Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!princeton!orsvax1!pyrnj!caip!lll-crg!seismo!mcvax!cernvax!jmg
From: jmg@cernvax.UUCP (jmg)
Newsgroups: net.micro.cbm
Subject: C-Power 2.9 / Trim bug
Message-ID: <283@cernvax.UUCP>
Date: Mon, 3-Mar-86 06:27:19 EST
Article-I.D.: cernvax.283
Posted: Mon Mar  3 06:27:19 1986
Date-Received: Wed, 5-Mar-86 04:26:27 EST
Reply-To: jmg@cernvax.UUCP (Mike Gerard, DD Division, CERN.)
Organization: CERN, Geneva/Switzerland
Lines: 26
Keywords: C

Maybe this problem is known across the water, but it caused me pain and
suffering, not to mention complaints from the family that I was on that
bl--dy computer again. Anyway, with Trim as distributed on my C-power
version 2.9 compiler it seems that the statement
    if (!(bitval & 1))
goes the wrong way when bitval (an int) is hex 100, 200, 400 etc.
Some of you could verify it with

 main()
{ int bitval;
  char bit;
  for (bit=0 ; bit<15 ; bit++)
  { printf("bit = %2d, bitval = %4x\n",bit,bitval);
    if (!(bitval & 1))
      printf("!(%4x & 1) = %4x is true\n",bitval,!(bitval & 1));
    else
      printf("!(%4x & 1) = %4x is false\n",bitval,!(bitval & 1));
  }
}

It is correct before trimming, wrong after!

Also, does anyone know the exact circumstances under which one gets
a syntax error for something like

    i = x[j = y];