Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.csd.uwm.edu!gem.mps.ohio-state.edu!ginosko!uunet!hsi!aati!mlfarm!ron
From: ron@mlfarm.UUCP (Ronald Florence)
Newsgroups: comp.emacs
Subject: broken help functions
Message-ID: <290@mlfarm.UUCP>
Date: 29 Aug 89 03:10:39 GMT
Organization: Maple Lawn Farm, Stonington, CT
Lines: 58

I would very much appreciate help with the following rather bizarre
problem:

We recently compiled a new version of GNU Emacs 18.50, running under
Xenix 2.2.3 on an IBM ps2/80, and have begun to encounter intermittant
strange behavior from the C-h functions.  For example, sometimes if I
enter "C-h b" with debug-on-error turned on, the command fails and the
debugger backtrace comes up with

     Signalling: (wrong-type-argument sequencep boundp)
       describe-bindings()
       call-interactively(describe-bindings)

Another time, I entered "C-h l" and got:

     Signalling: (wrong-type-argument sequencep t)
       substitute-command-keys("Type \\[delete-other-windows] to remove help window.")
       byte-code("..." [standard-output function t get-buffer-window funcall message substitute-command-keys one-window-p "Type \\[delete-other-windows] to remove help window." "Type \\[switch-to-buffer-other-window] RET to restore old contents of help window."] 6)
       print-help-return-message()
       byte-code("..." [standard-output nil t "*Help*" princ key-description recent-keys move-to-column 50 search-forward " " newline print-help-return-message] 10)
       view-lossage()
     * call-interactively(view-lossage)

The elipses represent strings of 8-bit octal characters.  If I do not
have debug-on-error set, the minibuffer messages when the help
functions fail are generally "Wrong type argument: sequencep, S" where
"S" can be all sorts of lisp symbols.  Frequently the message will
include "#<EMACS BUG: ILLEGAL DATATYPE (#o N)..." where "N" can be
anywhere from 012 to 037777777777.

When "C-h l" did not fail and I got a lossage list, it was bizarre:

     [ ESC B [ ESC B [ ESC A [ ESC A [ ESC B [ ESC B [ ESC 
     B [ ESC t p l v l C-x RET [ C-h C-x C-] C-s b C-x RET 
     k C-x RET k ESC F ! C-h C-x C-h C-x C-h 4063275 4063280 
     4063280 4063280 6699672 6094941 5505108 4915275 4325442 
     3735609 4063275 4063280 4063280 7289496 6684774 6094941 
     5505108 4915275 4325442 3735609 4063275 4063280 7879320 
     7274607 6684774 6094941 5505108 4915275 4325442 3735609 
     4063275 27001244 26411411 25821578 26149244 26149249 
     26149249 26149249 -33539432 32965111 32375278 31785445 
     31195612 30605779 30015946 29426113 29753779 29753784 
     29753784 -29345128

When a help function fails, print-help-return-message() often also
fails with messages like "Type C-x 971 to remove help window" or "Type
C-x 0 to remove help window".

It appears that none of the C code or lisp code (including help.el)
has been changed, and several recompiles have not cured the problems.
I have discovered no problems with any lisp functions other than these
intermittant problems with the help functions.

I'm stumped, and would very much appreciate suggestions by email.
Thanks in advance.
-- 

Ronald Florence			...{hsi!aati,rayssd}!mlfarm!ron