hakanson@ogicse.ogi.edu (Marion Hakanson) (02/05/91)
Anyone else done the above combination, or am I the unlucky pioneer?
I built with pretty much all default answers to Configure, except I did
use the "-YPOSIX" option to pick up the gid_t and uid_t typedefs. The
first time I made the mistake of using "-O -Olimit 2900" in order to
be able to optimize those big files (eval.c, toke.c & something in a2p),
and eval.c took something like more than 10 hours to optimize (the
poor optimizer got to about 28M in virtual size, only 12M of RAM).
Anyway, the result failed a few tests, so I rebuilt with just "-O",
since a couple of failures were in op.eval. This time I got exactly
the same results (included below) as before, although it took a heck
of a lot less time to find out (:-).
===============================
. . .
comp.cpp......../bin/sed: Can't find label for jump to ''
FAILED on test 0
. . .
op.eval.........1..10
ok 1
ok 2
ok 3
ok 4
ok 5
ok 6
not ok 9 0
FAILED on test 9
. . .
op.s............1..51
#1 :$x: eq :$x:
ok 1
. . .
ok 39
not ok 40
FAILED on test 40
# skipped 41..51
. . .
lib.big.........../lib/bigint.pl did not return a true value at ./lib.big line 2.
FAILED on test 0
Failed 4 tests.
===============================
Rebuilding without "-O" allowed all tests except "comp.cpp" to pass,
and that appears to be a failure due to an inadequacy of the Ultrix-4.1
"sed" command. Rats, and I like to use Perl to get away from such
stupid problems.
--
Marion Hakanson Domain: hakanson@cse.ogi.edu
UUCP : {hp-pcd,tektronix}!ogicse!hakanson