On Tue, Apr 22, 2003 at 03:56:36PM -0400, Nathan J. Williams wrote:
> David Maxwell <david@vex.net> writes:
> 
> > Of course, one spin is "Test your application on NetBSD. Its correct
> > pthread implementation will help you identify places where you have
> > misused the thread APIs. If your application depends on behaviour
> > of a specific vendor's implementation, it will eventually break when
> > their implementation changes."
> 
> "Correct" is not the right word to use there; it's entirely legal for
> an implementation to ignore the class of errors that I recently turned
> into assertions. It's also legal to fire up GNU Chess or call your
> parents long-distance on the phone.
> 
> "Strict" might be a better description.

Ah yes, there are different cases of problems that 3rd party apps might
have. I was thinking of the 'dependence on scheduling order' problem,
where the app has made assumptions based on the particular ordering one
thread implementation consistently provides. 

For other classes of programming errors (or looseness), 'Strict' is a
better word.

-- 
David Maxwell, david@vex.net|david@maxwell.net -->
An organization gets what it rewards.
			      - Perry Metzger