Re: Sporadic problems with *.KBD defs under OS/2

[Robert Holmgren holmgren@xxxxxxxx]

** Reply to note from xywrite@xxxxxxxx Mon, 14 Dec 1998 09:23:26

> I dropped an NI into every nn= line in my KBD file--except for those in
> the Alt table, which has the text macros, where for whatever reason the
> calls to the macros sometimes get fouled up if the NIs are there.

It's worth noting that when NI was first introduced, XyQuest
specifically intended it for ALT keys, or TABLEs that include the ALT key
+ another key(s) -- and nothing else.  So to exclude it from ALT keys is the
opposite of the intention. What exactly gets "fouled up"?

Also, I'd try another tack, which is to not use the names ALT or CTRL in
your TABLEs. For example, change ALT to ALLT, or CTRL to CTL (or LCTRL
and RCTRL, if you separate them), in a universal fashion -- still using
NI, to cover all bets. Also try LSHIFT and RSHIFT, so that no bare
"SHIFT" appears in any TABLE. That's the way I'm set up, I've tried to
insulate myself from any potential DOS intercept -- and I'm not
experiencing any keyboard problems whatsoever since the beginning of time
(or as long as I can remember anyway). If I _want_ DOS|OS/2 to intercept
the keystroke, say with Alt-Escape or Ctrl-Escape or Alt-Home, I just omit
that spec in my KBD file by commenting it out, e.g.:

;1=

A third thing that I do -- and I think this is crucial -- is to state
all permutations of multiple-key TABLEs. In other words, I don't just
say:

TABLE=ALLT+LSHIFT

Instead, I say:

TABLE=LSHIFT+ALLT,ALLT+LSHIFT

This _works_ (for me, at least): cascaded (arpeggiated) either way (Alt
first, or Shift first), or absolutely simultaneous. I just tried it about
fifty times, and I cannot get an error EXCEPT if arpeggiation is
elaborately delayed, i.e. I wait a second or two between the Alt and the
Shift keystrokes. I'm running v4.017, OS/2 v4 FixPack 9, a variety of
machines (right now a P2/266), with Avant Stellar and OmniKey 102 keyboards

Try implementing this strategy. (Make a backup of your KBD file first!)
Let me know your results.




---------
Robert Holmgren
holmgren@xxxxxxxx
---------