bug: missing finish-output after writing prompt in REP

[hoehle@zeus.gmd.de (Joerg Hoehle)]

Hi,

debug.d:read_form() should call FINISH-OUTPUT after writing the prompt.

The following doesn't work nicely:
> (make-buffered-output-stream #'(lambda (s) (princ s *terminal-io*)))
#<BUFFERED-OUTPUT-STREAM #<CLOSURE :LAMBDA (S) (PRINC S *TERMINAL-IO*)>>
> (setq *standard-output* *)
#<BUFFERED-OUTPUT-STREAM #<CLOSURE :LAMBDA (S) (PRINC S *TERMINAL-IO*)>>
()  ; <- my input
>   ; <- the prompt
NIL ; <- the output

I'm wondering why it worked so far.  I suppose that the std C library
flushes stdout itself before reading something from stdin or that no
buffering at all was involved.

Here's an *untested* diff to read_form():
*** debug.d	Mon Apr 15 16:29:35 1996
--- /tmp/debug.d	Tue Mar 11 14:11:43 1997
***************
*** 27,30 ****
--- 27,31 ----
  #         (terpri ostream)
  #         (write-string prompt ostream)
+ #         (finish-output ostream)
  #       )
  #       (let* ((eof-value "EOF")
***************
*** 77,80 ****
--- 78,82 ----
          #endif
          write_string(ostream_ptr,STACK_2); # (WRITE-STRING prompt ostream)
+         finish_output(STACK_4); # (FINISH-OUTPUT ostream)
        }
      # Prompt OK
Rather use finish_output(*ostream_ptr) ??


I'm still using CLISP-1996-05-30, but the CHANGES.LOG file doesn't
mention anything about this bug, so I must assume it's still there.

	Jo"rg Ho"hle.
Joerg.Hoehle@gmd.de		http://zeus.gmd.de/~hoehle/amiga-clisp.html