[RFC][PATCH] krb5 => libverto main loop

[Greg Hudson]

On Sat, 2011-08-20 at 15:48 -0400, Ken Raeburn wrote:
> I've always liked that (except for the com_err mess), but I've always
> expected that eventually the external dependencies would have to
> become more advanced.  It took us quite a while to start demanding C89
> support, and after more than a decade we don't require C99.

MSVC doesn't implement very much of C99, so requiring C99 would mean
switching to MinGW for the core Windows build.  I don't know exactly
what the implications of that are since we generate DLLs and wouldn't
want to require DLL consumers to use MinGW as well.
http://www.mingw.org/wiki/MSVC_and_MinGW_DLLs suggests that at a minimum
we'd need to use a Microsoft tool to generate a lib file usable by MSVC.
If that's the end of it then it might be okay (although requiring the MS
SDK in addition to MinGW sounds a little annoying), but converting would
still be a lot of work.

> Maybe it's time to add more dependencies or at least stop supporting
> some much older configurations (lack of threads, lack of IPv6, ancient
> Tcl versions for tests, etc).

It's probably time to deconditionalize the threading and IPv6 code, yes.

(As for the issue at hand, I will take charge of doing what we
reasonably can to minimize the pain of relying on an external event
loop, but I'm not sure exactly what that is yet.)