Re: [apache-plusplus] Applications for HTTP servers (was "Process model for C++ Apache")

[Dean Gaudet <dgaudet@arctic.org>]

On Fri, 29 May 1998, Bret wrote:

> 	I think if you'll double check, it was said that this is true for _static_
> content.  Any type of dynamic content does _far_ more than this, like fire
> up a Perl interpreter for example.  If all you're shipping out is static
> content (aka a porn site :) ), then this would become significant...
> otherwise, for any kind of requrest for dynamic content, more than this is
> involved...

Much dynamic content is actually static.  The dynamic part is a decision
as to how to set up the output... and that decision is usually valid for
some time or over a broad range of inputs.  This is why it's sometimes a
win to build a site with a squid front-end inverse caching an apache
back-end.  Or... see my [NEWTOY] post to new-httpd about flows... a
dynamic response can become a cached flow, and behave static after that
point.  (The dynamic content generator has more control over invalidating
the cached flow than it does over invalidation in the HTTP-proxy sense...) 

Incidentally, I tried Michael's suggestion in splitting up my flow
prototype.  When I'm more awake I'm actually going to try with a single
type of thread though, so I can see a comparison.

Dean