[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


Note that it seems to be impossible to mail to this
person's address.


Received: from MIT-MC by MIT-OZ via Chaosnet; 12 Jan 84 17:44-EST
Received: by UCB-VAX.ARPA (4.22/4.19)
	id AA05266; Thu, 12 Jan 84 14:42:50 pst
Message-Id: <8401122234.AA06362@ucbpopuli.CC.Berkeley.ARPA>
Received: by ucbpopuli.CC.Berkeley.ARPA
	(4.13/4.9.5) id AA06362; Thu, 12 Jan 84 14:34:41 pst
Date: 12 January 84 17:33-EST
Subject: emacs bugs

Wolfgang N. Naegeli
308 Fernow Hall
Cornell University
Ithaca, New York 14853-0188
(607) 539-7084

Dear Dr. Stallman:

I wrote to you on 22 March 1982 about some problems we have with
EMACS.  I also phoned and left a note for you with one of your
colleagues.  I realize that you certainly have many more important
things to do than responding to questions.  I tried very hard to
find solutions and found a few, but menawhile additional problems
have arisen.  I would very much appreciate if you could just help
me with the journaling problem.  All the rest is not really

Also, would you please send me a copy of AI memo 519a,
"EMACS, the Extensible, Customizable, Self-Documenting Display Editor."

Thank you very much!

Yours sincerely,
Wolfgang N. Naegeli

                          BUGS (?) IN EMACS

Journaling    does not work as described in the manual, according to
        which the journal file should be made fully up to date on the
        disk after every 50 characters.  The journal file is opened when
        I initiate my EMACS (as requested in my init file), but nothing
        is written to it until and unless I issue the M-X End Journal
        File command. But that is not what I want.  Occasionally I have
        had some characteristics of the EMACS environment listed that
        were set at startup, and maybe even the first few commands and
        characters of input, but usually there is nothing in the file
        at all when I check at the next logon, and I have not been able
        to determine under what circumstances that minimal recording occurs.

        Despite the fact that I have set the file retention code for the
        journal file to 3 in TWENEX, only the latest one is saved.

Mode 2 Timeout     in MODE2 library does not work.  If this command is
        executed, I get an error message at the first time of update:
        QNB Q-Register not buffer - attempt to select a string or number ?
        after that there is no further attempt at updating at argument

Display Mode Line Inverse (variable),     if set to 1 causes only the
        standard, but not the second mode line to be inverted.
        It probably as just been forgotten by the other of MODE2 library
        to check for the value of this variable.

Abstract Redefinitions  and ...

Wall Chart    abort at certain command characters that are supposedly
        undeterminable, such as ( which I have redefined to execute
         the keybord macro function from the VT100 library, which I
        have copied to my own library.

Make Page Permutation Table    does not check for the value of
        variable Page Delimiter.  Thus it is useless if anything other than ^L
        is used as a page separator.

Push to Exec,    if done from a secondary EMACS causes the system to go
        into half-duplex.  See example described under A) below.

Calling EMACS with the edit (ed) command leaves the main buffer empty
        and gives a second buffer if a file is specified as an
        argument of ed.  If used without argument, main buffer is

C-X C-V and calling EMACS with ed first displays the end of the buffer,
        then clears display and immediately reads in the beginning.
        Particularly on slow terminals this is anoying.  Is there a way
        to supress this display of the end?

Minibuffer Size (variable),    if set to -3 with library SLOWLY is
        ignored.  Minibuffers continue to display at the top instead of

M-O (Set Screen Size),   if invoked for the first time without argument,
        gives just 1 line instead of defaulting to 3 as the manual says.
        Even when used with more lines, at the next invocation it reverts
        to 1 instead of giving the same number of lines as when last
        invoked and as it should accroding to the manual.

        Also, with just a 1 line screen display C-V does no longer work,
        it does not scroll, though M-V still does.

