I feel the purpose of 'scheme-script' is perceived differently
among people.
If I'm writing applications/libraries that are used widely, I
think it is *my* responsibility to guarantee certain level of
user experience---such as performance, response time, memory
consumption, etc. So naturally I'd specify which
implementations I'm targetting at, and I'd optimize for them.
However, in that case, I'll also specify other bunch of stuffs
(like minimum requirement of CPU, memory, OS, etc.) anyway, and
include some configuration/installation tools that tailor
the script to the user's environment. Such tools can replace
#!-line at installation, so it doesn't really matter what is
written there initially. (BTW, "#!/usr/bin/env scheme-script"
has a classic security risk when the user has unsafe directories
in his/her PATH.)
When it matters is, probably, to exchange some reference
implementations or proof-of-concept type scripts. In such case,
my answer to Mr. Sperber's question is 'yes'---it's ok as
far as it runs. the #!-line is merely a declaration that
the script conforms r6rs, and it happens to be interpreted by
unix as an interpreter specification.
If the editor's intention of #!-line specification is like
the latter case, I'm ok with it. Having such a declaration
is certainly convenient. If that's the case, however, I'd
like such intention to be explained in the spec, as it seems
to have created some confusion.
--shiro
From: Michael Sperber <sperber_at_informatik.uni-tuebingen.de>
Subject: Re: [r6rs-discuss] [Formal] "#! /usr/bin/env" is not "portable." It's Unix-specific.
Date: Thu, 23 Nov 2006 19:11:28 +0100
> Abdulaziz Ghuloum <aghuloum_at_cs.indiana.edu> writes:
>
> > On Nov 22, 2006, at 1:34 PM, Michael Sperber wrote:
> >
> >>
> >> John Cowan <cowan_at_ccil.org> writes:
> >>
> >>> It's not uncommon to have more than one Scheme installed; who gets
> >>> to be "scheme-script"? By default, the last Scheme installed, which
> >>> is a silly policy.
> >>
> >> Why does it matter, as long as it runs your program?
> >
> > What do you mean? Will all conforming implementation be of the same
> > quality?
>
> I didn't mean anything, I was asking a question. (And it wasn't as
> rhetorical as you seemed to think.)
>
> Your point is taken that, in the absence of a standard location for
> the entry point, the implementation quality is uniformly low, namely
> non-existent.
>
> If it is present, implementation quality may vary wildly (as does the
> implementation quality of just about any program mentioned, in, say
> POSIX), but administrators (and even users) can at least choose, if
> they care, and not think about it if they don't.
>
> --
> Cheers =8-} Mike
> Friede, V?lkerverst?ndigung und ?berhaupt blabla
>
> _______________________________________________
> r6rs-discuss mailing list
> r6rs-discuss_at_lists.r6rs.org
> http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss
>
Received on Thu Nov 23 2006 - 15:09:55 UTC