Le Mon, 16 Mar 2026 10:39:12 -0400,
Peter McGoron <xxxxxx@mcgoron.com> a écrit :

>  From my quick look at the generated grammar, the operators are
> treated in the implementation as syntax-rules literals.

Alas gram-expr.scm and generated-expr.scm have to be removed because
generated grammar can not handle correctly the comparison cases.

The current implementation of expr (expr.scm) relies on symbol only not
on their binding so overloading should be possible.

> In my understanding of the R6RS, something that is treated as a
> symbol is matched using `syntax->datum` and `symbol=?`, while pattern
> literals are matched using `free-identifier=?` (R6RS §11.19,
> Libraries §12.4), which only matches on the symbolic name of the
> identifier in certain scenarios.
>
> In the R7RS, the specification of matching a pattern literal is "P is
> a literal identifier and E is an identifier with the same binding"
> (R7RS §4.3.2).
>
> So to be more correct for both R6RS and R7RS implementations, the
> SRFI should say "operators are matched as literals in a syntax-rules
> declaration" or something like that, and leave out the symbol part.
>
> -- Peter McGoron