Closed Bug 1331899 Opened 7 years ago Closed 7 years ago

Update web-platform-tests to d74108d5e0d21d1f815e2eb4dac7db056b2a89ca

Categories

(Testing :: web-platform-tests, defect)

defect
Not set
normal

Tracking

(firefox53 fixed)

RESOLVED FIXED
mozilla53
Tracking Status
firefox53 --- fixed

People

(Reporter: jgraham, Unassigned)

References

Details

Attachments

(11 files)

59 bytes, text/x-review-board-request
Ms2ger
: review+
Details
59 bytes, text/x-review-board-request
Ms2ger
: review+
Details
59 bytes, text/x-review-board-request
Ms2ger
: review+
Details
59 bytes, text/x-review-board-request
Ms2ger
: review+
Details
59 bytes, text/x-review-board-request
gps
: review+
Details
59 bytes, text/x-review-board-request
Details
59 bytes, text/x-review-board-request
Details
59 bytes, text/x-review-board-request
Ms2ger
: review+
Details
59 bytes, text/x-review-board-request
Ms2ger
: review+
Details
59 bytes, text/x-review-board-request
Details
59 bytes, text/x-review-board-request
emorley
: review+
Details
      No description provided.
Comment on attachment 8827857 [details]
Bug 1331899 - wptrunner changes to support v4 manifests,

https://reviewboard.mozilla.org/r/105460/#review106270

::: testing/web-platform/harness/wptrunner/metadata.py:303
(Diff revision 1)
>      id_test_map = {}
> -    exclude_types = frozenset(["stub", "helper", "manual"])
> -    include_types = set(manifest.item_types) - exclude_types
> -    for test_path, tests in test_manifest.itertypes(*include_types):
> +    exclude_types = frozenset(["stub", "helper", "manual", "support", "conformancechecker"])
> +    all_types = [item.item_type for item in manifestitem.__dict__.itervalues()
> +                 if type(item) == type and
> +                 issubclass(item, manifestitem.ManifestItem) and
> +                 item.item_type is not None]

This is... clever. Can we do something less clever? `manifest.item_types` seems like something that would be more obvious. Why does that no longer work?

::: testing/web-platform/harness/wptrunner/update/sync.py
(Diff revision 1)
>  
>      def create(self, state):
>          from manifest import manifest
>          state.manifest_path = os.path.join(state.metadata_path, "MANIFEST.json")
> -        # Conservatively always rebuild the manifest when doing a sync
> +        state.test_manifest = manifest.Manifest("/")
> -        state.old_manifest = manifest.load(state.tests_path, state.manifest_path)

This seems to be used in `CheckoutBranch` in `update/upstream.py`. Why can we remove it?

Oh, you're fixing that in the next commit. That's not great, though I suppose none of these commits work in isolation.
Attachment #8827857 - Flags: review?(Ms2ger) → review+
Comment on attachment 8827858 [details]
Bug 1331899 - Update wpt sync for manifest changes,

https://reviewboard.mozilla.org/r/105462/#review106276
Attachment #8827858 - Flags: review?(Ms2ger) → review+
Comment on attachment 8827859 [details]
Bug 1331899 - Fix rename handling in wpt sync process,

https://reviewboard.mozilla.org/r/105464/#review106278
Attachment #8827859 - Flags: review?(Ms2ger) → review+
Comment on attachment 8827860 [details]
Bug 1331899 - Fix wpt manifest iterator to return the test path,

https://reviewboard.mozilla.org/r/105466/#review106282
Attachment #8827860 - Flags: review?(Ms2ger) → review+
Comment on attachment 8827864 [details]
Bug 1331899 - Stop navigation timing test loading a random page from the internet,

https://reviewboard.mozilla.org/r/105478/#review106284

Make sure this lands as a move, not rm + create.
Attachment #8827864 - Flags: review?(Ms2ger) → review+
Comment on attachment 8827865 [details]
Bug 1331899 - Ensure that asserts in referer-policy tests occur in a test(),

https://reviewboard.mozilla.org/r/105480/#review106290
Attachment #8827865 - Flags: review?(Ms2ger) → review+
Comment on attachment 8827867 [details]
Bug 1331899 - Update fetchlogs script to work with taskcluster,

https://reviewboard.mozilla.org/r/105484/#review106322
Attachment #8827867 - Flags: review?(emorley) → review+
Comment on attachment 8827859 [details]
Bug 1331899 - Fix rename handling in wpt sync process,

https://reviewboard.mozilla.org/r/105464/#review106530

Interesting code you have here.

