Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!gamma!epsilon!zeta!sabre!petrus!bellcore!decvax!genrad!panda!talcott!harvard!cmcl2!rna!cubsvax!winston
From: winston@cubsvax.UUCP (Ken Winston)
Newsgroups: net.lang.f77
Subject: Errors in f77 -O option
Message-ID: <446@cubsvax.UUCP>
Date: Wed, 5-Mar-86 07:13:46 EST
Article-I.D.: cubsvax.446
Posted: Wed Mar  5 07:13:46 1986
Date-Received: Fri, 7-Mar-86 07:57:10 EST
Organization: Columbia Univ. Bio. CG Fac., NY
Lines: 29


It is well know that there are bugs in the optimizer in the f77
compiler (ours is on MicroVax II ULTRIX). I have seen various scary
warnings about these errors, but no actual list. I was beginning to
think the whole thing was overblown when I ran into one the hard way
(I only lost a day tracking it down).

Does anybody have a list of the -O problems on f77? Which ones (if any)
are fixed in the new version of f77?

Here is a program that acts differently when compiled with/without -O.
The "0." gets a little too optimized in the write statement, and
disappears altogether.

	dimension array(20)
	open (1,file='binar',access='direct',recl=80)
	do 10 i=1,10
10	array(i)=i
	write(1,rec=1) (array(i),0.,i=1,10)
	close (1)
	open (1,file='binar',access='direct',recl=80)
	read(1,rec=1) array
	write(6,*) array
	stop
	end

Ken Winston				Winston Capital Systems, Inc.
...{cmcl2,rna,cubsvax}!wealth!ken	250 West 75th Street, Suite 3A
+1-212-382-3450				New York, New York  10023-1757