Subj : Re: pthreads-win32 question... To : comp.programming.threads From : Randy Howard Date : Fri Feb 11 2005 03:02 am In article , davids@webmaster.com says... > Except there's no portable way to know how much stack space anything > takes. Although that's true, it's not really the point of my question. If those working on pthreads-win32 didn't want it to be used, then ...ATTR_STACKSIZE should not have been defined, or is there more to it than this? Similarly, having PTHREAD_STACK_MIN = 0 on a platform that does use a stack seems broken. Since phtread_attr_getstacksize() does not return ENOSYS on this implementation, that would seem to also indicate that they (at least at some point) thought that these functions might be useful to someone. You may be arguing that it didn't make sense to put these defines and interfaces into the standard in the first place, but since they are there, and since it claims to be supported, it's hard to imagine that zero is the right answer for that implementation, portability issues ignored for the moment. I'm not arguing against the point your repeating, I'm simply trying to understand why they would bother to define them if they aren't useful. A safer implementation would seem to be just removing the defines, and having the functions pertaining to changing stack size, addresses, etc. all return ENOSYS. Otherwise, it still appears to be broken, within the confines of its inherent unportable evilness. :-) -- Randy Howard (2reply remove FOOBAR) "Making it hard to do stupid things often makes it hard to do smart ones too." -- Andrew Koenig .