Re: Boost.Compat

From: Hans Dembinski (hans.dembinski_at_[hidden])
Date: 2023-05-02 09:08:05

> On 30. Apr 2023, at 01:06, Andrey Semashev via Boost <boost_at_[hidden]> wrote:
> 
> I really don't like us having
> multiple copies of the same components for the sole purpose of reducing
> dependencies (in the case of latch - reducing tests dependencies, not
> even library dependencies).

Me neither, but like everything in life it is a trade-off. I want to both reduce duplication and dependencies in Boost.Histogram, which is a rather high-level library that builds on a lot of low-level Boost functionality. To reduce dependencies, I used to have my own implementation of stuff like variant or span, which I eventually replaced with boost::span in Boost.Core and Boost.Variant2.

This is partially repeating what others have said: I see utility in Boost.Compat, because the scope of Boost.Compat is well-defined, while the scope of Boost.Core isn't. I have had discussions with the Boost.Core devs over what should be in Boost.Core or not. Knowing the scope of a boost library helps both with making requests that have a chance to be accepted and also makes it more clear for users what to expect from this library.

Regarding the reviewing of Boost.Compat: A light review is sufficient, since we can skip the discussion of the design/scope/etc, but it looks like we want to have some initial review. We are currently discussing scope and rules for this library, these should be approved by a formal review.

Just my two cents,
Hans