Am Fr., 24. Jan. 2025 um 19:15 Uhr schrieb Wolfgang Corcoran-Mathe
<xxxxxx@sigwinch.xyz>:
>
> Hi Marc,
>
> Thanks for your extensive review. I’ll respond to your earlier message
> after I’ve had a chance to think about it carefully.
>
> On 2025-01-24 18:41 +0100, Marc Nieper-Wißkirchen wrote:
> > I would like to make the following hopefully constructively read
> > suggestion that gives implementations enough freedom but also allows
> > implementations to address the points I raised in the previous post:
> >
> > (1) Provide only a single procedure
> >
> > (generate-symbol [pretty-name])
> >
> > that returns a symbol (in the sense of `symbol?') that may or may not
> > be interned and that is guaranteed to be different in practice from
> > any other symbol (*). An implementation may use PRETTY-NAME for
> > printing purposes.
>
> This is very similar to Chez’s gensym objects, and here is my complaint
> about them: They are two-headed symbols. Only Chez supports symbols
> with pretty and unique names, as far as I know, whereas uninterned
> symbols are supported by a majority of Schemes.

My suggested reduced proposal should be compatible with all Schemes.
Firstly, it is not specified where the pretty name goes (it is solely
for diagnostic purposes). Secondly, the uninterned symbols of the
current SRFI 258 draft would also implement `generate-symbol'; the
pretty name would just be the textual name.