Jeff Kell writes:

> Tabs are largely related to the curses library, where much strangeness
>  is seen between the 3000/9000 (with most any emulator) and get a bit
>  more strange with QCTerm in the 3000/9000 environment.  For the best
>  example of this, use a non-standard terminal screen layout, e.g., 96
>  columns and 48 rows.  Reflection on the 9000 behaves correctly (the
>  curses initialization apparently figures this out and presents the
>  proper status response) while QCTerm does not.  Neither work correctly
>  on the 3000.  You will see this in vi, more, patchix, etc.

But I also received the following privately:

=======================================

As far as i can see the following should happen after logging in:

- tset in /etc/profile or $HOME/.profile is used to set the
terminal type (telnetd should handle getting the the terminal type
for a telnet connection) and do some basic initialisation of the
terminal (as defined in terminfo).  This includes setting up the
tab-stops (HP-UX also has a separate 'tabs' utility for this).

- When a curses application is started (such as vi or HP Eloquence)
it will set the terminal into "curses mode". The ESC sequence
is defined terminfo (smcup and smkx).

- When a curses application is ended, the terminal is reset to
normal mode (rmkx, rmcup).

On a related note: The Linux implementation of the curses library
(screen handler) has bugs with hp terminals (or more general with
terminals with have the xmc#0 setting and use invisible switch
characters). As a result, screen enhancements may be mangled in some
cases. We had a look in the curses source code but the fix is not
easy.

The same used to apply to some curses versions on HP-UX (they should
work almost correctly these days).

======================================

The bottom line of this problem seems to be that there are significant
variations, O/S version to version, in regard to how the Linuxes/Unices of
the world react in this regard. As always, we are ready and willing to adjust
QCTerm's behavior to whatever is the most common convention -- if we can
determine what that most common convention is.

In that regard, it does us no good for you to write and simply say that
QCTerm isn't working properly. To fix a problem, what we need is a
step-by-step list of keyed inputs and access to a machine that is causing the
problem to be made visible. If we can get these, we're not only pleased to
fix the problem, we can generally find it and correct it in a relatively
short time. Most problems we are encountering now are tending to be very
simple things, most of which are extradefinitional -- essentially cultural --
more than they are bugs. The terminal type negotiation is characteristic of
that kind of problem. It certainly was never part of any HP terminal design.
Rather, it was imposed by the nature of the internet, and even there, it
doesn't seem to be consistently done.

Nonetheless, keep those cards and letters coming in (along with complete
explanations of the problems) and we'll keep adjusting QCTerm.

Wirt Atmar

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *