Optimizing of explicit temporary storage

[Mike Stump]

On Oct 11, 2004, at 11:44 PM, Kai Henningsen wrote:
> Instead, what you'd need is to use some sort of language guarantees 
> about  
> the behaviour of the allocation functions themselves (new, delete, 
> malloc,  
> free) so the decision becomes local.

Or a simple statement in the language definition that pairs can be 
optimized away in the right situation, even if there are side 
effects...  For example, a user defined new/delete that does a printf 
could be considered ok, and that the compiler can get rid of pairs 
output lines.

Making this a local decision would be a good language direction for C++.