Separate the W3C web-platform-tests from the Blink layout tests. |
||||
Issue descriptionThis is a suggestion Dirk mentioned in https://bugs.chromium.org/p/chromium/issues/detail?id=609853#c20. There are two parts to this suggestion: 1. Move the imported code from third_party/WebKit/LayoutTests/imported/web-platform-tests to blink/web-platform-tests, because: - the path would be shorter/simpler - the W3C tests can be considered separately; they come from a different source and eventually we might wish to import them in a simpler way (e.g. pulling in the web-platform-tests repo via DEPS) 2. Run the W3C tests in a separate step on the waterfall and on try jobs, because: - they could be run in parallel - they use a different http server (wptserve vs apache server) - if they're taken out of the LayoutTests directory (part 1) then they won't be run automatically with the others
,
Mar 30 2017
Is this still being considered? I'm a fan of this idea.
,
Mar 30 2017
We haven't talked about it recently; it doesn't sound like a bad idea to me, but I'm not sure if it's worth the work that it requires :-/
,
Mar 31 2017
Which of the benefits would still be realized if we did this today? I think the parallelization isn't a problem any more because of layout test swarming? The path length is also short enough now that we could import all tests, right?
,
Mar 31 2017
Good point, sounds like the main benefit would now be "logical separation" -- it would imply that the web platform tests and layout tests are separate things. Although, maybe this isn't a good idea, since: - We run web-platform-tests basically the same way as other layout tests, with the same test harness, so for practical purposes for Blink, web platform tests really are a subset of layout tests. - We want to encourage people to migrate existing tests into web-platform-tests, and this seems more straightforward if it's just a change of test path, and not a change of where/when the tests are run.
,
Mar 31 2017
It's probably not worth the effort ...
,
Mar 31 2017
Yeah, that's probably the case. Alright, closing this issue. |
||||
►
Sign in to add a comment |
||||
Comment 1 by qyears...@chromium.org
, Oct 21 2016NextAction: 2017-01-01
Status: Available (was: Unconfirmed)