Re: [boost] Core libraries should separated from experimental libraries

Subject: Re: [boost] Core libraries should separated from experimental libraries
From: Tom Brinkman (reportbase_at_[hidden])
Date: 2009-11-21 16:33:31

C++ is an for expert system and "boost" is a place for the most
advanced users of c++ to share experiences.

If you don't agree with that basic statement,
we don't have much to talk about.

I know that calling some one "average" at a particular skill is a loaded
statement.  It implies a sense of superiority that is uncomfortable to some.

However, "average" only comes with a great deal of practice and hard work.  Most
of us, only ever become average at a handful of things in our life, if
we are lucky.
Actually, calling someone average at something is highly complimentary.

If you are uncomfortable with the term "average", take your sensibilities
elsewhere.

No back to the issue at hand:

Boost needs to separate out the core libraries from the experimental libraries.

The libraries within boost that have stable api's should separated from those
that don't.

Boost needs a system similar to debains.  People need to know what parts
of boost are backwards compatible.

At the same time, we need to allow for complete interface changes to the
experimental libraries, to encourage and nurture those libraries that
are still in active development, but still very usable.

I don't see this as a difficult issue to comprehend at all.  Its quite
surprising to me that it gets so much push back.

This is the single reason why so many people are so resistant to using boost.

Hopefully somebody will take up this issue, as I fear without it, boost will
become trivialized after the next release of the c++ standard.

I'm just saying what people are thinking in the trenches.

>From my perspective, as the former review wizard, and long time follower
of boost, the current situation of having a dozen libraries in the
queue, and probably
a dozen more waiting behind the scenes is unacceptable.

If we distributed those libraries currently waiting to be reviewed as
part of the
non-stable part of boost, we could be much more liberal about what
libraries get accepted.

The important review would then be determining when a library is going to be
switched from the non-stable to the stable portion of boost.

Tom Brinkman