Lists Home |
Date Index |
On Monday 10 June 2002 8:38 pm, Aaron Skonnard wrote:
> Ok, ok, I take it back. HTTP is probably trivial to implement (having
> not implemented it fully myself). I brought it into the discussion
> because it's something we all use yet I'm sure we could still find a
> group of developers that thinks it technically sucks for whatever
As is my destiny on XML-DEV, I will now point out that HTTP sucks. It's based
on TCP, and until recently (and even then not with the trivial
implementations people throw together) set up an entire TCP connection to
then send a single request packet and get a single response packet. Even when
used in 'persistent mode' it then introduces a framing system that carefully
emulates the underlying packet-based nature of the network which TCP does a
lot of work to carefully hide away.
HTTP has a lot of confusing options that nobody seems to implement. Just look
at the list of error codes for a quick run-down of them.
It seemed to be at its best around 1.0, before they put in all that
"Connection:" stuff to try and make up for the misuse of TCP.
Check this baby out:
"Hybrid TCP-UDP Transport for Web Traffic"
Alaric B. Snell
http://www.alaric-snell.com/ http://RFC.net/ http://www.warhead.org.uk/
Any sufficiently advanced technology can be emulated in software