[racket-dev] Packaging

I think that the versioning problem is an important and hard one, and
the obvious first place to turn this infrastructure work into a
research result. Racket gives you a significant opportunity that
others would not have (for a certain class of solutions, at least).
Even better, we have significant experience with one end of the
spectrum and, of course, there are lots of other attempts out there
that we should also be able to extract experience from if we dig into
linux distros and their communities, etc.

Robby

On Sat, Feb 19, 2011 at 3:53 PM, Jay McCarthy <jay.mccarthy at gmail.com> wrote:
> Eli, you have a lot of very good points that I will try to respond to
> individually later.
>
> I think the highest bit issue you bring up is conflicts. (Multiple
> versions are a special case of conflicts between packages.)
>
> I see three options
>
> 1. Totally disallow conflicts
> 2. Make conflicting installations irregular
> 3. Make conflicting installations normal
>
> I think (1) is epitomized by a single location that packages get
> copied into as they are installed.
>
> I think (2) is epitomized by something like (1) with a secret place on
> the side for conflicts to be stored.
>
> I think my proposal is (3).
>
> I think the way I proposed to deal with conflicts dovetails nicely
> into dealing with module privacy. I like the way that keeping things
> in separate directories always makes it easy to deal with version
> control. I also like the decoupling of package names and the modules
> they provide.
>
> I'd like to know where our community stands on this issue, because if
> most are happy with (1), then I think the packaging system could be
> much simpler. I personally think (2) is more complicated than (3).
>
> Jay
>
> --
> Jay McCarthy <jay at cs.byu.edu>
> Assistant Professor / Brigham Young University
> http://faculty.cs.byu.edu/~jay
>
> "The glory of God is Intelligence" - D&C 93
> _________________________________________________
>  For list-related administrative tasks:
>  http://lists.racket-lang.org/listinfo/dev
>