cursor.request

[jakearchibald]

Requests for cursors are different as they can succeed many times. When you call cursor.continue() the success of that operation is provided in the request returned by source.openCursor().

If I want to give an object store to another piece of code, I can just pass the object store. However, if I want to pass the cursor elsewhere, I also have to pass the related request.

Should cursor.request return https://w3c.github.io/IndexedDB/#cursor-request?

[inexorabletash]

I've run into wanting this for prollyfills. SGTM.

Should be easy to spec/test/ship.

[jakearchibald]

Yeah, I ran into this trying to make wrap() and unwrap() functions that 'upgrade' IDB objects. Works for everything except cursors.

[jakearchibald]

I'll take this on.

I've been wanting to document how standards are developed for a while, so @surma and I will film ourselves updating the spec and writing tests.

[inexorabletash]

@asutherland @aliams @beidson - any feedback on this?

[aliams]

No objections here. It sounds like we're just exposing a property that is already initialized as part of the cursor and should live as long as the cursor.

[asutherland]

Did my thumbs-up not say it all? ;) This sounds good from a Gecko/Firefox perspective. We have the request available and Jake's magical upgrade use-case sounds both reasonable and essential to help move IDB forward from an API that only an implementer could love to the favorite API of developers everywhere!

[beidson]

yah this is fine.

[inexorabletash]

Thumbs everywhere! Thanks all!

Thanks @jakearchibald and @surma - looking forward to the PRs. Blink change should be trivial once WPTs are in.