[webkit-dev] Exporting WPT tests

[Mike Pennisi]

Jeff has just created a document to explore what this tool might look like:

https://bugs.chromium.org/p/chromium/issues/detail?id=691653#c3

Youenn, this sounds like it's right up your alley!

On Fri, Apr 28, 2017 at 11:44 AM, youenn fablet <youennf at gmail.com> wrote:
> Hi Mike,
>
> Thanks for the information.
> It is really great to see Safari be integrated in the bots :)
> https://cs.chromium.org/chromium/src/third_party/WebKit/Tools/Scripts/webkitpy/w3c/wpt_github.py
> seems like a really good potential candidate for WPT upstream.
>     y
>
> Le ven. 28 avr. 2017 à 08:25, Mike Pennisi <mike at bocoup.com> a écrit :
>>
>> Hi Youenn. My name is Mike, and I've been working with Google for the past
>> 4
>> months or so to improve various aspects of the Web Platform Tests
>> project (more
>> on that here [1]).
>>
>>  > The only constraint I know of is that the test does not give flaky
>> tests from
>>  > WPT Chrome/Firefox bots.
>>
>> The full set of validation steps are described in the project's
>> `.travis.yml`
>> file [2]. That's a bit tough to read even if you're familiar with
>> TravisCI (we're
>> working on it!), but from WebKit's perspective, the only other relevant
>> check
>> is for file linting. It's not very opinionated (mostly limited to
>> objective
>> concerns) but still something to be aware of.
>>
>> Also note that we're very close to including both Edge and Safari in the
>> set of
>> browsers used to identify flaky tests! [3]
>>
>>
>>  > We do not have yet the tooling to automate the creation of a WPT
>> GitHub PR
>>  > from a WebKit patch that lands.
>>
>> I've recently been migrating tests for Service Workers from the Chromium
>> project to WPT. The process in place there is pretty slick. (Context for
>> other
>> folks on the list: it's able to create commits that exclude
>> Chromium-specific
>> files [4] and then submit GitHub pull requests from those, merging when CI
>> passes [5]. The patch Youenn mentioned is based on those files.)
>>
>> I'm wondering if we can avoid duplicating effort by making a standalone
>> tool.
>> It might even be the kind of thing we could host in the W3C GitHub
>> organization--whose to say that Edge (for example) wouldn't benefit from
>> that,
>> too? I would love to be involved in that implementation.
>>
>> But I'm getting ahead of myself :) I've CC'd Jeff Carp and Quinten
>> Yearsley of
>> the Chromium team since they are currently working with that tooling.
>>
>> So what do you folks think? Would it be practical to share code like this?
>>
>> [1] https://bocoup.com/blog/diving-into-the-web-platform-tests
>> [2] https://github.com/w3c/web-platform-tests/blob/master/.travis.yml
>> [3] https://github.com/w3c/web-platform-tests/pull/5231
>> [4]
>>
>> https://cs.chromium.org/chromium/src/third_party/WebKit/Tools/Scripts/webkitpy/w3c/chromium_commit.py
>> [5]
>>
>> https://cs.chromium.org/chromium/src/third_party/WebKit/Tools/Scripts/webkitpy/w3c/wpt_github.py
>>
>> _______________________________________________
>> webkit-dev mailing list
>> webkit-dev at lists.webkit.org
>> https://lists.webkit.org/mailman/listinfo/webkit-dev