[Asterisk-Dev] lock.h change on 10-28 (rev 1.39) means you can never use pthread_mutex_t?

[steve at daviesfam.org]

On Tue, 1 Nov 2005, Kevin P. Fleming wrote:

> This was done for two reasons... consistency, and to allow DEBUG_THREADS
> to be used even when external modules are being built against Asterisk
> headers and then loaded. Previously, if an external module was used that
> was compiled using pthread_mutex_t but the rest of Asterisk was
> expecting those items to be ast_mutex_t (and DEBUG_THREADS was turned
> on), bad things would happen.


Well, I've been bitten by the DEBUG_THREADS mismatch with external 
modules, so understand about that.

Nevertheless, I do feel this change is rather a curve-ball to send down so
late in the 1.2 process.  5 minutes before beta2 is shipped this is
changed; my Sirrix-using users start to grab it, and ... oops ... compile
errors.  Steve?  Sirrix?  Don't you even test that your code compiles?

I don't think every pthread_ call is masked by an ast_ equivalent.  For 
instance pthread_setcanceltype() - are calls like this in danger of being 
masked out in the future?

Steve