Re: [boost] 1.29.0?

From: David Abrahams (dave_at_[hidden])
Date: 2002-09-17 09:52:52

From: "Douglas Gregor" <gregod_at_[hidden]>

> What are we doing with 1.29.0? I thought we were stabilizing the codebase
for
> a release soon,

We are.

> but apparently that's not the case because the MPL/type
> traits merge is not something we would do right before a release...

This is probably my fault. The reasons for this are a little complicated.

V2 of the Boost.Python library is almost ready for release. It's been a
year in the making, and when we released 1.28.0 I promised that would be
the last release of Boost.Python v1.

We had two branches of Boost.Python v2, one of which was dependent on the
mpl-development branch of MPL (an early version, no type traits
modifications), and the other which was dependent on essentially what
Aleksey wants to release. Understandably, Aleksey doesn't want an early
version of his code to go out in any boost release, even if it isn't an
official part of the release (e.g. no doc links). I discussed this problem
with the other moderators (including John Maddock, the type_traits author),
and we decided that it would be OK to merge Aleksey's refactored
type_traits before we branch for release. Given a few days to stabilize any
compiler issues, we thought it would be OK. Of course, we ran some
regressions to see that things weren't badly broken. Most things seemed to
have stayed the same. There were a few improvements.

Unfortunately, we couldn't run regressions on all platforms/compilers that
Boost users are running on. Also, I have to say, I underestimated the
potential impact of the type traits changes. It looked to me like the
changes were almost entirely restricted to factoring out the individual
components, and the addition of some template specializations that non-MPL
users would never touch.

Anyway, I hope most of those problems are stabilizing quickly.

> I would strongly prefer to have wrapped up 1.29.0 this week, released,
then
> gone on with destabilizing merges.

I think that's still possible. The plan is to branch Wed or Thursday, and
release *probably* sometime shortly after Beman returns from travelling on
October 5th, assuming things are in good shape.

> We haven't released since May, and there
> are already a huge number of changes (6 new libraries, with major
upgrades to
> 3 other libraries); on the personal side, I really want the Function
changes
> to get out into a released version of Boost before the proposal is
discussed
> in Santa Cruz so that the new syntax is available in a stable version of
> Boost.

Understood. We'll try to work with you to achieve that.

-----------------------------------------------------------
           David Abrahams * Boost Consulting
dave_at_[hidden] * http://www.boost-consulting.com