Re: [reSIProcate] Documentation push
Excellent question! Also one you don't hear much with open source.
Having just wrestled my way through a DNS project, I'll gladly throw in
my two cents.
The class documentation in resip is very good. One reason I picked
resip over SIPx is that it has a very intuitive object-oriented layout,
the architecture seemed robust, and in particular the DNS implementation
seemed to fully match the RFCs. Also, resip is better than most at
providing examples in the test directories. SIPx also seemed to have
more dependencies and was tougher to build on my box (an old MontaVista
distro), while resip was very easy.
However, a document or comment somewhere that says "you have to do it
this way", rather than letting me assume the design in testDns.cxx was
more of a suggestion, would help. I lost a lot of time thinking that
the DnsInterface object really didn't need to be a separate thread,
since I already had multiple calling threads and the resip code is
reentrant. Turns out I needed to keep DnsInterface in a separate thread
and serialize the access. I don't remember SIPx having that much going
on under the hood, guess I'm curious now what it's missing.
Other areas where I lost time werent resip's fault, and the list here
helped a lot. The worst one was that the rest of my code uses
RogueWave, and I had to get the resip classes to build with that. I
just had to add some #includes here and there and make sure I
compiled/linked everything against the right libs, but figuring out the
weird errors I got was a pain. It was very easy to use
build/Makefile.pkg for conditional compiles!
It's you good you put the work into this, since there are probably
plenty of people out there like me who just need the DNS stuff or some
individual component. If the "worker" classes are high quality, at some
point people like us may just decide "if we're already using resip for
xxx, we might as well use the whole stack."
Dave
-----Original Message-----
From: Byron Campen [mailto:bcampen@xxxxxxxxxxxx]
Sent: Monday, January 29, 2007 11:17 AM
To: resiprocate-devel
Subject: [reSIProcate] Documentation push
In thinking about things that we can do to boost interest in the
resiprocate project, I have come to the conclusion that the primary
barrier to entry is the quality/quantity of documentation. Are there
others that agree with this assessment?
Best regards,
Byron Campen