Re: Cistron-Radius Users
On Fri, 15 Mar 2002, Russell Coker wrote:
> The problem with multithreading is that debugging can be a bitch. Core files
> don't support multi-threading, debuggers have problems with it too, so too
> often you end up doing printf() debugging.
> The positive thing about threaded programming is that there's no cost to
> transferring data between threads. You want to malloc() in one thread and
> free() in another, no probs. This adds some extra freedoms to the design.
> Would you like to have one thread for reading data from the net, a pool of
> threads doing database lookups, then another thread writing responses? It's
> no challenge to code.
> Mainly I think it's what you're most experienced with. You obviously have
> more skills in fork() based programming so I won't try and convince you to
> change your programming style. But threaded programming can work equally
> well if you've had some practice.
This is about the best and most well-balanced counter opinion I've heard
so far. Indeed, the zero-copy properties are more than nice.
I'll admit that I'm a bit extreme in this -- I'd sooner create a whole
infrastructure based on shared memory and signals to achieve zero-copy
communications between processes than just do what's obvious: go
> > If you're interested, see http://www.openradius.net for another
> > approach.
> Are you going to package it for Debian?
I definitely plan to. I'm near release 0.9.2 which fixes some minor bugs
and a memory leak, and I'd love to package this one or the next.
Thing is, I'm not too familiar with the whole process and policies. I
love what these do for Debian, as I get to experience its advantages
almost daily, but need to learn a bit more about this.
> Also are you planning to attend the Debian meeting in den Haag on Saturday?
I'd like to -- I live *very* near there. Where can I find more
information? It isn't announced as an event on debian.org.
E-Advies / Emile van Bergen | email@example.com
tel. +31 (0)70 3906153 | http://www.e-advies.info