[comp.emacs] Trouble with TeleVideo 921 and EMACS???

Thomas.Epperly@EDRC.CMU.EDU (03/16/88)

I have been having some trouble using GNU EMACS on a televideo 921
terminal.  The problem comes when I use ^N or ^P to scroll the screen
(i.e. ^N off the bottom of the page or ^P off the top).  The screen gets
garbled and it gives me a search failed message when I have not initiated
a search myself.  Can someone suggest a change to the termcap or something
like that?  If so I am interested in any suggestions you have.
Tom Epperly
te07@edrc.cmu.edu
or te07@andrew.cmu.edu

jr@PEBBLES.BBN.COM (John Robinson) (03/25/88)

>> It seems that
>> the (set-input-mode nil t) doesn't always work because it is evaluated
>> before the emacs is dumped.

One of the unfortunate parts of the Unix terminal design is that the
state manipulated by stty is volatile, and not part of the program
(yes, yes, changing this is nowhere near as easy as I just made it
sound).  So executing this form before dumping won't have any effect
on the state your terminal aquires when you restart the dumped emacs.

Emacs would need more code to fix this.  It copes correctly with
terminal state over a suspend (^Z), so it could also in principle do
this around a dump.  Given the predilection expressed in the files
accompanying the GNU emacs distribution, I wouldn't count on this
appearing.

I would also question whether putting this into a dumped emacs is ever
the right thing.  Given a choice of modes, as you have with most
terminals, raw mode is what emacs prefers.  If you insist that your
site "default" be cbreak, you are penalizing the users with capable
terminals that wish to run in raw mode.  So leaving this as a user
customization is still, I think, the right thing.  Making the
information available to users that want the customization is fine,
and this news list helps do that (together with the PROBLEMS and
INSTALL files).

Another possibility would be something like novice that offers to
change your default environment (my mucking in your .emacs file) to
put in the necessary initialization.  Real cute would be detection of
^S^Q pairs in the input at early stages of an emacs session which
provoke an automatic novice interaction.  Maybe it could all be buried
in incremental search in fact.  Hmmmmm...

/jr