Prev: Call for Paper The International Journal of Computer Science (IJCS)
Next: vector:erase iterator requirements
From: Paul Bibbings on 6 Feb 2010 04:19
Olivier <olivier.grant(a)gmail.com> writes:

> On Feb 5, 8:47 pm, Paul Bibbings <paul.bibbi...(a)gmail.com> wrote:
>> Olivier <olivier.gr...(a)gmail.com> writes:
>> > Daniel, your solution is definitely very elegant, and I looking
>> > forward to C++0x for this. But there is still something I don't seem
>> > to get with rvalues. Take the following code which is an extension of
>> > your solution :
>>
>> > 3
>> > 4 class Bad
>> > 5 {
>> > 6 public:
>> > 7 explicit Bad( int const &i ) : i_(i) { }
>> > 8 explicit Bad( int const && ) = delete;
>> > 9
>> > 10 private:
>> > 11 int const &i_;
>> > 12 };
>> > 13
>> > 14 class Bad2
>> > 15 {
>> > 16 public:
>> > 17 explicit Bad2( Bad const &t ) : t_(t) { }
>> > 18 explicit Bad2( Bad const && ) = delete;
>> > 19
>> > 20 private:
>> > 21 test const &t_;
>> > 22 };
>> > 23
>> > 24 int main( )
>> > 25 {
>> > 26 int i(23);
>> > 27
>> > 28 //Bad t0(23); // <- Kaboom! Expected.
>> > 29 Bad t1(i);
>> > 30
>> > 31 Bad2 t20(Bad(i)); // <- No Kaboom ? Why ?
>> > 32 Bad2 t21(t1);
>> > 33
>> > 34 return 0;
>> > 35 }
>> > 36
>>
>> > I would have (naively) expected to be able to do the same thing with
>> > non built-in types. But this compiles fine with GCC 4.4 (line 28 blows
>> > up if uncommented). So what i'm not getting is why does 23 bind to an
>> > rvalue reference, where a call to Bad(i) does not ? What am I
>> > overseeing ?
>>
>> Try replacing your line 31 with:
>>
>> Bad2 t20 = Bad2(Bad(i));
>>
>> If you then still need help working out why your original line 31 seemed
>> to be accepted by the compiler when this one isn't, replace the original
>> with:
>>
>> Bad2 t20(Bad i);
>>
>> which is, in truth, a mere notational alteration that doesn't affect its
>> semantics at all in this instance.
>
> Ok, now I'm completely confused :)
>
> Bad2 t20(Bad i);
>
> This is totally ill-formed to me. I'm not even sure how the compiler
> should/does interpret that. But :
>
> Bad2 t20(Bad(i));
>
> This isn't. Why is the compiler choking on this? How is it
> interpreting this statement? Is this GCC specific?
>
> Sorry, I know this is slightly off-topic, but still interesting I
> thinnk.

It is indeed interesting, and very important to be aware of. You can
get the background by searching on the internet for C++'s "most vexing
parse." A quick introduction by Danny Kalev on InformIT is:

http://www.informit.com/guides/content.aspx?g=cplusplus&seqNum=439

where he puts it in the context of what he (optimistically) calls C++09.

Regards

Paul Bibbings

--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]

 | 
Pages: 1
Prev: Call for Paper The International Journal of Computer Science (IJCS)
Next: vector:erase iterator requirements