In message <[email protected]>, [email protected] writes:
>
> Agreed, however I could have opened several simultaneous connections
> for the graphics, even before I had gotten the initial HTML response.
> That delay can therefore be masked, unless the additional connection
> establishment packets add to the loss.
>
> Joe
Then you send a bunch of SYNs and FINs without any data if there is no
inlines as is the case on a good percentage of pages.
You are starting to argue in favor of a great deal of additional
inefficiency in order to avoid a performance problem that occur in the
presence of loss with already inefficient HTTP 1.0 methods.
Curtis
This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:28 EST