hpux and /lib/pa1.1/libm.a ; markets graphs

[Rick Riolo]

Yes, hpux has more than one version of libm.a .
I'm not sure I understand the full story, but as I read
things the pa1.1 version has things like fabs and rint
which were not officially ANSI standards, just in some
"future directions" specs (at least that was so whenever hpux did this 
work). so they put these in a special version of libm.a. 
But I have also seen news items about how some systems
need patches, otherwise rint won't work right!!
(so first you apply the patch, then relink everything
that uses it...argh!).
So...you might want to check out your rint to be
sure it works...supposedly when it doesn't, it just
gives the same huge value for everything, so its easy to tell.
(or so they say....)

Anyway, since the /lib/pa1.1/libm.a is supposed to be
a superset of the usual libm.a, I think Nelson's 
suggestion  is probably best for the hpux default lines in the 
Makefile.conf, i.e, just OTHERLIBDIR=-L/lib/pa1.1/
(but I haven't tried it yet...).
(For hpux's cc on 700 series machines, that is the libm
it uses by default.)

re apps/market graphs:
When I run it on my hpux displaying on its own screen,
some of the graphs data parts are about 0.5cm 
wide---unreadable, while the part that displays the correspondence
between colors (the squares or the lines) and variables
are readable.
Note that the Price vs Time graph is ok: the data portion
is a couple of inches wide, so one can see it the lines vs t.
Its only the other two that have the data portion
scrunched horizontally.

Just now I ran apps/market on the same hpux as above, but
displaying on another screen (on a decstation), and the
data portions of the frames are now about 1.5cm wide...that
is for the histogram Usage of predictors, its readable just fine,
for the accuracy vs t, I can now see the lines, but they
are still squeezed into a small horizontal space
(and of course the numbers labeling time are all written
on top of each other).

Its all rather odd.  I ran a bunch of the blt test
and they see to all look ok.
I especially don't understand why the data parts would
be 3x bigger on my dec screen than on the hpux screen
(the screens are about the same size).

Anyway, I'm not sure this should be on the top of 
your list (at least not until maybe someone else
using hpux can try this out, or until I can try
to write soemthing of my own to exercise these graphers.)

 - r

Rick Riolo                       address@hidden
Program for Study of Complex Systems (PSCS)
1061 Randall Lab     University of Michigan
Ann Arbor MI 48109-1120
http://pscs.physics.lsa.umich.edu/rlr-home.html