Re: terminfo change for screen/eterm?

[G. Branden Robinson]

At 2023-01-02T12:23:37+0100, Sven Joachim wrote:
> On 2023-01-02 05:15 +0000, Sam James wrote:
> 
> > In Gentoo, https://bugs.gentoo.org/698318 was reported to us where
> > a patch (https://698318.bugs.gentoo.org/attachment.cgi?id=593684).
> > was given to avoid Emacs malfunctioning within screen.
> >
> > It seems Debian's been applying this for a while (and still is).
> 
> The rationale behind it seems to the Debian keyboard policy[1], which
> was put into place somewhere around 1997/1998 after much discussion
> about the backspace and delete keys on the debian-devel mailinglist.

Yes.  The blame for Debian changing its xterm terminal description to
configure ^? as the delete character can be laid at my doorstep, I
think.

> Basically its most important point is "backspace key must not invoke
> the help menu in Emacs", which as a long-term Emacs user I totally
> agree upon.

As I recall, ^? is also what actual DEC VT-series terminals emit when
the key-that-is-in-the-backspace location is pressed.  (Unfortunately,
my VT520 and 525 have been in storage for years.)

But--and Thomas can surely correct me here--I think X Consortium or even
MIT Athena xterm (the program goes back to 1984, before X itself), while
otherwise emulating a VT100 (or thereabouts), deliberately used ^H for
this key instead.

In my opinion it made sense 25 years ago to restore alignment with DEC
VT terminals on this point, and still does today.  xterm spent several
years being fairly indifferently maintained until Thomas picked it up.
The "backspace color erase" debacle is a case in point, where Thomas's
predecessors screwed up both xterm and ncurses by bashing on the code
like ignorant apes.

> Whether these changes would be appropriate upstream I do not know,
> since screen runs on many other systems besides GNU/Linux.

That one's above my pay grade; it seems to me like a lot depends on
the intended and actual architecture of a terminal multiplexer--how
transparent does it try to be?  If perfectly transparent, you don't
_need_ a terminal type for screen (or tmux).  But since the terminfo
database has much material for both, I guess perfect transparency has
proven intractable.

I used to use screen constantly, but I gave up on it some years ago
because that lack of perfect transparency concealed terminal behavior
from me that I really needed to see in my capacity as a groff developer
(q.v. grotty(1)).

Good to see Dan Jacobowitz is still around.  I still remember how he got
my then-wife's PowerMac 7200 booting Linux for the very first time on
the show floor at LWCE-NY 2000.  I hope this email finds you well!

Regards,
Branden

signature.asc
Description: PGP signature