There is a 42's bug...
Message #5 Posted by Vieira, Luiz C. (Brazil) on 22 Feb 2003, 8:53 a.m.,
in response to message #1 by R Lion
... and J. F. Garnier (if I am not wrong) have already mentioned some experiences with it. Also, Graig Finset's HP42S database mentions something about it. I extracted part of the file: (may I, Craig?)
BUGS/ROM-Versions::
Bug/2 (as in 1/2 a bug): Clear REGS (do CLV "REGS") and execute \GS+.
RCL now recalls the contents of registers 00 through 120,930 (indirect
above 99). You convert a hexadecimal address to a register number by
dividing by 16 and adding a constant (183 on the reporter's system,
but may vary). Exit this mode by setting SIZE to any non-zero amount.
Interesting addresses:
1FFC2 to 1FFFF copyright notice
40300 to 40309 speed nybble, contrast nybble, clock
Do COMB or PERM with X > Y. You correctly get "INVALID DATA". Now do
X<>Y. Do COMB or PERM again. It should work. Instead, you get
INVALID message again. EXIT + \v/x does not clear this condition.
However, any arithmetic operation it! [reported by Rich Homard,
b645zjo@utarlg.uta.edu; posted to Comp.sys.handhelds on 29 Oct 1991]
John W. Loux adds: Some of the earlier HP42S calculators had this
problem. This problem was fixed at one point. However, in a later
HP42S update, the problem reappeared. It seems that a combination of
hardware and software caused the problem. When the hardware problem
was fixed, the software problem reappeared! As far as I know, there
are no plans to fix the "bug" on the basis of the simple workaround.
The problem resurfaced in late REV B calcs and all REV C --- somewhere
between #3011 and #3055 is my understanding.
Hope this helps.
Best regards.
Luiz C. Vieira - Brazil
| 
Go back to the main exhibit hall