Peter,

Thanks for your comments.

On 2026-07-08 07:36 -0400, Peter McGoron wrote:
> 1. Given how easy it is to detect the zero bytevector and the zero
> string, the SRFI should be required to signal an error. I don't
> think there are any extension opportunities there either.

Debatable.  (vector-copy v 3 1) is UB in R7RS, but I can't think of
any sane way of handling it other than to raise an exception.

The bigger issue here is the question whether the caller or the callee
should be checking arguments.  An R6RS fan would say every standard
procedure must check its arguments.  A bare-metal traditionalist would
prefer to do any such checking explicitly, and to disable all checks
when the need for speed strikes (which is, usually, every morning at
9 AM).  R7RS allows either approach.

One SRFI can't resolve this issue.  I will blow with the wind, as
Captain Renault[1] might have said, and the prevailing wind happens to
be from R7RS.

> 2. I think it would be good to specify what it means to position a
> cyclic port.

Good idea.  I'll add this to the next draft.

> 3. SRFI 271 (which is still in draft) should reference this SRFI to
> comment on ways to initialize ports.

Yes.  I've committed a paragraph about this to the SRFI 271 document.

[1] https://en.wikiquote.org/wiki/Casablanca_%28film%29

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