On 2021-06-20 20:23 +0200, Marc Nieper-Wißkirchen wrote:
> PS There are some other references to the type "exact-integer" (not
> explained), which all should probably be updated to "fixnum".

The return types of fxmapping-count and fxmapping-size should indeed
be 'fixnum', and not 'exact-integer'.  The explanation of the other
occurrences of 'exact-integer' as a type is that I was preferring it
to 'fixnum' for the types of arguments to higher-order functions,
apparently in the name of generality.  I realize this was a mistake,
and again a result of confusing covariance and contravariance.

It's not exactly compelling, but I will fix it for consistency.

> PPS Unrelated to this but related to my ongoing discussion with Shiro: SRFI
> 224 is silent with respect to multiple returns from the higher-order
> procedures. Given that SRFI 224's objects are purely functional, the purist
> approach of full compatibility with call/cc makes sense (and which we also
> have with all R7RS base procedures).

Agreed on approach, but what changes would you suggest to make this
clear?

--
Wolfgang Corcoran-Mathe  <xxxxxx@sigwinch.xyz>

"Nothing can transcend its smallest elements." --CEO Nwabudike Morgan