GNU bug report logs - #40373
[PATCH] guix: new command "guix run-script"

Previous Next

Full log

Message #11 received at 40373 <at> debbugs.gnu.org (full text, mbox):

From: Konrad Hinsen <konrad.hinsen <at> fastmail.net>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: 40373 <at> debbugs.gnu.org
Subject: Re: [bug#40373] [PATCH] guix: new command "guix run-script"
Date: Thu, 02 Apr 2020 09:13:23 +0200

Hi Ludo,

> What about making that just a new ‘-s’ flag for ‘guix repl’ (just like
> Guile’s ‘-s’ flag)?  ‘-q’ should always be implied when doing that.
>
> Now, this wouldn’t be usable as a shebang due to the fact that only one
> argument is allowed, unless we do some extra argument tokenizing.

That is one reason why I opted for a separate command.

The other is that I am in tutorial-driven development mode: I need "guix
run-script" in order to be able to insert my own scripts (for analyzing
dependencies) into a Guix tutorial for an upcoming MOOC. So I need to
make sure that people can run my scripts easily, but also that they
understand what they are doing. A command that does something else than
its name suggests, with a similarity that is only visible to experts,
is no good for use in a tutorial.

BTW, I opted for a lengthy name ("run-script" rather than just "run" or
"script") according to the principle that short words should be left for
frequently used concepts (a principle respected by human languages, but
also by Lisp tradition).

I am of course aware that much of the code in "run-script" is the same
as in "repl", which is not good. But I'd rather think about a better
framework for code sharing among Guix scripts than about pushing too
much semantic differences into obscure options. An example would be
reusable "option clusters", such as "options for Guile" or "options for
channels".

Cheers,
  Konrad.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.