Am Mi., 2. Nov. 2022 um 13:49 Uhr schrieb Marc Nieper-Wißkirchen
<xxxxxx@nieper-wisskirchen.de>:
>
> Am Mi., 2. Nov. 2022 um 13:28 Uhr schrieb Marc Feeley <xxxxxx@iro.umontreal.ca>:
> >
> >
> > > On Nov 2, 2022, at 6:36 AM, Marc Nieper-Wißkirchen <xxxxxx@nieper-wisskirchen.de> wrote:
> > >
> > > In any case, when we want to support the reading and writing of
> > > records, some solutions (and probably the simplest ) require that a
> > > record-type registry must be maintained anyway:
> > >
> > >> (define-record-type foo (nongenerative ...) ...)
> > >> (gc)
> > >> (read)
> > > <external representation of record type foo>
> >
> > Please don’t force implementations to use a (centralized) registry.  Decentralization is necessary for code mobility.
>
> No centralized repository is needed; at some point, one just has to
> transfer the definition of a record type from node A to node B at some
> point.  This can happen by serializing a record-type definition and
> sending it over the wire, or one can use the Gambit approach by
> effectively sending the record-type definition each time along with a
> record.

PS: I found your comment very valuable; any lexical format for record
types should at least optionally allow to define the record type along
with it.  I hadn't thought of it before.