Am So., 6. Nov. 2022 um 22:03 Uhr schrieb Shiro Kawai <xxxxxx@gmail.com>:
>
> Ah, Gauche already uses #[...] for another purpose.
> Could it be an extension of srfi-10, e..g #,(<rtd> ....)?   Oh, wait, that conflicts with R6RS. Hmm...

The problem is that any sensible lexical syntax will probably conflict
with some existing syntax of some implementations.

Maybe SRFI 237 should mention the possibility of a

#!srfi-237

(or similar) directive (like #!fold-case), which presence would enable
SRFI 237-syntax in case the implementation does not support it in its
default reader mode.

> By the way, if the record type has a protocol clause, would the reader bypass it?

Yes. It's just a raw copy of the record's fields (and that's why only
non-opaque records are supported).

PS: I still have to think about how to match unknown records with
syntax-case (if the generalization of syntax objects mentioned in SRFI
is adopted).

>
>
> On Sun, Nov 6, 2022 at 6:40 AM Marc Nieper-Wißkirchen <xxxxxx@nieper-wisskirchen.de> wrote:
>>
>> An external representation for non-generative, non-opaque records
>> (allowing simple serialization and deserialization) is now specified
>> in my latest commit (in the personal repo).