Contemplating re-design and re-write (was: Backward compatibility)

Subject: Contemplating re-design and re-write (was: Backward compatibility)
From: Mateusz Loskot (mateusz_at_[hidden])
Date: 2018-07-05 16:27:22

On 5 July 2018 at 16:59, Christian Henning <chhenning_at_[hidden]> wrote:
> Mateusz Loskot <mateusz_at_[hidden]> wrote:
>> Christian Henning <chhenning_at_[hidden]> wrote:
>> >
>> > Same goes for features already in gil like SFINAE which can
>> > be expressed better these days in c++17.
>>
>> Prrr, hold your horses :-), we've just entered C++11.
>>
>> I don't mind switching to C++14 or even C++17, but it should come with
>> clear rationale. BTW, Switch to C++11 was the easiest one, it's just
>> obvious and necessary at numerous levels.
>>
>> I personally am not interested in re-writing GIL just for the sake of
>> using the latest & greatest C++, however exciting this idea may be.
>>
>> Again, I'd be careful about aiming for re-design and re-implementation
>> - it may require new review or it may not, who knows. I don't know.
>>
>
> I think and that's only my personal opinion,

Sure, we all do it here.

> take the existing concepts and
> go to the boost community to get some idea on how to proceed.
> It's all about learning and it should be fun at the same time.

I look at the recent months of activity in GIL.
We can't get 2-3 people regularly testing, fixing, committing or
but we expect to have time for complete rewrite.
I'm not even remotely optimistic.

I'm also hardly excited about the idea of throwing a bunch of concepts at the
Boost community asking: what can we do about it with C++17 or later.
IMHO, it makes sense to ask (very) specific question, but it does not
to ask "how to proceed". I bet even the current GIL concepts would be
thoroughly redesigned as outcome of such brainstorm, and that would be
completly different library.

Meanwhile, I personally will try to keep solving actual issues in the
current GIL.

Best regards,
-- 
Mateusz Loskot, http://mateusz.loskot.net