Hallvord R. M. Steen [:hallvors] Reporter
	Description • 11 years ago

Test cases:
http://w3c-test.org/web-platform-tests/master/XMLHttpRequest/data-uri-basic.htm
http://w3c-test.org/web-platform-tests/master/XMLHttpRequest/data-uri.htm

	Hallvord R. M. Steen [:hallvors] Reporter
	Comment 1 • 11 years ago

Per spec, for normal GET requests to data: the .status should be 200 and .statusText should be "OK"

	Boris Zbarsky [:bzbarsky]
	Comment 2 • 10 years ago

The links in comment 0 are 404.  Updated links:

http://w3c-test.org/XMLHttpRequest/data-uri-basic.htm
http://w3c-test.org/XMLHttpRequest/data-uri.htm

	Boris Zbarsky [:bzbarsky]
	Comment 3 • 10 years ago

The tests seem to have some issues, by the way.  Specifically:

1) data-uri-basic.htm expects data: fetches to not have a content-length header, whereas such a header is in fact desirable last I checked.

2) data-uri-basic.htm expects .responseText to return null, but this is not a nullable attribute in the IDL.

3) data-uri-basic.htm expects a POST to a data: URI to fail, but I see nothing in the fetch spec requiring that behavior.

4) data-uri.htm expects various methods to fail for data:, but I see nothing in the fetch spec requiring that behavior.

	Boris Zbarsky [:bzbarsky]
	Comment 4 • 10 years ago

In any case, the status and statusText parts of this are fixed by bug 716491.

	Boris Zbarsky [:bzbarsky]
	Comment 5 • 10 years ago

Ah, I see, the method stuff is in the basic-fetch parts of fetch.  And I seem to recall some disagreement on the spec list about the content-length which never really reached a resolution.  The null thing still seems bogus to me.

	Hallvord R. M. Steen [:hallvors] Reporter
	Comment 6 • 10 years ago

(In reply to Boris Zbarsky [:bz] from comment #3)
> The tests seem to have some issues, by the way.

Good catch. Fixes here:
https://github.com/w3c/web-platform-tests/pull/1239

	Hallvord R. M. Steen [:hallvors] Reporter
	Comment 7 • 10 years ago

(Fixes do not cover points 3 and 4 because the tests were apparently correct per http://fetch.spec.whatwg.org/#basic-fetch . This spec also says that only content-type (and not content-length) should be generated for data: - Boris seems to think this is a bad idea. Anne?

	Anne (:annevk)
	Comment 8 • 10 years ago

As far as I can tell we did reach a conclusion: http://lists.w3.org/Archives/Public/public-webapps/2013OctDec/0107.html

	Boris Zbarsky [:bzbarsky]
	Comment 9 • 10 years ago

Hmm.  OK.  We presumably need a Gecko bug on that bit then, yes?

	Hallvord R. M. Steen [:hallvors] Reporter
	Comment 10 • 10 years ago

I reversed the relevant test (but this actual failure is hidden by some other failures in builds w/o the bug 716491 fix.)