[Gcl-devel] humble q? about 'listen'

[Peter Wood]

Hi

The 'humble' in the subject line is because of my stupid 'bug' report
about #'sleep :-)

Anyway -

My question is: Is #'listen reliable on x86 Gnu/Linux?

In the following if #'prep-term and #'deprep-term put the terminal
into and out of cbreak, and (|u7|) sends a cursor position request to
the terminal: (on an xterm this is the chars #\^[ #\[ #\6 #\n).  The
format of the return is like a scanf format 'u6=\E[%i%d;%dR' where %i
means decrement the 2 numbers signified by %d, and the other chars are
literals ie #\^[ #\[ 'decf 'number #\; 'number #\R ...

(defun test-3 ()
  (prep-term)
  (|u7|)
  (unwind-protect
      (do ((ret nil (push (read-char) ret)))
          ((char= (peek-char) #\R) (progn (read-char) (nreverse ret))))
    (deprep-term)))

(defun test-4 ()
  (prep-term)
  (|u7|)
  (unwind-protect
      (do ((ret nil (push (read-char) ret)))
          ((not (listen)) (nreverse ret)))
    (deprep-term)))

It appears that #'test-4 unpredictably sometimes returns only a list
containing the first char of the terminal response (actually
#\Newline, since that is waiting in the input) and sometimes returns
the correct list of (eg: (#\Newline #\^[ #\[ #\2 #\4 #\; #\1 #\R) )

#'test-3 appears to be reliable and returns eg (#\Newline #\^[ #\[ #\4
#\9 #\; #\1) - obviously the numbers will vary.

>(term::test-4)

(#\Newline #\^[ #\[ #\1 #\3 #\; #\1 #\R)

>(term::test-4)

(#\Newline #\^[ #\[ #\1 #\7 #\; #\1 #\R)

>(term::test-4)

(#\Newline #\^[ #\[ #\2 #\1 #\; #\1 #\R)

>(term::test-4)

(#\Newline #\^[ #\[ #\2 #\5 #\; #\1 #\R)

>(term::test-4)

(#\Newline #\^[ #\[ #\2 #\9 #\; #\1 #\R)

>(term::test-4)

(#\Newline)

>^[[33;1R

#'test-3 can run for a long time without problems (100000 invocations
where the cursor gets sent to a random position on the screen and a
cursor-position-request is issued.  Maybe more, I got bored :-)) 


Is it a problem with #'listen?  The docs say it might not work on some
platforms due to lack of support in the OS.  Is Gnu/Linux amongst
those?  Maybe my usage is incorrect?

Regards,
Peter

(and sorry about the long post)