SRFI-22 -- What won't work shivers@xxxxxx (20 Mar 2001 18:01 UTC)
Re: SRFI-22 -- What won't work sperber@xxxxxx (20 Mar 2001 18:45 UTC)
Re: SRFI-22 -- What won't work Marc Feeley (20 Mar 2001 19:21 UTC)
Script names shivers@xxxxxx (20 Mar 2001 21:52 UTC)
Re: SRFI-22 -- What won't work sperber@xxxxxx (21 Mar 2001 08:13 UTC)

Script names shivers@xxxxxx 20 Mar 2001 21:52 UTC

   > Olin> My advice: make these program names *short* and highly coded
   >
   > Suggestion for this specific case?

   Yes.  Forget about all the variants, call the executable "scheme" and
   **require** that it support SRFI 0.  Also, add the features "r4rs", "r5rs",
   and "ieee-1178-1990" that can be checked by cond-expand.  End of story.

Check me on this, please. Suppose we do it your way. We have multiple
scripts scattered over the system; some are R5RS, some are IEEE, etc.
All of these scripts fire up the same interpreter. So this one interpreter
has to be able to provide these multiple interfaces on demand, right?

That would work for differences that occur above the text->sexp parsing
step, but would not help you with lexical/sexp-syntax differences. Do
we know that the relevant options are identical at this level?

Another small point: calling the script "scheme" seems like a bad idea.
You probably want a name that won't conflict the default interactive
Scheme system on the system. "scheme-script" had that property. For something
shorter, maybe scheme0 (for "Scheme with SRFI 0"), or scmrun or scmscript
or srun.

Don't nobody tell me I am exercising tasteless judgement in over-shortening
names. I know they are ugly names. I only point out that with a 32-char
limit, you are really up against the wall.
    -Olin