Feature req for GPGME

[Robert J. Hansen]

> The Windows way :-)

I've seen it in UNIX code, too, so we can't dismiss it out of hand as 
being Windowsish.  I agree it's not ideal, though.

> Without accessor fucntions we won't have the ability to generate the
> data on demand, instead we have to populate the struct with all
> possible data, most of them not needed by allpications.

I'm already doing this in my C++ bindings.  For a 50-key ring, creating 
the keydb representation takes under a second.  Haven't tried it on 
anything larger.

> GpgmeCtx and GpgmeKey are independent.  I fail to see how you approach
> is better - you also have to make sure that KS is released.

Yes, but if you don't release KS, all you have is a memory leak.  If you 
don't release a GpgmeKey, you're leaking a resource which shouldn't and 
mustn't be leaked.  But I'm an admitted paranoid about these things.  :)