From: John (EBo) David (ebo_at_[hidden])
Date: 2000-12-09 07:08:03
John Maddock wrote:
> >If you don't want to change regex right now, it can be done later.
> >The main thing is to make sure we are agreed on a set of directory names
> >that are flexible enough to carry us far into the future. I really
> >hope someone does a much smarter CVS, and soon!
> No kidding :-)
> I don't mind making the changes, I just want some time to check everything
> through afterwards, and more to the point, I don't want you to issue a
> point release while I'm half way through the changes :-)
> I want to be sure that we only do this once as well.
> Is everyone happy about using "/libs/libname/src", or would "/src/libname"
> be better?
that depends... is Boost a single library? If so then I would prefer
If Boost is a collection of libraries then my personal preference is
"/libs/libname/src" with the associated "/libs/libname/include",
"/libs/libname/doc", "/libs/libname/tests", etc. But if this is the
case then there is some confusion as to what "boost/boost" is to be...
My *real* preference would be, and I know I am probably going to get
flaimed for this, to set up the directory structure something like:
boost-x.y.z/boost -- containing accepted stuff. implies frozen code!
Likewise duplicate the same methodology in:
boost-x.y.z/boost_pending -- stuff pending review. implies frozen
boost-x.y.z/boost_development -- stuff being developed and out for
Using the above directory tree you could potentially have three
different version of a class availible: the stable version, one that is
current under review, and one that is playing with ideas that are not
ready for consupmption for a formal review. I would also use this to
help facilitate testing the review code (in boost_pending) against
proposed review changes (in boost_development).
well, you asked, and that is my $0.03 worth (inflation you know...
things have gone up)...
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk