Certificate Management stuff

[Ivan R. Judson]

> If you want to spearhead an effort bring m2crypto up to 
> speed, that's fine with me; I can supply test code for the 
> various bits of pyopenssl that I use, but will forge ahead 
> using the tools that work now.

I'd be happy to start moving this direction, but I'd want commitment that
once it was in place we could move to it as a replacement. I don't want to
fracture our development externally, let alone internally. Given the
statement in the other email that everything is there, I can assume what's
there is a complete set of requirements then, right?

I think this is in our long term interests, since it means we can offload
support to the M2Crypto developer, which means a smaller code base for us to
maintain. Given our resources and their allocation (not to support
necessarily), it's prudent that we try to move this direction as much as
possible. I'm not opposed to us using *any* tools internally, but in terms
of released code, I definitely would like to reduce the amount of code
(sheer quantity) that we have to support to code that is a large percentage
authored *by us*. Seems a better use of our resources.

Tests would be exceedingly cool in either case ;)

--Ivan