Re: Boost, not LEWG

From: Zach Laine (whatwasthataddress_at_[hidden])
Date: 2020-06-28 20:31:31

On Sun, Jun 28, 2020 at 2:42 PM Emil Dotchevski via Boost
<boost_at_[hidden]> wrote:
>
> On Sun, Jun 28, 2020 at 11:56 AM Zach Laine via Boost <boost_at_[hidden]>
> wrote:
> > Raising issue on this list with the way WG21 does things accomplishes
> > nothing.  Our choices as a separate Boost entity are: 1) get involved
> > directly and individually with WG21 and fix things, for our individual
> > definitions of "fix"; and 2) accept the reality that WG21 actually
> > wants to move even faster than it already does* and make it easier for
> > Boost to accept new libraries before they are standardized, thereby
> > influencing the quality of those submissions.
>
> I disagree. Boost does not need to do anything, the problem we're
> discussing is a committee problem. Whether they see it this way or not,
> this is the truth.

I agree that this is a WG21 problem.  It *is* the truth.  But so what?
 What does knowing that truth actually do?  If WG21 does not see
things this way, the fact that it is the truth is irrelevant.  Many
people have raised this issue in WG2 meetings.  It has fallen on deaf
ears, for a variety of reasons.  Again, knowing the truth and
revealing it on a mailing list will not change the behavior in the
LEWG meeting room.

> As long as the committee is willing to "standardize" innovation -- and
> there's no indication that they intend to stop, if anything the process is
> accelerating -- there is nothing we can do to have library authors go
> through the hassle of Boost.

Here I disagree.  I've had several people ask me how to submit things
to Boost in WG21 meetings, and I've given them a rundown, and offered
some degree of hand-holding.  When only one of them (JeanHeyd)
actually followed through, I asked the others why they did not.  I got
answers in line with the suggested changes I've mentioned in this
discussion.  Those changes won't get all WG21 submitters to use Boost,
but I think it will be better than 1/N.  Also, regaining the
reputation as *the* proving ground for future standard libraries would
help too.  That relies partially on getting WG21 submitters'
involvement, but also on doing as Jeff suggested earlier -- having
backported items from C++X in Boost for use in C++<X.

Note that all of these measures would enable more people to submit to
Boost, whether they were WG21 participants or not.

> Specifically, it is not clear to me that the
> lower volume of Boost activity is a problem

We seem to be getting lower and lower numbers of reviewers.  I think
that's a problem, and I think increasing the number of active
participants may help solve that (or at least I hope so).

> , nor I think we need to attract
> the kind of developer who views Boost as an obstacle to getting their
> library standardized ASAP.

I care about the quality of submissions to Boost, and I don't want to
lower that regardless of what WG1 does or does not do.  I also care
about stopping untested library additions from moving through LEWG.  I
think Boost *could* be a partial solution to that.  I don't think
those goals need to be at odds.

> This isn't something that can be fixed from within. The last thing we
> should be doing is lower the bar in Boost.

You're talking about two different things.  We can make it easier to
get into Boost (overcoming technical obstacles and better documenting
the process) without lowering the quality bar for reviews.

Zach