[comp.emacs] Is KSH ^T wierd enough to vote no on 1003.2a?

daveb@Ingres.COM (Dave Brower) (04/10/91)

Posix 1003.2a is balloting (qI get a vote), and one of the things it
is standardizing is the behavior of "set -o emacs" mode for command line
editing.  This would be great, except for the questionable historical
interpretation of control-T in the KSH.

(A) For reference, GNU emacs (and most of the others I've used) switches the
current and the previous character, moving the point forward.  This lets
you drag the previous letter forward.  It's also intuitvely  right for
when you realized a transposition after typing the second character that
should have been first.  With the point on '2' below,

	before		after
	1324		1234
          ^                ^
(B) In set -o emacs mode, KSH switches the current with the next character,
and moves forward.  This has the effect of letting you drag the current
character forward, but means to correct a transposition, you need to 
back up a character.  This is what Posix.2a proposes to do.

	before		after
	1324		1234
         ^                ^
(C) In set -o gmacs mode (which .2a is not supporting), control-T switches
the previous two characters without touching the current one.  So, with
the point on '3' below,

	before		after
	213		123
	  ^               ^

The question is whether the KSH behaviour so objectionable that it's
worth arguing about.  Keep in mind, you are going to be living with the
result in your shell for ever.

Trick part:  The ballot process is recalcitrant to toss out existing practice
and stuff from base documents without strong support.  Posix started
started from the KSH book, which documents what I consider odd
behaviour. It's therefore up to balloters to make the case that despite
the existing practice, etc., etc., that it really ought to be changed.
This might be a tough hill to climb.

Some info I'd like (as well as plain old opinions) is:

(1) Does KSH emacs mode really match Montogomery emacs ^T?
(2) Does KSH gmacs mode really match Gosling emacs ^T?
(3) How do CCA or Unipress handle ^T?
(4) Does the BRL shell with editing mode have an emacs mode with ^T?
    How does it work.

Please MAIL your thoughts to me (daveb@ingres.com), as I'm going on
vacation and won't be reading news until after the ballot deadline on
4/24.

thanks,
-dB