> The problem with brute forcing is a browser. With you own computer it's not
> an issue to search for a file, but with browser based interpreter (like
> try.schmeme.org) everything should be in right place if this is per git repo
> it will be easily to fetch the repo and get a file like package.scm from root
> directory of a package so see the detail of the package.
>
> Snowball may be problematic when defining package, because I think it's more
> for a libraries, if each file is in same directory.
>
> So if you have package with (sort nice) (copy nice) how you will find nice.scm
> file or whatever the extension is? On server/computer this maybe not a
> problem because traversing the files is fast. But I can't imaging poking in
> browser to see if file exists.

We could list the library names in the package metadata, and then give a
rewrite rule for how to get their filenames:

(package foo
   (libraries (foo aaa)
              (foo bbbbbb)
              (foo cccc dddd))
   (rewrite-library-name-to-file-name (foo <rest> ...)
     (string-append (string-join "/" '(<rest> ...)) ".sld")))

[The above rewrite syntax is just a poor illustration of what it could be.]