Re: [boost] Proposed changes to operators.hpp

From: Joel de Guzman (djowel_at_[hidden])
Date: 2002-07-03 08:09:38

David Abrahams wrote:

>From: "Daniel Frey" <d.frey_at_[hidden]>
>
>>See also Scott Meyers "More Efficient C++", Item 4.7. If you read the
>>book, you might get the impression I misunderstood it completely, but
>>please read Scott's errata for this item, available at:
>>
>>http://www.aristeia.com/BookErrata/mec++-errata_frames.html
>>
>
>And there it says:
>
>  7/19/96 sdm   109  In July 1996, the standardization committee
>9/26/96
>                      decided that named objects may be treated
>                      essentially the same as unnamed objects for
>                      purposes of performing the return value
>                      optimization.  I added a footnote to this effect.
>
>
>Which leads me to believe that there is no "NRVO", just an "RVO".
>Furthermore, I believe that more compilers implement the RVO for unnamed
>objects than for named objects. So why aren't you using this implementation
>instead?
>
>friend const T operator+( const T& lhs, const T& rhs )
>{
>  return T(lhs) += rhs;
>}
>

It seems that reading all these, this optimization is *very* compiler 
specific.
My immediate insight suggests to me that the best approach is a, ummm,
a MACRO (do I hear yuck? :-). Such a macro might be able to switch
the most efficient implementation for a particular compiler based on boost's
config switches. After all, all of us at least agree that we need to 
optimize
away the return value if possible. I suggest that the OP use his 
benchmark and
write such a macro with specific implementations that take into account the
quirks of known compilers.

Cheers,
--Joel