Garrett Rooney <[EMAIL PROTECTED]> writes:
> Friedrich Dominicus wrote:
>> This may sound a bit ridicolous, however I wonder how serious you are
>> about the portability of the code used for libapr. Is it wanted that
>> compiler/llibrary internals of specific compilers are used? Or is that
>> considered a bug?
>
> It needs to be portable ;-)
>
> Seriously, if you have some way to make something a bazillion times
> faster by using a gcc extension then I suppose it might be considered,
> but you'd also have to provide a non-gcc version of the code for cases
> where that wasn't possible. People compile APR with all sorts of
> compilters.
How about this code then?
/* #ifndef _WIN32_WCE
if (((*new)->td = (HANDLE)_beginthreadex(NULL,
attr && attr->stacksize > 0 ? attr->stacksize : 0,
(unsigned int (APR_THREAD_FUNC *)(void *))dummy_worker,
(*new), 0, &temp)) == 0) {
return APR_FROM_OS_ERROR(_doserrno);
}
#else
*/
if (((*new)->td = CreateThread(NULL,
attr && attr->stacksize > 0 ? attr->stacksize : 0,
(unsigned int (APR_THREAD_FUNC *)(void *))dummy_worker,
(*new), 0, &temp)) == 0) {
return apr_get_os_error();
}
// #endif
The latter works on Windows XP and probably others as well whereas the
first one uses a compiler specific internal symbol
Regards
Friedrich
