Re: [boost] [contract] concept error messages

Subject: Re: [boost] [contract] concept error messages
From: Lorenzo Caminiti (lorcaminiti_at_[hidden])
Date: 2012-10-09 03:47:25

On Mon, Oct 8, 2012 at 11:00 AM, Matt Calabrese <rivorus_at_[hidden]> wrote:
> On Mon, Oct 8, 2012 at 1:56 PM, Matt Calabrese <rivorus_at_[hidden]> wrote:
>
>> It's theoretically possible, to an extent, when I get automatically
>> generate archetypes from the pseudo-signatures working and if people were
>> to use a Boost.Generic macro when writing their constrained algorithms. I
>> was working toward generating archetypes last year, but it's now on hold
>> until I finish fixing explicit concept maps, which is taking a little bit
>> longer than expected. I wasn't expecting to use archetypes to the full
>> extent that they were used in N2914 though because it's complicated to do
>> in a library. What will probably end up happening first in that area is
>> archetypes will be automatically generated from concept definitions, but
>> people will have to manually work with them. With algorithms that have
>> simple constraints specified as template< ((Regular)) T > void foo( T );
>> it's definitely feasible, but with complicated clauses, it would be
>> difficult.
>>
>
> Something that's interesting -- if both of our libraries end up supporting
> automatic generation of archetypes, we could coordinate how they are looked
> up and the way we represent them internally so that a single macro for
> writing algorithm would work with both libraries. I feel like we're both a
> ways away from that at the moment though.

Agree it'd be good to coordinate and agree this stuff is probably far
away into the future ;)

Thanks,
--Lorenzo