On Mar 25, 2010, at 4:50 PM, Ian Hickson wrote:
It seems that the consensus is now leaning towards changing the spec
again
to include the overhead, but I haven't changed it yet because I
don't want
to be flip-flopping back and forth -- if we change this, I don't
want to
change it back. I think the use cases I know of are addressed
whether we
include overhead or not, so from an objective perspective I don't
think it
really matters, which makes this more of an opinion thing. I'd
encourage
anyone else with an opinion on this to make their opinion known.
(Yes, I'm
actually _asking_ for people to suggest colours for this bikeshed!)
As an author, I'd rather not know about the overhead. To me it would
be confusing to queue up 10 bytes and ask for bufferedAmount and get
back 15.
The sample scripts are doing one of two things. Either they are
trying to prevent saturating the network and they are doing this by
testing for bufferedAmount being equal to zero. Or, they are trying
to saturate the network given the existing interfaces. I believe
there are other options to allow the application to maintain an
asynchronous interface and achieve saturation that would consume less
cpu overhead. I'm not sure if we want to open that can of worms in
this thread or not.
pedz