273235 2024-04-25 00:25:29 -0700 REGRESSION(277673@main+): [import-w3c-tests] -ref.html files are no longer imported 2024-05-02 11:11:07 -0700 1 1 1 Unclassified WebKit Tools / Tests WebKit Nightly Build Unspecified Unspecified RESOLVED FIXED InRadar P2 Normal --- 1 ntim gsnedders ap gsnedders jbedard ryanhaddad webkit-bug-importer oldest_to_newest 2031107 0 ntim 2024-04-25 00:25:29 -0700 ... 2031108 1 webkit-bug-importer 2024-04-25 00:25:41 -0700 <rdar://problem/127035082> 2032646 2 gsnedders 2024-05-01 17:12:08 -0700 From the old code: ``` def is_reference_filename(self): # From tools/manifest/sourcefile.py in WPT repository # https://github.com/web-platform-tests/wpt/blob/22f29564bb82b407aeaf6507c8efffdbd51b9974/tools/manifest/sourcefile.py#L405 reference_file_re = re.compile(r'(^|[\-_])(not)?ref[0-9]*([\-_]|$)') return "/reference/" in self.filename or bool(reference_file_re.search(self.filename)) ``` (Introduced in bug 234510) The WPT code this is copied from has the filename without the extension at this point, which means it matches `foo-ref` but not `foo-ref.html`; as such, we concluded `foo-ref.html` wasn't a reference and imported it as a support file. *However* we do correctly check for `/reference/`, so we do sometimes exclude the original filename. This was arguably a bug in the prior patch, as it was only after that we started importing both `-ref.html` and `-expected.html`. So we've got two questions here: 1. What is the correct behaviour? Should we be importing both -ref.html and -expected.html? (Probably yes — in the git world this doesn't bloat the repo as the files are identical.) 2. If we want to import `-ref.html`, do we *also* want to import files in `reference`? (Probably also yes — the inconsistency is the weirdest part of this!) 2032672 3 gsnedders 2024-05-01 17:59:52 -0700 Pull request: https://github.com/WebKit/WebKit/pull/28025 2032837 4 ews-feeder 2024-05-02 11:11:05 -0700 Committed 278271@main (ec1b41b9a866): <https://commits.webkit.org/278271@main> Reviewed commits have been landed. Closing PR #28025 and removing active labels.