> Second: I bounced this idea of rusty yesterday, and he seemed to like
> the idea. Basically, I think after this next release of Scoop (which I
> think we might as well make 1.0), we should branch Scoop into -STABLE
> and -CURRENT branches, much how FreeBSD does its development.
> ...

The only problem I see with it is the CURRENT codebase quickly dropping
the old code that has made developing for Scoop that much harder, and
replacing it with spiffy clean new stuff.  Then new features getting added
that rely on the new whizbang code.

You can end up with a couple things here: once that code is tested you
backport it to STABLE, with the cleanups as well; or the new features
don't get to be used generally until Scoop 2.0 because getting them into
1.x would be too much work.

Optimally the former would happen, of course.  But from what I remember of
the codebase, with boxes relying on internal functions, it could get ugly
quick, and the latter could prevail.  Also I think if that happens once,
it would probably snowball and happen from then on.  Then you'd end up
like Slash and have to write a 20K line conversion script to go from 1.0
installs to 2.0...

Of course, I haven't contributed code in over a year, so this is just an
unweighted opinion :-)

-Andrew