On 04 Mar 2001, Kai Großjohann wrote:
>> And this reminds me of the todo item which was sent to me by
>> Michael Kifer: use out of band transfer for large files and inline
>> transfer for small files.
Daniel Pittman <[EMAIL PROTECTED]> 05 Mar 2001 10:13:16 +1100
> Worth thinking about... which I will do. I don't know how easy or
> practical it will be, though. Is this really going to be /that/ much
> more efficient?
I suppose it would be quite a bit more efficient ... once you
partitioned "small" and "large" :-) For each pair (client, server) ?-)
There might be a useful default value, but ... allow me to make an
alternate suggestion. To determine that value would require use of a
suitable profiler, which would need to be portable enough to run on
all the various client platforms on which you'd need to test to get an
empirically-reliable result. Instead of doing the (large amount of)
required testing, why not just release the profiler, and let folks
determine the appropriate value for their local conditions?
FWIW, [EMAIL PROTECTED]
