test-webkitpy: fix webkitpy.layout_tests.port.mac_unittest.MacTest.test_skipped_file_paths
Created attachment 80526 [details] Patch
This is one way to fix the test that adam disabled in r76428. It takes the approach that this is a platform-dependent "integration test" and will work correctly for the version of the Mac it is running on. An alternative would be to rewrite this test to be version-independent and simply compare against what we think the list of skipped files is supposed to be for whatever versions we wish to support. I don't have a strong leaning either way. Thoughts?
Comment on attachment 80526 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=80526&action=review > Tools/Scripts/webkitpy/layout_tests/port/mac_unittest.py:46 > # FIXME: This test does not appear to be correct. It seems to receive Should remove this comment. > Tools/Scripts/webkitpy/layout_tests/port/mac_unittest.py:60 > + expected_paths.append('/LayoutTests/platform/mac' + port.version() + '/Skipped') Assuming that we had complex fallback logic (e.g. mac-snowleopard falls back on mac-leopard, but mac-leopard doesn't fall back mac-tiger), then it seems like we'd want to force port.version() to fixed value(s), so that we could get full coverage of those combinations regardless of what platform the test is run on. However, we don't actually have complex fallback logic. Still, I think overriding version() is the way to go.
Created attachment 80532 [details] make version-independent per review feedback from mihaip
(In reply to comment #3) > Still, I think overriding version() is the way to go. Ok.
Comment on attachment 80532 [details] make version-independent per review feedback from mihaip View in context: https://bugs.webkit.org/attachment.cgi?id=80532&action=review > Tools/ChangeLog:8 > + handles the version of the platform it is running on correctly. ChangeLog should be updated since the platform that it's running on no longer matters.
Committed r77039: <http://trac.webkit.org/changeset/77039>