Re: Making the keyboard work

[Thor Lancelot Simon <tls%rek.tjls.com@localhost>]

On Sat, Feb 21, 2009 at 04:55:10PM -0500, Greg A. Woods wrote:
> At Fri, 20 Feb 2009 22:50:20 +0200, Alan Barrett wrote:
> Subject: Re: Making the keyboard work
> > 
> > I recommend option 2: change wscons to behave like xterm-color (or some
> > other suitable well-known terminal type).  However, I won't be doing the
> > work.
> 
> I would prefer not to have yet another implementation of an attempt to
> mimic some old terminal that had too many of its own old bugs and
> quirks, even if that intermediate implementation is now widely known and
> used (as Xterm is).

Yes, this was my objection to "option 2": that xterm and xterm-color
have wildly variant termcap and terminfo entries on different systems
and this is particuarly ugly with an ANSI terminal with more than 24
lines (which is precisely what wscons is, by default) because this can
lead to the cursor being stuck in row 25 and other major annoyances (sigh).

> Further, if sshd allows $TERM to be set by the client, then perhaps it
> could also allow $TERMCAP to be set by the client as well?

This was in fact my proposal -- though actually I would like to see
login (possibly via login.conf) set $TERMCAP from $TERM if $TERM is
set when login runs.  The problem is that our ssh client doesn't
forward $TERMCAP by default, just $TERM (and $ROWS and $COLUMNS, which
is why xterm users don't notice other systems' busted xterm termcaps
too often).  But we could fix that easily enough.

I don't know what remote systems' sshds are likely to with $TERMCAP;
and I'm not certain sysv-curses systems like Solaris will do anything
with $TERMCAP even if they receive it; I don't have one handy to test.
Meanwhile, we don't have a source of terminfo entries to feed into
$TERMINFO.