- From: Domenic Denicola <notifications@github.com>
- Date: Tue, 03 Mar 2015 10:49:30 -0800
- To: whatwg/streams <streams@noreply.github.com>
- Message-ID: <whatwg/streams/issues/253/77008418@github.com>
> In Blink, we don't have it yet. Hmm OK. I thought it would just be a natural extension of the cross-worker transfer (from structured clone). But I can understand if not. Do you think there's a potential theoretical problem or is it just the practical issue of when work gets done? > Request, Response, etc. would have stream factory methods. OK. I will try to think through this idea more then. > From operationStream, I'd like to incorporate: I will continue this over on the operation stream thread. > This asymmetry still looks a little odd. We could make enqueue() return a promise that isn't fulfilled until the corresponding read() is called. But I am not sure what the use case is. > I don't insist that we should make read() return Operation, but do we need to keep each write() return a promise for returning the result of individual write()? It seems pretty desirable to know if a chunk has been uploaded or written to a file. For example that's how you would do upload progress (in the case of a fast source and slow sink). It's OK though IMO if write() returns some kind of structure like `{ ack(), finished }` instead of just returning a promise directly. Unlike read(), where the return value is always important, for write it is only used sometimes, and other times just using `ws.closed` is enough. > If write() may return after writing only partial data, write() needs to return how many bytes have been written. I don't think write() should return after writing only partial data. But maybe this is because of my ignorance of the native APIs involved. I will go do some reading on them to make sure that what I envision fits with them. --- Reply to this email directly or view it on GitHub: https://github.com/whatwg/streams/issues/253#issuecomment-77008418
Received on Tuesday, 3 March 2015 18:50:54 UTC