You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks, had a look at the tests. Also played around a bit more with it. Edge seems to indeed (incorrectly) not serialize the origin in the URL. But it's still possibly to tease out what origin it associates with the blob URL by trying to XHR the resulting URL. And with that it seems that only an XMLHttpRequest instance created using the top (web-platform.test:8000) origin is able to read the blob. So at least edge isn't using the current settings object, but other than that the various settings objects always confuse me. Is that actually the incumbent settings object, or is that entry settings object? Either way it still seems reasonable to change the spec to "current" settings object.
Very interesting. Maybe you could submit extra tests so that we test both the serialization and XHR's ability to access. But yes, I agree we should go ahead with changing the spec and merging the tests.
Activity
Ms2ger commentedon Feb 1, 2017
Probably uses current global in Gecko
Add tests for which global is used in blob URL origins
domenic commentedon Feb 4, 2017
I've written some (quite fun...) tests which confirm that current seems more correct: web-platform-tests/wpt#4722
@mkruisselbrink can I hand this over to you for test review + spec fixing?
mkruisselbrink commentedon Feb 6, 2017
Thanks, had a look at the tests. Also played around a bit more with it. Edge seems to indeed (incorrectly) not serialize the origin in the URL. But it's still possibly to tease out what origin it associates with the blob URL by trying to XHR the resulting URL. And with that it seems that only an XMLHttpRequest instance created using the top (web-platform.test:8000) origin is able to read the blob. So at least edge isn't using the current settings object, but other than that the various settings objects always confuse me. Is that actually the incumbent settings object, or is that entry settings object? Either way it still seems reasonable to change the spec to "current" settings object.
domenic commentedon Feb 7, 2017
Very interesting. Maybe you could submit extra tests so that we test both the serialization and XHR's ability to access. But yes, I agree we should go ahead with changing the spec and merging the tests.
domenic commentedon Feb 7, 2017
web-platform.test:8000 is the entry global, BTW; see https://github.com/w3c/web-platform-tests/pull/4722/files#diff-1449b34f69ff3a67cb389f72c59c4496R10
Add tests for which global is used in blob URL origins (#4722)