apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "David Reid" <dr...@jetnet.co.uk>
Subject Re: cvs commit: apr/include/arch/beos threadproc.h
Date Tue, 11 Sep 2001 20:24:09 GMT
+1 from me.

(No offence taken :))

david

> [This is not directed at you David. Your patch reminded me that we
>  have this problem everywhere.]
>
>
> On Tue, Sep 11, 2001 at 01:33:39PM -0000, dreid@apache.org wrote:
> >    APR_DECLARE(apr_status_t) apr_thread_exit(apr_thread_t *thd,
apr_status_t *retval)
> >    {
> >        apr_pool_destroy(thd->cntxt);
> >   - exit_thread ((status_t)retval);
> >   + exit_thread ((status_t)(*retval));
> >    return APR_SUCCESS;
> >    }
>
> I still think we should change this prototype to:
> >    APR_DECLARE(apr_status_t) apr_thread_exit(apr_thread_t *thd,
apr_status_t retval)
> There is no good reason that the return status is a pointer. Where is
> that memory supposed to be allocated from? It better not be allocated
> from the thread pool, since that was destroyed on the previous line (not
> just on beos, but all other platforms too IIRC).
>
> I'd like to supply a patch that changes apr_thread_exit() to simply take
> (apr_status_t retval) as the return value, any takers?
>
> -aaron
>
>


Mime
View raw message