Re: GSS-APIv3 sketch

[Nicolas Williams <Nicolas.Williams at sun.com>]

On Thu, Nov 12, 2009 at 09:06:01PM +0100, Volker Lendecke wrote:
> On Thu, Nov 12, 2009 at 01:42:51PM -0600, Nicolas Williams wrote:
> > On Thu, Nov 12, 2009 at 07:54:56PM +0100, Volker Lendecke wrote:
> > > Let me point out another thing: By using threads and a
> > > purely blocking API, you force everyone into threads. If you
> > > do it async, it is very easy to create blocking wrappers
> > > around the async calls which then can be used in threaded
> > > programs. The other way round is a lot harder.
> > 
> > That ship has sailed: Solaris 10 and up has a unified process model, and
> > it's a threaded process model.
> 
> So GSSAPI is a Solaris only thing?

How does that follow?  I'm advising you that on Solaris you can expect
core libraries to launch threads, and you can expect them to use
pthread_atfork() and atexit() propoerly.

Note that I'm NOT saying "you can't have a libevent-/whatever-based GSS
async extension".  I'm not being hostile to what you propose nor to you.

I'm saying that I think we must have a callback-based GSS async
extension as _one_ choice, and probably the required-to-implement choice
because that one makes the fewest assumptions about operating systems.

Nico
--