Text 1063, 88 rader
Skriven 2008-12-23 09:45:20 av Mike Luther (1:117/3001.0)
   Kommentar till text 1062 av Marc Lewis (1:396/45.0)
Ärende: Numlock oddity.
=======================
At the risk of be thought of as opknockshus, chuckle.

 ML> Only under eCS (2.0RC5), NOT under Warp4:

 ML> The Boxer Text Editor (OS/2 version, B2.exe) with the normal
 ML> default.key keyboard file loaded, the Numlock key works
 ML> backwards for the function keys surrounding the numeric keypad
 ML> (/, *, -, +).  When doing a line copy using Atl+L to mark a
 ML> line, the normal position (ON) on the Numlock key will cause
 ML> bizarre results when pressing the function keys to either cut,
 ML> copy or paste the marked line.

 ML> I have the startup numlock disengaged in the eCS desktop menu,
 ML> but that hasn't made any differnce.  When installed, eCS setup
 ML> was told it was a normal type of enhanced US keyboard.

 ML> This problem does not show up on my Warp 4 machines nor the
 ML> windoze machines with any program.

 ML> Any idea of where to look for a correction of this annoying
 ML> issue?

The eCS development crew.  More to the point, does this same issue show up if
you use a different brand of keyboard?  And, lastly, are you using some type of
KVM switching unit?  The reason I ask this question is that I have discovered
that some of the latest motherboards and even this latest Modular Industrial
Systems 9455 SMB system, are sensitive to any form of residual mouse or
keyboard voltage level which may be present on the cable plug-ins when the
power is actually turned off for these units!

For example, I needed a manual unit switching boxlette for test purposes for
this relay rack monstor in addition to the Rose KVM switch I use. OK, it swaps
the monitor and keyboard, but not the mouse lines between an older AMD 700Mhz
large rack box I use with mobile drive trays to circulate between OS/2 and
WIN-XP test drives.  That unit, when used with this manual hard line switcher,
actually transfers the voltage from one of the keyboard lines backwards to the
relay rack or other case unit when the rocker switch is set to the relay rack
unit and the AMD unit is still on!

You can actually see one of the red voltage monitor lights on MIS 7062 Passive
Backplane still red even though the complete power is gone from the Tri-Mag
power supply!  The light goes off if I switch away from the relay rack case
back to the AMD box.  Wierd.

OK, here is the issue.  If there have been any hardware changes made in the
passive backplane PCI slot devices or whatever, and that external tag voltage
was left on the keyboard line was still present, the whole MIS 9455 SMB will
*NOT* reboot at all!  You can momentarily pull the keyboard plug just before
you turn on the AC power line to the case, or,you can switch away from the case
back to the AMD unit when you first flip the power switch.  You can let the
power come up on the case and then flip the switch back to that case and
everything works fine!

And .. if there have been no changes in the hardware mix in the relay rack
case, you can actually power off and power on with the error voltage still on
the keyboard line this way,  It will still work.

However .. I long ago found out that there are other motherboards which have
very strange keyboard issues where this happens to discrete other boxes plugged
into the bench for work purposes on this same switch. Actually, some
motherboards won't even operate at all with the manual switch tied to the Rose
KVM operarations here are present together with voltage left on line from that
AMD box which is normally used for the number four position on the ROSE lines. 
The only way these motherboards will work is if you use a separate manual
keyboard plugged into the test box with its video with the Rose KVM, on that
manual switch too. Once the system is up you can swap back to the Rose KVM
keyboard and use it. But you absolutely cannot count on correct keyboard action
if it ever touches the rack switching system until after it is totally booted.

Now .. if you aren't using the manual switch, nothing has ever failed like this
when the Rose KVM is actually doing the whole switching job.

I sincerely doubt that you are seeing anything as wierd as this Marc,But things
do happen this way.  You are probably seeing a keyboard driver error in the
development work with the eCs 2.0 release.  And you might note.  From what I
think I have learned, there is a complete new Fix Pack involved with this 2.0
release, the escaped FP 6 that got out about a year ago.  Check the kernel and
keyboard driver dates.  I wouldn't be at all surprised to see that they are
newer than even those which are present for MCP2 latest everything.  If you
need to I can furnish the driver dates and file sizes here for comparison.


--> Sleep well; OS/2's still awake! ;)

Mike @ 1:117/3001

--- Maximus/2 3.01
 * Origin: Ziplog Public Port (1:117/3001)