popov wrote:
> And sometimes it happens that compiler A does not realize the problem but
> compiler B does realize it. Compiling with different compilers and looking
> to all warnings is some kind of quality assurance.

If you have to rely on warnings to do your quality assurance you are in 
really serious trouble!!!  You also happen to make a very academic argument, 
as the "benefit" you present happens to be the biggest problem: It is 
possible, and does actually happen, that code "fixing" a warning on one 
compiler causes one on another.  Not to mention that code commonly 
introduced to "fix" compiler warnings does a whole lot of harm, in 
particular when done in the form of C-style casts.  And in other cases the 
warning cannot be eliminating without code duplication, which you would have 
noticed when compiling reasonably complex template code and getting tons of 
unused variable or argument warnings...

	Thorsten

Post a reply to this message