[
Lists Home |
Date Index |
Thread Index
]
"Bullard, Claude L (Len)" wrote:
>
>...
>
> One of our developers looking at HTTP and UDP
> said, "But sometimes we need to push the data
> to lots of different places and that isn't
> HTTP!"
I don't have time to get into a protocol re-engineering session but it
is simply untrue that it is impossible to push information to multiple
places using HTTP.
> ... BTW: reliability is not simply
> about delivering packets but sometimes
> about getting them there on time, so what
> Al was saying is relevant if it affects the
> ability of the push dispatch to get the
> right messages to the right places on time.
I think that someone else made the point that there are better ways than
just spewing UDP packets and hoping. Maybe you shouldn't be using a
packet switched network to start with! Anyhow, nobody considers HTTP a
replacement for UDP.
> VRML multicasting struggled with this mightily.
> How many real-time 3D gamers use HTTP, or the
> DIS simulations? Are they "off the web"?
Quake players think that they are surfing the web? I don't think so. If
you aren't using URIs then you aren't using the Web. If you're using
URIs but in a protocol that makes it hard or impossible to resolve them
then you're using the Web but not to its potential. The same goes if
you're using URIs but also using a bunch of private naming conventions
in method names.
Paul Prescod
|
