"1) The "Factors of Denominator" mode sometimes produces denominators
which are not factors. In these cases, the 32SII seems to display
the "Most Precise Mode" fraction, so this is a minor bug. Example:
SF 8
CF 9
8 /c
3
1/x

You see 0 1/3, but 3 isn't a factor of 8.

2) The "Fixed Denominator" mode (SF 8, SF 9) sometimes produces
denominators other than the specified one, along with incorrect
numerators. The amount of the error can be about +/- 1. HP is now
sending a manual addendum recommending that Flag 9 not be used at
all. Example:

SF8
SF 9
1
ENTER
ENTER
ENTER
8 /c
32
1/x (which is 0.03125)

You see v 0 7/8 (which is 0.875) but should see ^ 0 0/8. (Recall that
^ and v are the up- and down-arrows.)

Now press + repeatedly. You'll see:

v 1 7/8
v 2 1/2
v 4
v 4 3/4
v 5 3/5
v 6 1/2
v 7 3/7

etc... all of which are WAY off, and very few of which have a "fixed"
denominator of 8.

3) The "Most Precise" mode (CF 8, CF 9) and the "Factors of
Denominator" mode (SF 8, CF 9) sometimes violate the commutativity of
addition, and produce incorrect results. This only occurs when the x
value is close to 1/8 of the /c value, and the last two digits of the
LASTx register's exponent are nonzero. The worst error this can
produce is approximately +/- 0.125, which makes this a substantial
bug. Example:

CF 8
CF 9
8 /c
32
1/x

See v 1/8, but it should be ^ 0 0/8. Pressing:

0
+

gives ^ 0, but:

32
1/x
0
x<>y
+

gives v 0 1/8 . Addition should be commutative. The LASTx value is
causing the problems.

In late March, I called HP's Technical Support department about
problems with the fraction system (actually, at that time, I only knew
about bug 1). The HP employee stated that the 32SII can give
incorrect answers when using fractions (although she would not go into
any details about what problems were known), and that HP would likely
be instituting some type of exchange program this summer. She
suggested I call back after April 1 for details. I did call back,
left a message, and haven't heard back from them yet. "