Re: [Patch] First bits of the algo merge

[Paolo Carlini <pcarlini at suse dot de>]

Hi Howard,

and thanks for the review.

> I'm wondering if this shouldn't be:
>
>  __glibcxx_function_requires(_LessThanOpConcept<_ValueType2, 
> _ValueType1>);

[snip]

>> @@ -3635,11 +3691,16 @@
>>      includes(_InputIterator1 __first1, _InputIterator1 __last1,
>>           _InputIterator2 __first2, _InputIterator2 __last2)
>>      {
>> +      typedef typename iterator_traits<_InputIterator1>::value_type
>> +    _ValueType1;
>> +      typedef typename iterator_traits<_InputIterator2>::value_type
>> +    _ValueType2;
>> +
>>        // concept requirements
>> -      __glibcxx_function_requires(_LessThanComparableConcept<
>> -        typename iterator_traits<_InputIterator1>::value_type>)
>> -      return std::includes(__first1, __last1, __first2, __last2,
>> -               __gnu_cxx::__ops::less());
>> +      __glibcxx_function_requires
>> (_LessThanComparableConcept<_ValueType1>)
>> +
>> +      return std::includes(__first1, __last1, __first2, __last2,
>> +               __gnu_cxx::__ops::less<_ValueType1, _ValueType2>());
>>      }
>>
>    // Set algorithms: includes, set_union, set_intersection, 
> set_difference,
>
> And this:
>
>       __glibcxx_function_requires(_LessThanOpConcept<_ValueType1, 
> _ValueType2>)
>       __glibcxx_function_requires(_LessThanOpConcept<_ValueType2, 
> _ValueType1>)

You are spot on! Actually however, you are arguing about different
functions than I touched (as regards concept checks), about which I
meant to ask later ;)

Indeed, there is a whole series of functions, those that you pointed
out, which have got a very strong (too strong)
_SameTypeConcept<_ValueType1, _ValueType2> in the predicated version of
the algorithm (e.g., includes, set_*, etc.). Because of that, I believe
_LessThanComparableConcept<_ValueType1> is currently equivalent to what
you are suggesting.

Therefore, the real question is: shall we remove
_SameTypeConcept<_ValueType1, _ValueType2> from the predicated version,
in those cases? Let's reach an agreement about that and let's go ahead
with this part of the work!

Thanks,
Paolo.