Unclear whether document unloading should abort fetches

[domenic]

Porting from https://www.w3.org/Bugs/Public/show_bug.cgi?id=27347. Related: #1334.

Currently https://html.spec.whatwg.org/#navigate step 9, after beforeunload says

Abort the active document of browsingContext.

which in turn leads to https://html.spec.whatwg.org/#abort-a-document which says

Cancel any instances of the fetch algorithm in the context of document, discarding any tasks queued for them, and discarding any further data received from the network for them. If this resulted in any instances of the fetch algorithm being canceled or any queued tasks or any network data getting discarded, then set document's salvageable state to false.

Per the Bugzilla bug, as of 2016:

In short with recent versions:
Chrome and IE would never abort fetches just because of a started navigation.
Safari would abort "normal" fetches (ajax, loading iframes...) but not websockets
Firefox would abort "normal" fetches (ajax, loading iframes) and also websocket but not in all cases.

For a more detailed report i would need to invest more time...

The bug filer also notes that it's pretty weird for us to abort documents if the navigation is being caused by a download. Maybe we'd want to cancel ongoing fetches after we've actually updated session history or something.

[annevk]

See also #4483.

[gterzian]

Perhaps the issue should be renamed, and refocused, to "Unclear whether(or when) navigation should unload document"?

Is there really a question whether document unloading should abort fetches(related to document)?

[annevk]

I also noticed this and Firefox has gotten bug reports because of this:

• https://bugzilla.mozilla.org/show_bug.cgi?id=896666
• https://bugzilla.mozilla.org/show_bug.cgi?id=1683409

This surfaces primarily when a navigation results in a download or a 204/205.

[ricky11]

I had to delay my fetch request by 500ms in order to solve this Firefox bug, it would otherwise result in a NS_BINDING_ABORTED error in dev tools.