I agree with @mislav 's and @jimmywarting 's suggestions over in #572 that keeping the AbortController and AbortSignal polyfills separate from this library seems preferable.

Should we try to patch native window.fetch if it doesn't support aborting?

I think so. You may already be thinking this, but it would probably be a breaking change to the API, as this library would start polyfilling many more apps than it currently does. This might warrant a major version bump if this lib follows semver.

You should try calling xhr.abort to not wast bandwidth

It would be pretty cool to see this library update to accept a (duck-typed) AbortSignal and call xhr.abort whenever the abort event is dispatched.

A quick glance at the code suggests to me that this should be straightforward to implement. Perhaps the hardest part will be updating the feature detection code.

I wrote a quick snippet that seems like a step in the right direction. I tested it in Safari, Chrome, and Firefox, and got the expected results (no, no, yes, respectively).

What do folks think? ~~Should I code up a PR that implements something along these lines?~~ PR'd at #592

Implement abort #547

Description

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions