Bug 232386
| Summary: | REGRESSION(r284521) [GTK][WPE] many tests report parse errors as a result of the test files being interpreted as XHTML rather than HTML | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Arcady Goldmints-Orlov <crzwdjk> |
| Component: | New Bugs | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED DUPLICATE | ||
| Severity: | Normal | CC: | clopez, jtojnar, lmoura, mcatanzaro |
| Priority: | P2 | ||
| Version: | WebKit Nightly Build | ||
| Hardware: | PC | ||
| OS: | Linux | ||
| See Also: |
https://bugs.webkit.org/show_bug.cgi?id=230797 https://bugs.webkit.org/show_bug.cgi?id=232471 |
||
Arcady Goldmints-Orlov
As a result of the change in r284521, numerous test files have started getting loaded as XHTML rather than HTML and the results of those tests end up being just a parse error message.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Michael Catanzaro
Of course, since r284521 was just a straight revert of r276635, I would expect the test expectations to go back to what they were prior to r276635.
The *correct* behavior would be "match whatever Safari does," but I suspect we never did. My recommended solution, from bug #230797:
"""
Of course, it would be *better* for WebKit to make content type decisions based on the DOCTYPE instead, but we can't expect a content sniffer to do that. We would have to implement it ourselves in WebKit, rather than farming out the work to GIO. I wonder how other ports handle this.
"""
This is what Bastien has recommended, as well.
Michael Catanzaro
(In reply to Michael Catanzaro from comment #1)
> Of course, since r284521 was just a straight revert of r276635, I would
> expect the test expectations to go back to what they were prior to r276635.
Maybe the shared-mime-info version upgrade is related? I didn't revert the shared-mime-info version upgrade. The addition of new parse errors could mean shared-mime-info is just doing a better job of detecting XHTML than it was before.
Arcady Goldmints-Orlov
I gardened some of the failures in r284936, and I am going to take a look at what happens if I revert the shared-mime-info change as well.
Lauro Moura
(In reply to Michael Catanzaro from comment #2)
> (In reply to Michael Catanzaro from comment #1)
> > Of course, since r284521 was just a straight revert of r276635, I would
> > expect the test expectations to go back to what they were prior to r276635.
>
> Maybe the shared-mime-info version upgrade is related? I didn't revert the
> shared-mime-info version upgrade. The addition of new parse errors could
> mean shared-mime-info is just doing a better job of detecting XHTML than it
> was before.
The shared-mime-info was bumped from 1.10 to 2.1 in r276635. Among other changes:
1.13:
* Prefer text/html to XHTML for *.html files
* Better magic for text/html files
* Fix SVG magic for files embedded in HTML
1.15:
* Fix some HTML files being detected as XML
(from: https://gitlab.freedesktop.org/xdg/shared-mime-info/-/releases )
Michael Catanzaro
We currently have two different bugs for this issue. This one is older, but I think there's a little more useful discussion in the other bug, so let's keep that one instead.
*** This bug has been marked as a duplicate of bug 232471 ***