On Fri, Sep 20, 2019 at 6:23 PM Lassi Kortela <xxxxxx@lassi.io> wrote:

Agreed. The fx procedures raise an exception, which I think is good enough
>

In R7RS it's an error (Chicken just assumes that what you pass in is a
fixnum, even if it's a pointer).

> But we have to guarantee that users can pass any Unicode codepoint into
> any of the procedures, and they can handle it without raising an
> exception. It's not nice if for example `ascii-string?` blows up
> mysteriously when given some string with huge Unicode values.
>

No, ascii-string clearly has to accept any string at all, but I think
that's a special case.  The other procedures should I think accept any
character or any exact integer 0-127.

Is it guaranteed in all of R6RS, R7RS-small and R7RS-large that
> char->integer always returns a fixnum?
>

R7RS-small theoretically allows non-Unicode characters to exist, with
codepoints from #x11FFFF onwards, and the largest fixnum that's actually
required to exist is #x7FFFFF (though few if any Schemes limit fixnums that
severely).  In practice nobody implements or uses such characters.

--
John Cowan          http://vrici.lojban.org/~cowan        xxxxxx@ccil.org
Overhead, without any fuss, the stars were going out.
         --Arthur C. Clarke, "The Nine Billion Names of God"