|
Boost : |
From: Beman Dawes (beman_at_[hidden])
Date: 2000-06-30 09:44:10
David Abrahams wrote:
>From: "Beman Dawes" <beman_at_[hidden]>
>
> ...
>>
>> In other words, the plan is to ease forward, learning and making
>> adjustments as we go.
>
>One of the main reasons for my proposal was that it was becoming
laborious
>to have individuals integrate and synchronize with each other. If you
>restrict write access too much, there isn't much benefit.
Agreed, for the long term. Short term, we need to build a bit of
experience and build confidence before we set policies. Maybe the
transition will go so smoothly the long term is only a few days from now,
but we aren't there just yet.
>I favor more of a free-for all - you can always decide that the DB is
junk
>and wipe everything back to the last official snapshot. This model was
>recently adopted for the development of Python's IDLE editor: too many
>potential improvements were getting lost because the "official
maintainers"
>didn't have time to review and integrate them. I see the same sort of
>things
>happening to us.
>
>Some other possibilities:
>
>1. new branches can be started and contributed to at will by anyone, but
>the
>main (release) branch stays inviolate until it is merged by a maintainer.
>2. Give everyone who has contributed to a released boost library write
>access.
Yes, both those approaches sound interesting. I had been thinking of 2)
independently.
>That said, we need to have a back-door for people like Paul. I'd rather
>they
>weren't "2nd-class citizens" in any sense, though. If the ftp-a-patch
>solution I've seen suggested works, that might cover it.
Yes. I have been involved in a couple of past projects which ran into
problems because there was no backdoor for those who couldn't or didn't
wish to access the repository directly.
--Beman
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk