219085---Import-the-fetch-request-redirecthtt.patch

Updated with long timeout in manifest.

Comment 10

[Ben Kelly [:bkelly, not reviewing]]

The e10s path is quite racy and we're unlikely to fix it in the short term.  Lets see if we can land this test for non-e10s only.

  https://treeherder.mozilla.org/#/jobs?repo=try&revision=f5dc82e18882

Comment 11

[Ben Kelly [:bkelly, not reviewing]]

I forgot to hg add the wpt .ini file in that last try push, but the test clearly passes in non-e10s.  I'm just going to land with the ini.

Comment 12

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/6d2641e05d2b

Comment 13

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/6d2641e05d2b