How about instead of assuming APR_SUCCESS we just leave it undefined?
This seems counter-intuitive, in that a worker function that prematurely
returns (w/o calling apr_thread_exit()) was probably exiting under error.
-aaron
On Wed, Aug 01, 2001 at 04:54:58PM -0000, [EMAIL PROTECTED] wrote:
> trawick 01/08/01 09:54:58
>
> Modified: threadproc/unix thread.c
> Log:
> if the thread returns NULL instead of the address of an apr_status_t
> then assume a status of APR_SUCCESS in apr_thread_join()
>
> Revision Changes Path
> 1.43 +1 -1 apr/threadproc/unix/thread.c
>
> Index: thread.c
> ===================================================================
> RCS file: /home/cvs/apr/threadproc/unix/thread.c,v
> retrieving revision 1.42
> retrieving revision 1.43
> diff -u -r1.42 -r1.43
> --- thread.c 2001/08/01 16:45:48 1.42
> +++ thread.c 2001/08/01 16:54:58 1.43
> @@ -189,7 +189,7 @@
> apr_status_t *thread_stat;
>
> if ((stat = pthread_join(*thd->td,(void *)&thread_stat)) == 0) {
> - *retval = *thread_stat;
> + *retval = thread_stat ? *thread_stat : APR_SUCCESS;
> return APR_SUCCESS;
> }
> else {
>
>
>
