I have discovered what I believe to be a problem in the second pass of
both the old and the new versions of the C compiler.  The problem is
attributable to the following ratfor source:

	#  cause /lib/c1 to fail
	#  W.E.Brown; 4-Nov-85

	program junk
	   parameter (TRIP = 0)
	   logical WantEv
	   integer x

	   WantEv = .true.
	   for (read *, x; x != TRIP; read *, x)
	      if (WantEv  .eqv. (0 == mod(x,2)))
		 print *, x
	   
	   stop
	end


When this is compiled via "f77 -d junk.r", the following results:

	ratfor  junk.r >junk.f
	f77pass1 -d junk.f /tmp/fort6926.s /tmp/fort6926.d /tmp/fort6926.x
	junk.f:
	   MAIN test:
	/lib/c1 /tmp/fort6926.x - /tmp/fort6926.a 
	line 8: No code table for op: !=

	compiler error.


Note especially the message about "No code table" -- this comes from
/lib/c1.

While it is certainly possible that f77pass1 is generating
something which /lib/c1 doesn't understand, I believe (from a somewhat
cursory inspection of the source of f77pass1) that the problem
does lie in /lib/c1.

I am writing to inquire whether this is a known bug, and whether there
exists a fix.  If not, I would ask that these symptoms be posted on the
net, with a request for assistance.  I just don't know the C compiler's
internals well enough to try to dope out this problem myself.