Re: [boost] is_class

From: Howard Hinnant (hinnant_at_[hidden])
Date: 2002-03-20 15:57:50

Agreed is_copyable<T> would be very nice.  What Rani's has_key_type 
might give us is a way to put a version number into an existing class 
interface.  One could check for a version_number type in a class, and if 
it isn't there, call it version 1.  If it is there, then the version 
number is T::version.  Given the class's version number, and its 
documentation, you can make assumptions on the class interface.

This might be a way to add to the interface of something like allocators 
without breaking everybody's existing allocators.  Containers could 
check the allocator's version number and then instantiate the 
appropriate code.

-Howard

On Wednesday, March 20, 2002, at 12:44  PM, David Abrahams wrote:

> I did some more thinking about that this morning, including reading the
> relevant part of the standard. It seems to me that we're still missing
> some very important predicates which we could get from a legality
> checker, like is_copyable<T>.
>
> -Dave
> ----- Original Message -----
> From: "Howard Hinnant" <hinnant_at_[hidden]>
> To: <boost_at_[hidden]>
> Sent: Wednesday, March 20, 2002 9:17 AM
> Subject: Re: [boost] is_class
>
>
>> On Wednesday, March 20, 2002, at 01:57  AM, David Abrahams wrote:
>>
>>> I wonder if this gives Howard his general "tell me if this
>>> construct compiles or not" compile-time check?
>>
>> I believe it goes a long way towards it.  This is extraordinary code.
>> I'm still trying to sort out the ramifications...
>>
>> -Howard
>>
>> _______________________________________________
>> Unsubscribe & other changes:
> http://lists.boost.org/mailman/listinfo.cgi/boost
>
> _______________________________________________
> Unsubscribe & other changes: 
> http://lists.boost.org/mailman/listinfo.cgi/boost
>