Re: New pthreads-w32 releases available: versions 2.3.0 and 1.7.0

[Ross Johnson <ross.johnson@homemail.com.au>]

UPDATE: The latest change to pthread_once.c (current CVS head) passes
all the tests on an MP system, particularly the two tests that failed
with release 2.3.0 (each was run hundreds of times). Thanks to Tim
Theisen for running the tests.

This means that people can have more confidence in 2.4.0 when it comes
out (version 1.7.0 still needs to be confirmed but it uses a different
model again - and the original change that caused v2 to fail doesn't
apply to v1).

On Thu, 2005-04-14 at 18:18 +0200, Alexander Terekhov wrote:
> [... named mutex and atrocities of MS impl ...]
> 
> It's not that hard to implement it using a map with pointers (to
> "normal" mutex + refcount structure and addresses of once_control
> variables as keys).

I won't be rushing a version 3 release now unless more problems require
it, but I will pursue this. It would be nice to leave the priority
hacking to the kernel.

As it's turned out, v1 and v2 can be fixed rather than abandoned, as far
as this problem is concerned at least. Other as yet unknown bugs aside,
the version differences are now just in efficiency and, in most
applications, should not even be noticeable.

Thanks.
Ross