Re: A proposal for a modern build-system

From: Hassan Sajjad (hassan.sajjad069_at_[hidden])
Date: 2024-03-22 12:39:42

>
> I am still hoping for a Modules v2 from the committee which has tangible
> benefits for the majority of users.
>

I am interested in what v1 misses that you want in v2. I believe because
of  https://lists.isocpp.org/sg15/2023/11/2106.php, the adoption will be
much quicker than you think.

On Fri, Mar 22, 2024 at 3:47 PM Niall Douglas via Boost <
boost_at_[hidden]> wrote:

> On 21/03/2024 17:34, Hassan Sajjad via Boost wrote:
>
> > I want to propose my build-system HMake for boost. Besides its already
> > state-of-the-art C++20 modules and header-units support, it will be the
> > first to support https://lists.isocpp.org/sg15/2023/11/2106.php or
> > https://lists.isocpp.org/sg15/2023/11/2146.php if any of this gets
> > implemented. These papers present ideas for faster module adoption and
> > avoiding redundant module compilations. These need support from
> > build-system. This way boost could be a breeding ground for C++20 modules
> > adoption. Also, I have good experience with boost's current build-system
> b2
> > as a good portion of HMake's current API is inspired by it.
>
> Unless it's improved cmake build support for Boost, I cannot say I am
> interested in yet another build system.
>
> I also think few would consume Boost as modules for many, many years yet
> to come. Precompiled headers if they work provide most of the build
> speed improvements, and everything else Modules v1 supplies has marginal
> utility for the vast majority of C++ use cases if I am blunt.
>
> I am still hoping for a Modules v2 from the committee which has tangible
> benefits for the majority of users.
>
> Niall
>
> _______________________________________________
> Unsubscribe & other changes:
> http://lists.boost.org/mailman/listinfo.cgi/boost
>