Giving the end-of-file object a read/write-invariant lexical syntax
smells like looking for trouble.

Am Mo., 21. Nov. 2022 um 17:40 Uhr schrieb Lassi Kortela <xxxxxx@lassi.io>:
>
> Hi, and thanks for commenting!
>
> >   "Chicken and Gambit support multi-line strings via the following
> >   syntax."
> >
> > STklos does the same. And it uses #[...], as MIT Scheme does, for
> > unreadable objects:
> >
> > stklos> (current-input-port)
> > #[input-virtual-port 7f641397b790]
> > stklos> display
> > #[primitive display]
> > stklos> (define-syntax b (syntax-rules () ((_) 'whatever)))
> > stklos> b
> > #[syntax b]
> > stklos> (find-module 'stklos)
> > #[module stklos]
> > stklos> (delay 1)
> > #[promise 7f77da492dc0]
> > stklos> (define-external c-fabs ((fd :double)) :entry-name "fabs" :return-type :double)
> > stklos> c-fabs
> > #[external-func fabs]
>
> Good to know. I wonder if there are more that use #[...], and where each
> convention comes from. I can find neither #[...] nor #<...> in Emacs
> Lisp, nor in the "pitmanual" (Kent Pitman's MacLisp manual, free to read
> on the web).
>
> Just noticed that MIT Scheme has:
>
> #<abc>
> ;Value: #[uri 12 "abc"]
>
> That has to come from the syntax <http://example.com/> recommended in
> some of the web RFCs for embedding URIs in prose.
>
> > Interestingly, in STklos, EOF and void are not unreadable. Type #eof at
> > the REPL and it goes away. :) Bigloo does the same with EOF, but the
> > name is #eof-object.
>
> Chez and Chicken too. Both use #!eof.
>
> The trouble is you can't disambiguate #!eof and end-of-file. You can
> e.g. write #!eof in the middle of a source file to confuse the reader.
>
> > Or, if you type (define x #void) in STklos you'll get the
> > same result as (define x (if #f #t)).
>
> Yes. https://docs.scheme.org/surveys/void-value/