Re: [Boost] [Describe] Summary Review Summary

From: Emil Dotchevski (emildotchevski_at_[hidden])
Date: 2021-03-16 22:42:42

On Tue, Mar 16, 2021 at 5:47 AM Rainer Deyke via Boost <
boost_at_[hidden]> wrote:
>
> On 15.03.21 22:37, Emil Dotchevski via Boost wrote:
> > The committee seems to be concerned more with internal and external
> > politics than with serving the community. If that wasn't true there
would
> > be ZERO library additions that haven't been battle hardened by being
> > deployed and established themselves as the defacto standard already.
>
> That's just bullshit.  Right or wrong, good or bad, there are plenty of
> reasons for pushing for a library addition without waiting for a
> third-party library to establish itself as a de facto standard that have
> nothing to do with politics.  For example:
>    - There is a profusion of different libraries solving this problem,
> none of them individually popular enough individually to become a de
> facto standard.  These libraries cannot talk to each other, splintering
> the community.

"Splintering the community" simply means that there is not yet a standard
way for doing something. Either all the available libraries have issues, or
the problem being solved has multiple valid solutions. In the former case,
develop a library that works better than the rest. In the latter case, the
fact that no standard emerges is a good thing -- adding yet another way
for doing something benefits nobody.

>    - There is a clear need for a library, but the community has not yet
> produced such a library.

Oxymoron.

>    - There is a clear need for a library, but the community /cannot/
> produce such a library because it requires compiler support.

Yes, I agree that things that require a language change or even an ABI
change are different. We can discuss that too, but my point is limited to
libraries.

> Even if you think that all of these reasons are bad reasons, you should
> still acknowledge that they may have good intentions behind them and
> that they are not necessarily politically motivated.

I'm not interested in discussing the authors' motivation, that is
irrelevant. When I mentioned politics, I meant that there are politics
involved in the decision making process in the committee, not that the
politics are motivating anyone.

> > The only thing they should be doing is rubber-stamping libraries that
are
> > already the standard for doing something. Instead, it's like a giant
tube
> > for force feeding us what we don't want (or else we would have adopted
it
> > already). For our own good of course.
>
> If a library is already an established de facto standard, then there's
> little point in adding it to the C++ Standard, and a good reason for not
> doing so: it splinters the community, which is the opposite of what a
> standard should do.

If the members of the committee think they can contribute a good library,
put it on github. If it is a clear improvement over libraries produced by
less qualified authors, people will pick it up, if not, not. Or, put it in
Boost, that makes the library easily accessible on any system which -- to
your point -- may reduce the benefit of standardization.