... to support more fine-grained commenting and collaboration.
-
-
Save inexorabletash/8c122c84a65de65c35b3 to your computer and use it in GitHub Desktop.
Thanks, @jakearchibald !
Promise properties should vend the same promise each time. So
tx.promise === tx.promise
.
Fixed in the "roughly equivalent to" example. Already present in the polyfill and normative text.
Guess this should be
await tx.promise;
Yes, I missed a few of those. Bleah. Added 'em in.
I think I prefer
.complete
, but earlier you call it.promise
.
I'd switched over but forgot when I made some edits. Corrected - it's .complete
everywhere.
Unfortunately they can throw errors, which would violate http://www.w3.org/2001/tag/doc/promises-guide#always-return-promises if they returned promises
Bleah, good catch. Will ponder.
One option for IDBCursor's advance()/continue() would be to have them return an IDBRequest which then has .promise
hanging off of it. The question is: what IDBRequest? Those methods currently reset the readyState of the original IDBRequest used to open the cursor from "done" back to "pending" and fire off new "success" or "error" events.
How about
cursor()
andadvance()
return the same IDBRequest originally used to open the cursor (NOTE: I've wanted this in other polyfills!)- Generate a new internal, unfulfilled Promise (the one returned by
.promise
) for the request at the same time as the readyState is reset
So rq.promise === rq.promise
would still hold, just not over time. Iteration would then look like:
async function getAll(store, query) {
let result = [];
let cursor = await store.openCursor(query).promise;
while (cursor) {
result.push(cursor.value);
cursor = await cursor.continue().promise; // only change
}
return result;
}
I still think I'd like
IDBRequest
andIDBTransaction
to havethen
andcatch
methods that proxy to an underlying promise. Seems like forgetting to add.promise
is going to be common.@domenic I guess this idea makes you sick in your mouth?