httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexei Kosut <>
Subject Re: NT and modules
Date Wed, 18 Jun 1997 04:53:40 GMT
On Wed, 18 Jun 1997, Rasmus Lerdorf wrote:

> I don't really see any benefit to this though.  In a pre-forking
> multi-process environment you get absolutely no performance benefit 
> from dynamically linking in modules at runtime.  The only benefit is in
> the convenience of not having to recompile to add a module.  This
> convenience comes at the expense of a much more difficult compile of each
> module.

IMHO, this convenience is worth it. Especially (and remember where
this thread started) in the personal computer market, where OSes don't
come with compilers, and no one would want to compile a server
anyway. I'm not saying that we should make it so people can't compile
modules into the server. For freeware Unix modules (like yours) that's
fine. However, for Windows, it's not. And for commercial modules, it's
not: If I don't want to give away my source, it's hard to sell an
Apache module. A working run-time module loader would help. Also, a
coherent API that doesn't change with every release. But I've written
about that before.

> It would be nice to get to the point where Perl is.  Having a standard
> mechanism to build a module and to make it available to Apache.  Then
> doing something like a "Use mod_rewrite" in the httpd.conf file, or even
> per-request in a .htaccess.  mod_dld is a very long way from something
> like that.

I don't see that. If mod_dld works on your system, "cc -o
mod_whatever.c" will built a module, and "LoadModule whatever_module
mod_whatever.o" will install it at runtime.

Alexei Kosut <>      The Apache HTTP Server

View raw message