Re: [boost] [xint] Third release is ready, requesting preliminary review

Subject: Re: [boost] [xint] Third release is ready, requesting preliminary review
From: DE (satan66613_at_[hidden])
Date: 2010-05-04 11:49:48

on 04.05.2010 at 19:36
 Stewart, Robert wrote :

> DE wrote:
>> i don't understand how is that different to
>>
>>   X operator- (const X &x)
>>   {
>>     X ret(x);
>>     //modify 'ret'
>>     return ret;
>>   }

> As the article states, the compiler can elide the copy in various
> cases.  With your version, the copy is unconditional.

now i'm confused
will the copy be elided in this case?

>> what about idiomatic
>>
>>   return X().swap(x);  //x is modified

> That incurs the default constructor overhead, whatever that may be,
> before the swap.
i see

>   That also requires that the function argument be a
> reference to non-const X and the caller would be justifiably surprised at the effect.
of course
or 'x' being an argument passed by value

but what about copy elision?

-- 
Pavel
P.S.
if you notice a grammar mistake or weird phrasing in my message
please point it out