Re: HOWTO keymap seperate users -- LINUX5250


On Wed, 7 Nov 2001, James Rich wrote:

> On Wed, 7 Nov 2001, Scott Klement wrote:
>
> > It seems to me that when you change it in .Xdefaults (or for that matter,
> > in /usr/local/share/tn5250/XTerm which is the .Xdefaults script that
> > xt5250 loads)  that you're actually changing the definition of what
> > "red" looks like, and what "yellow" looks like, etc.   when the code in
> > cursesterm.c says "yellow" you're printing green, etc.
> >
> > It seems to me that encoding this in .tn5250rc would be a better option.
>
> How would putting it in .tn5250rc be different?  Wouldn't both be changing
> the definition of "red" etc.?

Errr..  sorry, I wasn't very clear.   Right now, we're saying "when the
AS/400 sends the code for "GREEN", we'll print the curses color "GREEN",
and then when you put it in your x resources, you're saying "when curses
says "GREEN", print the color blue".

I guess I was trying to avoid changing the curses definition of green (but
still changing the AS/400's definition of green)

I don't like the idea that on a console-mode screen it'll print as green,
and with the same code it'll print as blue in an xterm.   It seems
reasonable to me that a system should be devised that works with any
future "terminal" object that we supply consistently.

>
> > (Unfortunately, there is no documentation, besides the mailing list, on
> > how to create a .tn5250rc -- this REALLY needs to be taken care of!)
>
> Okay - this is going into the HOWTO as well.  Really the HOWTO is more of
> a tn5250 HOWTO and not just a keymapping HOWTO.

Thank you... IMHO, this is important :)

>
> > The other problem with using xresources for this is that you'll need a
> > different config option when using console-mode tn5250, and yet another
> > option when using gnome-5250 or gtk-5250 (if they ever get finished)
> > and yet another (??) when using the SLang terminal object...
> >
> > Really... putting it in .tn5250rc seems like a better idea :)
>
> Except that we already have keymaps in .Xresources.  And the reason they
> are there is because (I think) the keymapping is xterm related (I don't
> know about console mode).  .Xresources really influences xterm - not
> tn5250.

Yep.  The issue is that we do not have the capability of reading the
keyboard directly in tn5250.  All we can read is the escape codes sent by
the terminal.

If someone logs on to your Unix box using a modem or serial terminal, or
telnet session, or ssh, etc...  all it receives from the other end of the
connection are the terminal escape codes.  It cannot directly read the
keyboards scan codes.   I would be possible for the local keyboard to get
the actual scan codes, but to do that in "console mode" would require
different code for each type of Unix which is undesirable.   Since, at
this time, we really only have a console-mode emulator (until someone
finishes the GTK versions) we're stuck with that way of working.

Consequently, rather than tell tn5250 what to read from the keyboard,
you have to tell xterm, since it's the program that's working directly
with the keyboard.

> Colors could be done the same way (that's why *pointerColor and
> *cursorColor work right now), by influencing xterm.  I'm just thinking
> these would be easy, simple changes that could be complete in an afternoon.
> Does xterm have a 'highlight' attribute or other attributes that could be
> used to define what colors certain characters should be?  I know there is
> a bold attribute (man pages are bold), so if tn5250 were to tell xterm
> "use the bold attribute here" and *boldColor (or whatever) was defined in
> .Xresources to be white, then we have a simple solution that works today.
> It may not be completely portable across console, gnome, Slang, etc. but
> it would work without much change.

Sure.  and you can use that now without having a discussion about it. :)
But, a method that works consistently across all interfaces seems like
something we should work towards when time permits, don't you agree?

> I don't know about gnome/KDE, but don't they use Xlib?  If so then they
> should be able to use .Xresources.

They probably can.  But then you need code in those emulators to read it,
and that means that tn5250 (at least the gtk versions) would need to read
it's configuration from two different config files.  Doesn't this seem
more complicated and awkward?

>
> > Should this be my next project?
>
> If you've got the time, we've got the crime... (apologies to the
> Bloodhound Gang)
>

Don't worry about me running out of things to do, because that's not
likely to happen :)   I'm just trying to prioritize...