Re: proposal: enhance and rename guile-tools

[Neil Jerram <neil@ossau.uklinux.net>]

Andy Wingo <wingo@pobox.com> writes:

> This is fun :)
>
> Here are a few threads the PLT folk had:
>
>    http://thread.gmane.org/gmane.lisp.scheme.plt.devel/2540
>    http://thread.gmane.org/gmane.lisp.scheme.plt.devel/2693/focus=2715
>    http://thread.gmane.org/gmane.lisp.scheme.plt.devel/2775

Well indeed, that pretty much covers the problem space, including my
thought above.

Some further Guile-specific thoughts, in no particular order.

- "gel" is short; and it's the name of Guile before it was Guile; and
  suggests sticking things together; and AFAIK isn't already being used
  for anything else; and has similar key positions as "gdo", so is
  equally easy to type.

- FWIW, the current differences between 'guile-tools THING . ARGS' and
  'guile -e "(@@ (scripts THING) main)" -- . ARGS" appear to be that

  1. the guile command eventually does (main (cons "guile" ARGS)),
  whereas guile-tools does (main . ARGS) (aka (apply main ARGS))

  2. the guile command runs the REPL afterwards.

- I think I might find guile-tools (as is) less bothering if its
  built-in commands (help, version and list) were rewritten as scripts
  themselves.  Then it would be clearer that the remaining code in
  guile-tools was just implementing the main-invocation convention for
  scripts/*.

(Incidentally I have a patch pending for scripts/README, to remove some
of the statements about that convention that are no longer true.)

On the overall point, of having a separate name instead of just "guile",
I think it just comes down to the observation that "because the `git'
experience seems definitive".  And I agree that it's hard to argue with
that.

Best wishes for reaching a decision on this!

      Neil