ISPELL sends out control characters that prompt the Concept-108 to
        reply with its answerback message, which in turn confuses ISPELL.
        A staccato of prompts and answers follows.  The job has to
        be started all over again after manually setting the
        answerback memory to null.  A more detailed description
        follows under example B) below.

        Neither the EMACS manual nor the ISPELL info file tells how to use
        multiple directories from within EMACS.  When using the command

M-X Command to Spell,   I get the error message:
        ?First nonspace character is not a digit
        and am left at the ISPELL prompt.  After using quit to return to
        EMACS the display is not restored but has to be redisplayed manually
        to get back to the right cursor position.

Exiting (with M-X Save All Files set in the Exit Hook) or visiting a new file
        will prompt for saving the buffer, but if one is not certain and
        attempts to check first by trying to abort the command with C-G
        one is left in an indeterminate state with neither EMACS nor TECO
        commands working and the cursor just switching between the display
        and the Echo Area ater Escapes.

Save Word Abbrevs   variable, if set to -1, does not cause incremental
        abbrevs to be written on exiting.  The init file needs to specify:
        Write Incremental Word Abbrev File for this to occur.


Calling EMACS from MM causes modeline to be updated four times in a
        row, or more precisely: initial display plus 3 updates.
        Perhaps the first three updates could be suppressed to
        save time, especially on slow terminals.

System messages are important to know, but would be much less annoying
        if they were displayed in the echo area.  It would also be
        nice to have the most recent message stored in a Q-register in
        case it is displayed at a moment when a clearing of the
        screen is in progress and it is lost before one can read it all.
        (I realize that such an enhancement would have to be implemented
        for the most part in the system software, which would have to
        check which terminals are in an EMACS session, but I trust that
        the ingenious minds behind EMACS are the most likely to come up
        with such an innovative feature.)

System commands:  It would be nice to have an extended command that
        would take any system command as an argument (hopefully with
        completion) and display responses (if any) in the echo area.
        This would avoid having to push to exec and pop back for just one
        simple command, and would save a lot of time, particularly on slow
        terminals by avoiding the need to redisplay the screen when popping

It would be convenient to have an ISPELL setup hook, to automatically
        load certain dictionaries,  maybe even depending on the
        fileextension and a variable that is set when any of the
        loaded dictionaries is augmented during a correcting run,
        so that one could check that variable and issue the necessary
        commands to dump those augmented dictionaries with the EMACS
        Exit Hook.

Do you know of a library for the Concept 108 that makes use of its 8 screen
        memory?  That might save a lot of time if one is on a slow modem
        and particularly when scrolling forth and back a lot and when using
        two widows.

I pushed from EMACS to EXEC, then after some file manipulations
I invoked MM and when I had to edit a message I used ^E (which according
to my MM.INIT is EMACS [my customized, dumped WNNMACS to be precise]).
When I wanted to look at a help file, I pushed again to EXEC.  After
getting the @ prompt I first thought I got hung up somewhere because
there did not seem to be a response, until I realized, that my commands
were actually received, but not echoed.  I only got the system output
displayed on my terminal.  It was really just like the system had switched
to half duplex while my terminal was still in full duplex. So I could not
see what I was entering.  As soon as I poped back, everything seemed to
be back to normal.  This is certainly not a sequence that I would use
very often, but I am curioous what happened.  In summary what I did was
EXEC ==> WNNMACS ==push==> EXEC ==> MM ==^E==> WNNMACS ==push==> EXEC ==pop==>
MM ==exit==> EXEC ==pop==> WNNMACS ==exit==> EXEC

Before displaying a word that it does not know, ISPELL
sends RUBOUT C-R space space.  When it gives options, it sends
first RUBOUT C-Q C-D C-A.  That is before option 0.
It does not send control characters, except CRLF, before giving
additional options, if
any, but after those (before telling the line number where
the objectionable word was found) it does RUBOUT C-Q C-Q C-A.
If there are no suggestions, it sends RUBOUT C-Q C-D C-A RUBOUT
C-Q C-Q C-A anyways.  Finally, before prompting with "==>"