Re: Boost, not LEWG

From: Jeff Garland (azswdude_at_[hidden])
Date: 2020-06-27 23:36:03

On Sat, Jun 27, 2020 at 4:12 PM Vinnie Falco <vinnie.falco_at_[hidden]> wrote:

> On Sat, Jun 27, 2020 at 3:46 PM Jeff Garland via Boost
> <boost_at_[hidden]> wrote:
> >  Ideally what Zach proposes will ship under the
> > Boost banner first.  Why?   So more than the maybe 20 people on the
> > committee will look at it before the ink is dried and it ships with
> > compilers.
>
> It sounds like the committee's workflow is defective. If more people
>

It has defects like any other process -- there is no perfect approach.  And
plenty of committee members agree (as you can see many on this list are
involved). In general, the committee responds only to people that bring
proposals -- that can be members of the committee or anyone in the
community.  If you write a paper and propose something the committee will
respond.

should interact with the library before it goes into the standard,
> here's a novel idea:
>
>     Don't put it in the standard yet.
>

That's correct, I really wish it were this easy.  So one of the 1st
questions a library will face is: is there an implementation?  And the
obligatory github repo is provided (do we go download and compile it - who
has the time?).   Next typical question:  Given the priorities of the
committee for the next release, do we want to spend time on this proposal?
I'm sure a few proposals fail this bar, but probably not nearly enough.
Part of the issue there being that c++ is forever behind in basic standard
library facilities compared to other languages - we still have technical
debt.

Of course that isn't it -- if your library proposal is small you have to
make it through a design study group, Library Evolution, and finally LWG.
It would be typically a ~2 year process.  You'd think that would afford
plenty of time for committee members to look at the details, but the world
is full of distractions.  If it's bigger....well read on...

> Why does Boost.Text have to go into the standard right away?

It doesn't, and I'd be personally surprised if it makes 23 (sorry Zach).

> Why can't
> it enjoy life as its own non-std library for a few years, the way that
> Asio did?

Many apologies to Chris K. on that from me (fun fact I was asio review
manager).  Without checking,  I think over a decade of his life has expired
working toward getting asio into the standard.  All of us owe Chris a huge
debt of gratitude for the endless hours he's spent.

> Plenty of users and companies can enjoy the Unicode library
> without it having to be in the std:: namespace.
>

While true, it's still very much surprising how many cannot. Being in std::
really is the ultimate c++ distro.

Jeff