[plt-scheme] planet require idea

How about not allowing the ".plt" ever, and instead of using : as the
mandatory thing to tell when a version spec is done, instead use a
colon as the (optional) signal that a version spec follows? Ie, if
there is a colon, then a version spec is expected to follow, but if
there is no colon, then you only see slashes?

Robby

On Thu, Mar 20, 2008 at 11:29 AM, Jacob Matthews <jacobm at cs.uchicago.edu> wrote:
> How about (in BNF form):
>
>  ;; new require spec ::= (planet SPEC)
>  ;; SPEC     ::= OWNER "/" PACKAGE VERSPEC ":" PATHSPEC
>  ;; VERSPEC  ::= "" | "/" MAJ MINSPEC
>  ;; MINSPEC  ::= "" | "/" PMINSPEC
>  ;; PMINSPEC ::= MIN | ">=" MIN | "<=" MIN | "=" MIN | MIN "-" MIN
>  ;;
>  ;; where PACKAGE is a string naming a package with or without the final ".plt"
>  ;;       and PATHSPEC is a normal file (possibly with path) specification
>
>  Some examples:
>
>  (planet planet/test-connection:test-connection.ss)
>  is equivalent to
>  (planet "test-connection.ss" ("planet" "test-connection.plt"))
>
>  (planet planet/test-connection/1:test-connection.ss)
>  is equivalent to
>  (planet "test-connection.ss" ("planet" "test-connection.plt" 1))
>
>  (planet planet/test-connection/1/0:test-connection.ss)
>  is equivalent to
>  (planet "test-connection.ss" ("planet" "test-connection.plt" 1 0))
>
>  (planet planet/test-connection/1/>=2:test-connection.ss)
>  is equivalent to
>  (planet "test-connection.ss" ("planet" "test-connection.plt" 1 (+ 2)))
>
>  (planet planet/test-connection/1/<=2:test-connection.ss)
>  is equivalent to
>  (planet "test-connection.ss" ("planet" "test-connection.plt" 1 (- 2)))
>
>  (planet planet/test-connection/1/=5:test-connection.ss)
>  is equivalent to
>  (planet "test-connection.ss" ("planet" "test-connection.plt" 1 (= 5)))
>
>  (planet planet/test-connection/1/1-8:test-connection.ss)
>  is equivalent to
>  (planet "test-connection.ss" ("planet" "test-connection.plt" 1 (1 8)))
>
>  (planet planet/test-connection/1/1-8:subpath/subfile.ss)
>  is equivalent to
>  (planet "subpath/test-connection.ss" ("planet" "test-connection.plt" 1 (1 8)))
>  and
>  (planet "test-connection.ss" ("planet" "test-connection.plt" 1 (1 8)) "subpath")
>
>
>
>
>  In all of these cases the initial "planet/test-connection" could be
>  replaced by "planet/test-connection.plt" with the same effect.
>
>  Does this look reasonable to people?
>
>  -jacob
>
>
>
>
>
>
>  On Thu, Mar 20, 2008 at 7:25 AM, Eli Barzilay <eli at barzilay.org> wrote:
>  > On Mar 20, Robby Findler wrote:
>  >  > On Thu, Mar 20, 2008 at 9:13 AM, Eli Barzilay <eli at barzilay.org> wrote:
>  >  > > On Mar 20, Robby Findler wrote:
>  >  > >  > There's extra stuff that fits into the parenthesized version of
>  >  > >  > the names that probably wouldn't fit into these names, but it
>  >  > >  > may be worth trying to come up with a syntax. Jacob?
>  >  > >
>  >  > >  I think so too -- just like the fact that there are some `(lib
>  >  > >  "...")'  names that don't fit into the symbol version (= when the
>  >  > >  filename does not have a .ss suffix).
>  >  >
>  >  > Well, I'd say that the missing planet names are more useful than those.
>  >
>  >  The principle is still the same...  Maybe something like making
>  >
>  >   (planet FILE-NAME PKG-SPEC [PATH ...])
>  >
>  >  shortened as
>  >
>  >   (planet <OWNER>/<PACKAGE>[:<V1>[:<V2>]]/<path>/...)
>  >
>  >     where <V1> is a number and <V2> is a number or `=<number>' etc
>  >
>  >  This should work out fine, since `:'s are not allowed in portable path
>  >  names.  It will even have the side benfit of requiring a few planet
>  >  packages in a single (planet ...) form.
>  >
>  >
>  >  --
>  >           ((lambda (x) (x x)) (lambda (x) (x x)))          Eli Barzilay:
>  >                   http://www.barzilay.org/                 Maze is Life!
>
>
> >  _________________________________________________
>  >
>  >
>  >   For list-related administrative tasks:
>  >   http://list.cs.brown.edu/mailman/listinfo/plt-scheme
>  >
>