httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kean Johnston <...@sco.com>
Subject Re: Philosphical help - module or CGI
Date Fri, 09 Jan 2004 18:46:30 GMT
Thanks for the response, Nick.

> SHared memory is not easy with Apache.  If you implement a pool /
> pointers in shared memory, then you're significantly advancing it.
> If you need shared memory pointers, you might be better-off writing
> a separate daemon, and connecting to your module.
Yeah thats what I have, I have a daemon process that runs alongside 
Apache that either a CGI or a module would connect to. It does indeed 
use pointers, but because this is not an application that needs to run 
on multiple architectures or needs to be highly portable (as long as it 
works on NetBSD I am happy) what I do is always bind the shared memory 
to the same virtual address, so the pointers inside the region are 
always the same, no matter which process attaches to the segment.

> I'd suggest a persistent daemon.  Perhaps prototype it as a single
> program, then separate off the shm via RPC.  Of course if your shm
> use doesn't involve pointers then it's all much simpler.
Yeah I do that already, perhaps it was not clear in my opriginal 
question. I dont actually use RPC, its too heavyweight, by conceptually 
thats what I do :)

> I agree with that.  Once you're up the curve it becomes just as
> simple as CGI, and gives you more flexibility and modularity.
Ok thanks ... I think thats what I will do. I always like learning new 
things so this will be fun as well as practical :)

>>I intend to implement this using httpd 2.0, if that makes any difference.
> 
> Yes, that's a far more powerful development platform than 1.x.
Actually I was wondering more if 2.1 has anything that would make this 
project significantly easier. I confess I've not be following 2.1 
development much.

Kean



Mime
View raw message