Help! Perl fails op.int test

[Gordon Burditt]

In article <1162 at devon.UUCP> paul at devon.UUCP (Paul Sutcliffe Jr.) writes:
>I've recently ported PERL to a Tandy 6000 (MC68000 CPU, Xenix/68k 3.0).
>Perl passes every test in the t/ directory *except* op.int, which tests
>the use of perl's int() function.  All four tests in op.int fail.  It
>appears that the crux of the problem is that "int(1.234)" does not
>return 1.000000 exactly.  The following will illustrate this:
>
>    $ perl -e "print int(1.234);"
>    1.0000002407468855
>    $
I get:
% perl -e 'printf "%50.40lf\n", int(1.234);'
        1.0000000000000000000000000000000000000000
but I'm using the 3.2 libc.a

>In short, what can I do to get perl's int() function to behave?
Fix modf().  See below.
>Or am I (quoting the man page) "at the mercy of [my] machine's
>definitions of various operations such as type casting, atof() and
>sprintf()."
You are.  But this bug has a fairly simple fix. (The 3.2 version already
has this in it).

# cd /tmp
# copy -m /lib/libc.a /lib/libc.old		(Just in case)
# ar xv /lib/libc.a modf.o
# adb -w modf.o
_modf+0x9e?x
_modf+158:	0x4241
		^^^^^^ If you get some other number, do not proceed.
_modf+0x9e?w 0x7200
_modf+158:	0x4241	=	0x7200
^D
# ar rv /lib/libc.a modf.o
Do not run ranlib on /lib/libc.a.  If you did, or you get complaints about
__.SYMDEF being out of date, you can undo the damage by running:

# ar dv /lib/libc.a __.SYMDEF

Convince yourself that the new version works, then remove /tmp/modf.o and
/lib/libc.old.

					Gordon L. Burditt
					...!texbell!sneaky!gordon

The preceeding information is provided by an individual, and it comes
with no guarantee, warranty, or support whatsoever, by any individual
or organization.