Should NULL be allow to be in string?

R7RS working group likes discussion, I guess.

The most recent topic is about string containing #\x0 (or #\null).

poll: invalid item #315 from 5th ballot

The #315 is about the following code won't raise any error.

(string-set! s i #\null)

For me, it seems totally fine. But for them, or even Unicode world, it's not fine. Really?

Well, either way the above expression on Sagittarius will be totally fine for any time. So, I don't have any intension to change current behaviour.

The reason I'm writing this article is I found a curious experience in the topic. How does the following code work on R6RS implementations? Original was about Chicken and Gambit.

(import (rnrs))
(define file (string #\a #\x0 #\b))
(when (file-exists? file) (delete-file file))

(let ((o (open-file-output-port file)))
  (put-bytevector o (string->utf8 "hello"))
  (close-port o))

The file is definitely invalid file path.

The results are like this;

Implementation	Result
Chez	Made 'a' file
Larceny	Made 'a' file
Mosh	Made 'a' file
Racket	Raised an error
Sagittarius	Made 'a' file
Ypsilon	Made 'a' file

I can't say which way should implementations behave.