Re: [w3c/IndexedDB] Behavior tested by transaction-abort-multiple-metadata-revert.html is not obviously specified anywhere (#109)

Agreed, this is underspecified. I'll propose text here.

The overall test was probably written to verify the steps in https://w3c.github.io/IndexedDB/#abort-upgrade-transaction which *revert* the clearing, which apparently implementations did, but the clearing you note in step 3 itself isn't described. I believe @pwnall verified that the behavior matched in Chrome and Firefox, so I guess we should keep it.

(I also agree in the abstract that it could go either way. I'm also not a fan of how "handles" revert following an abort rather than just holding stale data but I got outvoted on that one back in "v1". Oh well.)

The distinction between "object store handle" and "object store" (etc) is something that didn't exist in the "v1" spec and even more behavior like this was underspecified. There is almost certainly more I missed so please keep feedback like this coming! (Things like "Destroy store" could also be more rigorous; as I stare at it I'm not sure that actually has any meaning.)


-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/w3c/IndexedDB/issues/109#issuecomment-258557497