[OpenAFS-devel] pthreads, darwin, microsoft: untwisting the knot

[Marcus Watts]

From: Jim Rees <rees@umich.edu> writes:
> That seems pretty complicated.  Why not just have a separate
> rxkad_InitPthread routine and make applications call it?  Are there a lot of
> existing applications that would break?

All those applications today are calling rxkad_NewClientSecurityObject
or rxkad_NewServerSecurityObject .  Now, if those functions *could*
use pthread_once.  But even this isn't as simple as just having
the initializer in the first place.

It's only complicated because english isn't a good way to describe it,
and today's consequences *are* complicated.  The existing logic there
is extremely simple; this change won't make it much more complicated.  I'll
try to present a more details example of this soon -- one reason I was
hoping to run this by folks here was in the hopes of having actual
tested diffs.

				-Marcus