You may be interested in the code I've written in bug 1322769, which can convert a Git repo to Mercurial with directory renames and while preserving file copy/rename detection.
Comment on attachment 8827861 [details]
Bug 1331899 - Fix mozbuild for wpt manifest v4 changes,

https://reviewboard.mozilla.org/r/105468/#review106532
Attachment #8827861 - Flags: review?(gps) → review+
Comment on attachment 8827859 [details]
Bug 1331899 - Fix rename handling in wpt sync process,

https://reviewboard.mozilla.org/r/105464/#review106530

Yeah, this was a pretty similistic approach, but this specific code is trying to do the opposite i.e. pick out parts of specific Mozilla changesets (those affecting the wpt directory) and convert them into changes to apply to an upstream git repo.

It might well make more sense in the medium/long term to reuse the Servo sync code if it meets all the requirements (although I think a simpler solution here is OK; for example I don't think there's too much benefit in getting the full history of the upstream repository).
Comment on attachment 8827857 [details]
Bug 1331899 - wptrunner changes to support v4 manifests,

https://reviewboard.mozilla.org/r/105460/#review106270

> This is... clever. Can we do something less clever? `manifest.item_types` seems like something that would be more obvious. Why does that no longer work?

I'm going to land this now and then change it afterwards, since it works and I don't want more chances for my sync to conflict.
Comment on attachment 8827859 [details]
Bug 1331899 - Fix rename handling in wpt sync process,

https://reviewboard.mozilla.org/r/105464/#review106530

I will also be building functionality for exporting changes from sub-directories of mozilla-central to Git. Long term I'm optimistic we can switch WPT to this.
Pushed by james@hoppipolla.co.uk:
https://hg.mozilla.org/integration/mozilla-inbound/rev/b16ebc0970fb
Update web-platform-tests to revision 7071a3d128ff6610a57944d1b0aaabee97b3af5a, a=testonly
https://hg.mozilla.org/integration/mozilla-inbound/rev/6e93fdded128
wptrunner changes to support v4 manifests, r=Ms2ger
https://hg.mozilla.org/integration/mozilla-inbound/rev/496a7780461a
Update wpt sync for manifest changes, r=Ms2ger
https://hg.mozilla.org/integration/mozilla-inbound/rev/78932f311341
Fix rename handling in wpt sync process, r=Ms2ger
https://hg.mozilla.org/integration/mozilla-inbound/rev/beb43155b7a6
Fix wpt manifest iterator to return the test path, r=Ms2ger
https://hg.mozilla.org/integration/mozilla-inbound/rev/432c0dee9e82
Fix mozbuild for wpt manifest v4 changes, r=gps
https://hg.mozilla.org/integration/mozilla-inbound/rev/474e53d47c35
Fix lint errors, a=testonly
https://hg.mozilla.org/integration/mozilla-inbound/rev/53c4c315533a
Update web-platform-tests expected data to revision da8dc80ce9c2155e003f20816c0cb6f120c17db6, a=testonly
https://hg.mozilla.org/integration/mozilla-inbound/rev/da044e3791dc
Stop navigation timing test loading a random page from the internet, r=Ms2ger
https://hg.mozilla.org/integration/mozilla-inbound/rev/e019af713fb2
Ensure that asserts in referer-policy tests occur in a test(), r=Ms2ger
https://hg.mozilla.org/integration/mozilla-inbound/rev/e3eb95a2fa40
Capture the full viewport for reftest screenshots, r=gsnedders
https://hg.mozilla.org/integration/mozilla-inbound/rev/2f984d729523
Update fetchlogs script to work with taskcluster, r=emorley
Pushed by james@hoppipolla.co.uk:
https://hg.mozilla.org/integration/mozilla-inbound/rev/80bbd0f4ed81
Fixup Wr metadata missing Linux 32/Ubuntu 16.04 case on a CLOSED TREE, a=testonly
Blocks: 1334525
You need to log in before you can comment on or make changes to this bug.