Bug 169915 – [GTK] Several WPT tests are failing since they were added in the last update in r213882

• Component: WebKitGTK
• Status: RESOLVED FIXED

[Carlos Garcia Campos]

imported/w3c/web-platform-tests/XMLHttpRequest/getallresponseheaders-cl.htm
imported/w3c/web-platform-tests/XMLHttpRequest/getallresponseheaders.htm
imported/w3c/web-platform-tests/XMLHttpRequest/setrequestheader-content-type.htm

All of them fail because libsoup fails to load the requests because the response doesn't include the empty line after the headers. Other browsers allow that, so I've written a patch for libsoup in https://bugzilla.gnome.org/show_bug.cgi?id=780352.

While the patch is reviewed and we have a new libsoup version to depend on, we can probably add the patch to our jhbuild to make sure those test pass and don't regress.

[Carlos Garcia Campos]

Actually, with libsoup patched imported/w3c/web-platform-tests/XMLHttpRequest/getallresponseheaders-cl.htm passes, imported/w3c/web-platform-tests/XMLHttpRequest/getallresponseheaders.htm still fails, but it no longer timeouts and resultds are pretty much the same, and imported/w3c/web-platform-tests/XMLHttpRequest/setrequestheader-content-type.htm still fails with the same errors.

[Carlos Garcia Campos]

imported/w3c/web-platform-tests/XMLHttpRequest/setrequestheader-content-type.htm no longer works with the current version of wptserve, so I'm going to remove our expectations, since the test is skipped globally.

[Carlos Garcia Campos]

Our result for imported/w3c/web-platform-tests/XMLHttpRequest/getallresponseheaders.htm is valid with libsoup patched, it's just different to what CF does, but still valid.

[Carlos Garcia Campos]

Created attachment 305002 [details]
Patch

[Michael Catanzaro]

Comment on attachment 305002 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=305002&action=review

> LayoutTests/platform/gtk/imported/w3c/web-platform-tests/XMLHttpRequest/getallresponseheaders-expected.txt:3
> +FAIL XMLHttpRequest: getAllResponseHeaders() 1 assert_equals: expected "also-here: Mr. PB\r\newok: lego\r\nfoo-test: 1, 2\r\n" but got "ALSO-here: Mr. PB\r\nfoo-test: 1, 2\r\newok: lego\r\n"

IMO the test should be updated to print PASS for either result, so we don't have the text FAIL in our expected results unnecessarily.

[Carlos Garcia Campos]

(In reply to Michael Catanzaro from comment #5)
> Comment on attachment 305002 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=305002&action=review
> 
> > LayoutTests/platform/gtk/imported/w3c/web-platform-tests/XMLHttpRequest/getallresponseheaders-expected.txt:3
> > +FAIL XMLHttpRequest: getAllResponseHeaders() 1 assert_equals: expected "also-here: Mr. PB\r\newok: lego\r\nfoo-test: 1, 2\r\n" but got "ALSO-here: Mr. PB\r\nfoo-test: 1, 2\r\newok: lego\r\n"
> 
> IMO the test should be updated to print PASS for either result, so we don't
> have the text FAIL in our expected results unnecessarily.

Tests are not passing, but now the expected failure is the same than in all other browsers.

[Carlos Garcia Campos]

It seems webkit-patch doesn't know how to comment and close bugs with the new version of bugzilla.

Committed r214243: <http://trac.webkit.org/changeset/214243>