Another way to describe this scenario is that you have a number of distinct
products that end up checked out into a common source code tree structure.
In DCMS this can be handled by putting the more sensitive stuff in a
separate project and controlling access to that project. When checked out,
the various projects can share the same source tree structure.
I think I am arguing that if you have differentiated on access rights then
by definition you don't have one project anymore, and trying to squeeze the
whole into one project doesn't seem to fit very well.
However, I'm still pondering what you say. It's definitely thought
provoking.
> Here at MIT we have a source tree which
> is composed almost entirely of free software, but has a few
> directories containing proprietary stuff like AFS binaries. We'd very
> much like to make the free portions of the tree readable to anyone,
> but if we had to have the same permissions on all parts of a given
> branch, we'd have to restrict the whole repository and then make
> snapshots of the public parts available or some other kludge.
>
Received on Sat Oct 21 14:36:06 2006