Merge tests from css/vendor-imports into appropriate directories in css/ #8615
Comments
|
https://bugzilla.mozilla.org/show_bug.cgi?id=1423971 is a Mozilla bug for this. |
|
Looking at https://wpt.fyi/interop/css/vendor-imports/mozilla/mozilla-central-reftests, at the 3/4 column for tests that are failing only in one engine and thus would be highly valuable, I see: @dbaron, does this move need to happen from within the Gecko repo? Who would be able to do it? |
See web-platform-tests/wpt#8615 (comment) for why this directory is worth importing -- a bunch of Chrome-only failures. No-Try: true Change-Id: I8c1a0d0ce5fd7d630a5bf3196dd31e038d3e19f3 Reviewed-on: https://chromium-review.googlesource.com/1069149 Reviewed-by: Ian Kilpatrick <ikilpatrick@chromium.org> Commit-Queue: Philip Jägenstedt <foolip@chromium.org> Cr-Commit-Position: refs/heads/master@{#561113}
|
@jgraham do you support the non-wpt kind of reftests on Android in Gecko infra? Aside: It struck me today that the fix for reftests on mobile would be headless browsers on mobile. It probably doesn't exist, but is probably not very hard to support, and might make testing faster too. |
|
We're also blocked on running them on Windows and Mac and on all the issues referenced from https://bugzilla.mozilla.org/show_bug.cgi?id=1423971 (cited above by @gsnedders). |
|
And yes, non-wpt reftest run on Android in Gecko infrastructure (in addition to Mac, Windows, and Linux). wpt reftests only run in full on Linux, and a small subset of them run on Mac and Windows. |
|
https://bugzilla.mozilla.org/show_bug.cgi?id=1490969 was resolved about two months ago, deleting the original copy of these tests in Gecko, and updating /css/vendor-imports/mozilla/mozilla-central-reftests/README to say:
So it looks like it's now possible to do what this issue suggests directly as PRs in this repository: Merge tests from css/vendor-imports into appropriate directories in css/! |
These tests pass everywhere, and have no tricky dependencies. Part of #8615.
|
Have the tests already been moved? Maybe they've been copied but not deleted from the original location? It looks like my sync script broke when the directory was removed in mozilla-central, so it's possible the right thing to do here is to delete the directory rather than move anything. |
|
Oh, never mind, it looks like the end of the removal process was done differently from the rest of it. |
|
I tried a small and simple move in #26722. If that doesn't result in something blowing up I'll probably try to move some more. |
…ports into the main suite, a=testonly Automatic update from web-platform-tests [css-multicol] move tests from vendor-imports into the main suite (#26722) These tests pass everywhere, and have no tricky dependencies. Part of web-platform-tests/wpt#8615. -- wpt-commits: 2b631b86937343a916fcd49eb11a518b2b064f85 wpt-pr: 26722
…ports into the main suite, a=testonly Automatic update from web-platform-tests [css-multicol] move tests from vendor-imports into the main suite (#26722) These tests pass everywhere, and have no tricky dependencies. Part of web-platform-tests/wpt#8615. -- wpt-commits: 2b631b86937343a916fcd49eb11a518b2b064f85 wpt-pr: 26722
…ports into the main suite, a=testonly Automatic update from web-platform-tests [css-multicol] move tests from vendor-imports into the main suite (#26722) These tests pass everywhere, and have no tricky dependencies. Part of web-platform-tests/wpt#8615. -- wpt-commits: 2b631b86937343a916fcd49eb11a518b2b064f85 wpt-pr: 26722
…ports into the main suite, a=testonly Automatic update from web-platform-tests [css-multicol] move tests from vendor-imports into the main suite (#26722) These tests pass everywhere, and have no tricky dependencies. Part of web-platform-tests/wpt#8615. -- wpt-commits: 2b631b86937343a916fcd49eb11a518b2b064f85 wpt-pr: 26722
This is to make moving the tests easier, with fewer relative paths and less lint errors to ignore. As a drive-by fix, make use of XHTML-style <link /> markup more consistent within tests, which in every case meant dropping it. Part of #8615.
This is to make moving the tests easier, with fewer relative paths and less lint errors to ignore. As a drive-by fix, make use of XHTML-style <link /> markup more consistent within tests, which in every case meant dropping it. Part of #8615.
This is to make moving the tests easier, with fewer relative paths and less lint errors to ignore. As a drive-by fix, make use of XHTML-style <link /> markup more consistent within tests, which in every case meant dropping it. Part of #8615.
These tests all pass in Chrome, Firefox and Safari, except perspective-containing-block-dynamic-1b.html which fails in Safari: https://wpt.fyi/results/css/vendor-imports/mozilla/mozilla-central-reftests/transforms?run_id=5655021456719872&run_id=5745838489862144&run_id=5657645899841536 Part of #8615. Also clean up reftest.list for flexbox.
|
I agree it's not always easy to decide. I would say that as long as you can find a destination folder (feel free to ask if not sure) then just moving all tests without renaming or deduplication is fine. A reviewer might ask for some cleanup, but then they will have also helped you pinpoint what to clean up. |
…r-imports/ (feel free to!), a=testonly Automatic update from web-platform-tests Update warning about modifying css/vendor-imports/ (feel free to!) See web-platform-tests/wpt#8615. -- wpt-commits: 42d93de9be526a7548e18a75a949e8a7357ba26d wpt-pr: 26720
|
There's only one directory left: https://github.com/web-platform-tests/wpt/tree/master/css/vendor-imports/mozilla/mozilla-central-reftests It's only generated tests. Feel free to take it. |
|
@nt1m Not all heros wear capes! But for all I know you may be wearing a cape right now, so it's impossible to say if that's a relevant fact or not ;) |
cc/ @jgraham @dbaron
At the moment, the tests in
css/vendor-imports/mozilla/mozilla-central-reftests/cause a certain amount of confusion: they often get omitted when people are trying to run all tests for a certain module, and they often don't get noticed when people are looking to see if tests for a certain feature exists.The text was updated successfully, but these errors were encountered: