Alan Conway wrote:
(CCd qpid-dev for more feedback)
On Wed, 2007-01-24 at 13:39 -0500, Nuno Santos wrote:
Hi Alan,
while talking to Carl about how the trunk of APR already supports SSL
(though the stable version we're using does not), he mentioned that
we're trying to move away from APR and that you've been working on
that... my original thoughts were that since we depended on APR already,
that it might make sense to wait for the SSL-enabled release (depending
on the release timeframe, obviously), but if we're ditching APR then
I'll have to look elsewhere... any comments/suggestions on that?
It's high time we had a discussion about this.
APR is a nasty C wrapper that imposes function call overhead and uses
none of the features of C++ to make the API more type-safe, efficient or
usable. I'd like to find a decent C++ library for the job if there is
one.
http://asio.sourceforge.net/ looks promising - it's been accepted into
boost (which qpid uses a lot) and proposed for the next C++ standard.
http://www.cs.wustl.edu/~schmidt/ACE-overview.html has been around for a
long time and has a good reputation for speed in the CORBA world.
Does anyone have any comments on these or other general IO wrappers that
might be interesting for Qpid?
Even if we adopt a different/no wrapper for the qpid core, we can still
use APR for SSL but I'd like to take a bit of time to investigate
alternatives. Nuno would you be able to do a bit of reading while I get
AMQP 0-9 out of my hair?
Cheers,
Alan.
Alan,
Why not focus on the native posix one first?
Carl.
