Should we be pinning Blob URIs on fetches? about xhr HOT 6 CLOSED

That the URL can be fetched later is defined. The URL parser clones the Blob after all.

That open() errors or where we error exactly for closed blobs is not. If someone can work out what would be best there, including how that should work for FormData and the streaming case of directories I'd be most grateful. I agree though that failure for that should probably not happen during open()...

from xhr.

@mkruisselbrink @inexorabletash thoughts? What is the story for closed blobs these days?

from xhr.

Closed blobs are gone, but I created a WPT PR for the issue mentioned in OP. I think this is all defined already through the combination of URL and Fetch.

from xhr.

Bugs filed:

• https://bugzilla.mozilla.org/show_bug.cgi?id=1341609
• https://bugs.chromium.org/p/chromium/issues/detail?id=695031
• https://bugs.webkit.org/show_bug.cgi?id=168720

from xhr.

Huh. So what do we want to happen here?

from xhr.

@dmurph the way this works is that XMLHttpRequest invokes the URL parser during open(). And that ends up storing a cloned Blob object on the returned URL record: https://url.spec.whatwg.org/#concept-url-parser. That URL record gets passed to Fetch at some later point which uses the cloned Blob object.

from xhr.