[ksh93-integration-discuss] Small question about enabling the "multiline" mode...

[I. Szczesniak]

On 8/30/06, James Carlson <james.d.carlson at sun.com> wrote:
> Roland Mainz writes:
> > > The multiline option only works for terminals for which \E[A causes
> > > the cursor to move up one line.  I suspect that this is not the case
> > > for the terminal in your example.
> >
> > AFAIK all the terminal emulators shipped with Solaris (e.g. "dtterm",
> > "xterm", "gnome-terminal" and "konsole"), the native Solaris console on
> > SPARC and the Solaris/x86 console support that sequence (AFAIK it is
>
> Not if you switch to Tek mode.  ;-}
>
> > from vt100 or earlier (vt52 ?!)).
>
> No, not VT52.  That was pre-ANSI and used "ESC A" for the cursor-up
> command.  The VT100 and up used ANSI sequences, which include "CSI A"
> for cursor-up.  CSI can be rendered in 8-bit mode as hex 9B, or as
> "ESC [" in 7-bit mode.

I assume ksh93 could only use the 7-bit mode - the 8-bit mode would be
incompatible with multibyte output.
A quick fix may be to enable the multiline option in /etc/ksh.kshrc
only if if the terminfo command 'cuu1' is equal to  'CSI A':
$ infocmp | fgrep "[A"
        cup=\E[%i%p1%d;%p2%dH, cuu=\E[%p1%dA, cuu1=\E[A,

> The terminal emulators in Solaris are roughly VT102 (including
> scrolling commands that the VT100 didn't have).  But only roughly;
> there are many differences among the various implementations and
> between them and a real DEC VTxxx series terminal.  (Function key
> assignment and usage is one quite large area of difference.)
>
> In any event, it'd sort of be nice to use the system libraries to get
> the _right_ terminal sequences, just for those who are stuck using
> oddball devices.

David, is a reason why ksh93 can't use the system libraries?

Irek