From: Nicolas Fleury (nidoizo_at_[hidden])
Date: 2004-01-09 11:07:36
Joel de Guzman wrote:
> Again, please don't get me wrong. I'm just asking for opinions.
> Is a single monolithic release a better solution? Is it not a
> good idea for sub-libraries, such as Spirit, to have its own
> release cycle? Can't we have both? What are the pros and cons?
> Thanks and Regards,
For my part, as a user, I appreciate the current monolithic release,
since it's a good compromise between stability and frequency. However,
there's some situation where a separate release for some part would have
been appreciated. A recent example is Boost.Python. We don't want to
use not released code where I work and installing a monolithic group of
boost libraries simplifies things. We are using Boost.Python and latest
released version is incompatible with Python 2.3. I know we could take
the latest code and update our code, but we prefer to stay with released
stuff and stay with Python 2.2 until 1.31.
I guess Boost.Python is a different beast than most boost libraries, and
maybe the same thing applies to Spirit. If a boost library is clearly
independent of most boost other libraries, or at least of what could
change, maybe an upgrade release for that library could be possible.
But I'm not asking for anything, since I don't pay a penny for Boost and
I greatly appreciate all the wonderful work in it. I remember reading
about 1.31 in november, maybe even october, so I was wrongly expecting
1.31 by the end of last year; I don't know if it would be possible to
put a next release target date (approximated) on the web page.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk