Peter,
I once tried to write code that would issue no warnings, and ended
up going in circles -- changes that eliminated warnings on one platform
would generate them on another. Sounds impossible, but a good example is
one compiler knows that an exception or longjump jumps out of a
function, so a following return stmt is unreachable code, whereas
another compiler will know nothing about the exception and complain that
you aren't returning a value. Anyway, I agree we should try to turn off
warnings that are not warning us about real bugs. The trick is to be
cross-platform and not turn off the possibility of getting new useful
warnings in the future. If anyone knows the best way to do this or even
a solid "best practices" approach, please let me know.
-Roger
_______________________________________________
media_api mailing list
[email protected]
http://www.create.ucsb.edu/mailman/listinfo/media_api
