24023 2013-12-06 18:12:43 +0000 [Streams API] Returned promise of write() should be fulfilled with current available space size rather than how many bytes were written 2014-01-30 05:12:43 +0000 1 1 1 Unclassified WebAppsWG Streams API unspecified PC Windows NT RESOLVED FIXED P2 normal --- 1 tyoshino tyoshino feras.moussa mike public-webapps public-webapps-bugzilla oldest_to_newest 97257 0 tyoshino 2013-12-06 18:12:43 +0000 The returned value is not useful. I made this return the number of bytes actually written so that the writer can know how many bytes were written, but what the writer is interested in is how many bytes are writable now. 97261 1 tyoshino 2013-12-06 18:31:38 +0000 What we really should do is - consult dataSink (application specific logic) to determine how large (size, cost) a certain object is. dataSink (possibly asynchronously) computes the cost of a certain object written to it. pendingWriteQueue is built based on that cost. Non real application, generic consumer such as ByteStream introduced in the spec doesn't know how to compute cost for given object. So, it needs to be educated via e.g. write()'s second argument as I proposed in https://www.w3.org/Bugs/Public/show_bug.cgi?id=23977 . 99466 2 tyoshino 2014-01-30 04:28:01 +0000 (In reply to Takeshi Yoshino from comment #1) > What we really should do is > - consult dataSink (application specific logic) to determine how large > (size, cost) a certain object is. > > dataSink (possibly asynchronously) computes the cost of a certain object > written to it. pendingWriteQueue is built based on that cost. > > Non real application, generic consumer such as ByteStream introduced in the > spec doesn't know how to compute cost for given object. So, it needs to be > educated via e.g. write()'s second argument as I proposed in > https://www.w3.org/Bugs/Public/show_bug.cgi?id=23977 . Pasted wrong bug. This is the right one. https://www.w3.org/Bugs/Public/show_bug.cgi?id=24445 99467 3 tyoshino 2014-01-30 05:12:43 +0000 (In reply to Takeshi Yoshino from comment #2) > (In reply to Takeshi Yoshino from comment #1) > > What we really should do is > > - consult dataSink (application specific logic) to determine how large > > (size, cost) a certain object is. > > > > dataSink (possibly asynchronously) computes the cost of a certain object > > written to it. pendingWriteQueue is built based on that cost. > > > > Non real application, generic consumer such as ByteStream introduced in the > > spec doesn't know how to compute cost for given object. So, it needs to be > > educated via e.g. write()'s second argument as I proposed in > > https://www.w3.org/Bugs/Public/show_bug.cgi?id=23977 . > > Pasted wrong bug. This is the right one. > https://www.w3.org/Bugs/Public/show_bug.cgi?id=24445 This comment is just a mistake. Please ignore. Anyway, as https://www.w3.org/Bugs/Public/show_bug.cgi?id=24445 is changing the role of the Promise returned by the write() method, this bug is obsolete.