2.9bsd floating pt. simulation library problem

[Stan Tazuma]

We are having problems getting the floating point simulation library
to work on a PDP 11/23 running 2.9bsd.  We use the cc -f flag as
documented, but usually programs core dump or print out wild numbers.
Of the 2.9 programs which use floating point, none of them
can be compiled to work like the binaries off the tape (e.g. df, w,
lastcomm, etc.), except for la.c and la.

How is it that the distributed binaries work fine, yet we can't
compile the source code and reproduce the binaries (this is really
a more general problem that applies to non-fp programs, too) ?

Has anybody looked into the floating pt. problem and maybe found a
solution to it?  So far, I've found that very simple programs
using f-p will work correctly with f-p simulation.

Please followup to net.bugs.2bsd or respond to me via mail.

Thanks much,
Stan Tazuma
Teltone Corp.
	...uw-beaver!tikal!stan