From: David Abrahams (dave_at_[hidden])
Date: 2003-09-30 07:27:48
Beman Dawes <bdawes_at_[hidden]> writes:
> I've got some time opening up and would like to start working toward a
> 1.31.0 release.
> From the release manager's standpoint, there are really several
> aspects of getting a release ready:
> (1) Ensuring that the library code is reasonable shape. This is done
> by monitoring regression tests, and by monitoring the list looking
> for patches, bug reports, or other messages that directly relate
> to library code, and coordinating with library developers. This is
> very time consuming.
Don't forget the documentation.
Jeremy, Thomas, and I still have to do another revision to the
iterator library and docs.
> (2) All other preparation and coordination. This includes updating
> non-library-specific web pages, working with developers to
> minimize quality problems such as detected by the inspection
> report, coordinating with developers for promised updates, etc.,
> etc., etc. This is also fairly time consuming, although perhaps
> not quite as time consuming as (1).
> (3) The mechanical aspects of preparing a release. This is by far the
> simplest part, although this particular release is complicated by
> moving to a new host.
> For past releases, the release manager has had lots of help from the
> maintenance wizard and others, but has born the primary responsibility
> for the above tasks. On this release, I'd really like to split the
> responsibility. Perhaps by one person managing (1), while someone else
> manages (2). I'll volunteer for (2) and (3), but see the next
> While (3) is much simpler, the hosting is moving to SourceForge
> so it would be helpful if a Unix/Linux developer volunteered to
> write and run a release script that takes the distribution .gz or
> .zip and installs it on the host.
> Since Dave Abrahams managed the 1.30.1/2 release, I'd particularly
> like to hear his views on how to organize management of this
> release. Dave?
Well, for 1.30.2 I had a lot of help from meta-comm with the
regression issues, not to mention Martin Wille who was running tests
on Linux. I think it would be ideal to have at least 2 people
cooperating on (1). Wish I was brimming over with other ideas, but
I'm not. Sorry!
-- Dave Abrahams Boost Consulting www.boost-consulting.com
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk