I see. Thanks Glenn. A. TAKAYAMA
On Thu, Oct 27, 2011 at 12:30 PM, Glenn Maynard <[email protected]> wrote: > On Thu, Oct 27, 2011 at 11:51 AM, ATSUSHI TAKAYAMA > <[email protected]> wrote: >> >> So converting Blob to ArrayBuffer is only supported through an >> asynchronous API (FileReader's readAsArrayBuffer) except in Workers, >> but the other way round is synchronous? I thought the direction of the >> client side web was to make all API's that might involve disk access >> asynchronous. > > Creating a Blob can be synchronous because it doesn't actually (visibly) > copy or move any data around--that doesn't happen until you actually try to > do something with the data, and all APIs to do *that* are async. > > Even if an implementation needs to hit the disk to make a Blob out of an > ArrayBuffer for some reason, it doesn't have to wait for that to complete > before returning the Blob. It can just return the Blob and complete the > copy asynchronously, since it already knows everything it needs to about the > Blob (its size) in order to return it. > > -- > Glenn Maynard > >
