bug#11421: call-process behavior wrt. to read only buffers depends on coding system for read operations

[Lars Ingebrigtsen <larsi@gnus.org>]

Rob Browning <rlb@defaultvalue.org> writes:

> It appears that depending on the LOCALE setting, Emacs 23.4 may or may
> not signal a "buffer is read-only" error.  At least here, this can be
> demonstrated as follows.
>
> Save to ./test.el:
>
>   (let ((buf (get-buffer-create "xxx")))
>     (with-current-buffer buf
>       (compilation-mode))
>     (let ((coding-system-for-read 'iso-latin-1-unix))
>       (condition-case err
> 	  (progn
> 	    (call-process "/bin/cat" "/proc/cpuinfo" buf t)
> 	    (message "iso-latin-1-unix: No call-process error"))
> 	(error
> 	 (message "iso-latin-1-unix: call-process error: %s %s"
> 		  (car err) (cdr err)))))
>     (let ((coding-system-for-read 'utf-8-unix))
>       (condition-case err
> 	  (progn
> 	    (call-process "/bin/cat" "/proc/cpuinfo" buf t)
> 	    (message "utf-8-unix: No call-process error"))
> 	(error
> 	 (message "utf-8-unix: call-process error: %s %s"
> 		  (car err) (cdr err))))))

[...]

> which should produce something like this:
>
>   iso-latin-1-unix: call-process error: buffer-read-only (xxx)
>   utf-8-unix: No call-process error

When I try this in Emacs 27, I get:

Waiting for process to die...done
iso-latin-1-unix: call-process error: buffer-read-only (xxx)
Waiting for process to die...done
utf-8-unix: call-process error: buffer-read-only (xxx)

Can you still reproduce this bug on modern Emacs versions?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no