How to let racket-mode handle language extensions (the #lang directive)?

Question

I was wondering how to let racket-mode handle racket language extensions properly?

I tried the following file from the link above (test.rkt):

#lang datalog

ancestor(A, B) :- parent(A, B).
ancestor(A, B) :-
  parent(A, C), D = C, ancestor(D, B).
parent(john, douglas).
parent(bob, john).

ancestor(A, B)?

After C-c C-z, (which launches the REPL with ; Welcome to Racket v8.5 [cs]...), I was able to load the extension if I send the whole buffer (C-c C-c, or racket-run-module-at-point) to the REPL.

> 
ancestor(bob, john).
ancestor(john, douglas).
ancestor(bob, douglas).
test.rkt> 
; Closing REPL session because language's current-read-interaction returned EOF

Process *Racket REPL </>* connection broken by remote peer

But I cannot execute individual regions (and the REPL is terminated). If I just do C-c C-r (racket-send-region) on the first line (or few lines), including the #lang directive, I get an error:

; :2:0: read-syntax: `#lang` not enabled
;   possible reason: not allowed again inside a module that already starts `#lang`, or not enabled for interactive evaluation
; datalog: undefined;
;  cannot reference an identifier before its definition
;   in module: top-level
> 

How to configure Emacs racket-mode in order to make racket-send-region work in above?

(This is with Emacs 28, racket-mode from MELPA, under Ubuntu 22.04 LTS)

Answer 1

This probably isn't a configuration issue but an issue with #lang datalog and how it interacts (or not) with racket-mode. There are some more detailed discussions about it here and here which could provide another angle.