Here's another idea:

* Most packages are in the 'empty' category. So /gcc, /paludis and so
  on.

* But Ruby modules go in 'ruby'. So ruby/syntax etc. Ditto for other
  library things where name collisions are common. We'd also have
  vim/blah and probably aspell/blah. These categories are unimportant,
  so foo resolves to /foo (with a warning), not bar/foo.

* Name collisions that aren't caused by any of the above kind of things
  get solved by giving one of the packages a different name.

* On-disk layout doesn't have to correspond to the above (we at least
  want to use something like /default/ for the empty category). We can
  work this out later -- either we go with /packages/default/XX/xxblah
  (although that probably sucks thanks to kstupid kde kpackage knames
  and lots of libeverything) or we do something cleverer.

Then we start including trove (or tags, or whatever) categorisations
in packages to make searching easier.

-- 
Ciaran McCreesh

Attachment: signature.asc
Description: PGP signature

_______________________________________________
Exherbo-dev mailing list
[email protected]
http://lists.exherbo.org/mailman/listinfo/exherbo-dev

Reply via email to