Re: Vectors should not evaluate to themselves

[Dirk Herrmann]

On Fri, 6 Apr 2001, Martin Grabmueller wrote:

> > From: Mikael Djurfeldt <address@hidden>
> > Date: 05 Apr 2001 01:12:28 +0200
> > 
> > Martin Grabmueller <address@hidden> writes:
> > 
> > > according to R5RS, literally entered vectors do not evaluate to
> > > themselves and must always be quoted.  Guile does not detect this
> > > error:
> > > 
> > > guile> #(as)
> > > #(as)
> > > 
> > > Should this be fixed, and can it be fixed easily?
> > 
> > It should be fixed and it is easy to fix.  We only need to remove a
> > case in a switch at the top of the evaluator.
> 
> I have done that, but not committed yet.  As I have never touched the
> evaluator before, I am not sure if the attached patch is sufficient.
> Also, R5RS only defines ``Numerical constants, string constants,
> character constants and boolean constants'' (R5RS 4.1.2) as
> self-evaluating, so the following cases should be removed, too:
> 
>     case scm_tc7_smob:
>     case scm_tcs_closures:
>     case scm_tc7_cclo:
>     case scm_tc7_pws:
>     case scm_tcs_subrs:
> 
> and the `struct' part of the following:

Unfortunately, things are not that easy.  Bignums, real and complex
numbers are realized as smobs.  Thus, if you want that all numerical
constants (and not just immediate numbers) evaluate to themselves, you
have to be a little bit more careful here.

Actually, I don't understand the requirement that vectors don't evaluate
to themselves.  The fact that it is said explicitly in R5RS is an
indication that there must be a good reason for it, but I don't see why.

Best regards,
Dirk Herrmann