Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!think!ames!sri-spam!mordor!lll-lcc!lll-winken!csustan!helium!scott
From: scott@helium.UUCP (Scott Hazen Mueller)
Newsgroups: comp.sys.ibm.pc
Subject: Re: 386 unix (was Re: 386 demand paged virtual memory)
Message-ID: <185@helium.UUCP>
Date: Fri, 25-Sep-87 12:07:40 EDT
Article-I.D.: helium.185
Posted: Fri Sep 25 12:07:40 1987
Date-Received: Sun, 27-Sep-87 07:50:45 EDT
References: <125@snark.UUCP> <299@nuchat.UUCP> <358@netxcom.UUCP> <158@splut.UUCP> <380@micropen>
Reply-To: scott@helium.UUCP (Scott Hazen Mueller)
Organization: City of Turlock, CA
Lines: 103

In article <380@micropen> dave@micropen (David F. Carlson) writes:
>Are there any beta sites that can confirm or deny these rather reasonable
>claims?  Has anyone *ever* had a demonstrable, verifiable compiler bug
>with the Microport PCC-based compiler?
>David F. Carlson, Micropen, Inc.

Helium is a beta site for V/386.  I found a *something* in the compiler.
I wouldn't call it exactly a bug...  Here's what I sent in:
--- included text ---
There seems to be a bug in the compiler or assembler.  The following source
code:

typedef struct object {
    char posx, posy;
    char velx, vely;
    struct object *next, *prev, *contend;
    long energy;
    long mass;
    char type;
    char image;
    char strategy;
    char flags;
} OBJECT;

void their_smarts()
{
register OBJECT *obj;

setimage(obj, (obj->velx *= -1) < 0 ? '>' : '<');
}

compiles to the following assembler code:

	.file	"yukfoo.c"
	.version	"02.01"
	.data
	.text
	.align	4
	.def	their_smarts;	.val	their_smarts;	.scl	2;	.type	040;	.endef
	.globl	their_smarts
their_smarts:
	pushl	%ebp
	movl	%esp,%ebp
	pushl	%eax
	pushl	%edi
	leal	2(%edi),%eax
	movl	%eax,-4(%ebp)
	movl	%eax,%edx
	imulb	$255,(%edx),%dl
	movb	%dl,(%eax)
	testb	%dl,%dl
	jge	.L16
	movl	$62,%eax
.L17:
	pushl	%eax
	pushl	%edi
	call	setimage
	addl	$8,%esp
/ASMQ
	movl	-8(%ebp),%edi
/ASMEND0
	leave	
	ret	
.L16:
	movl	$60,%eax
	jmp	.L17
	.align	4
	.def	their_smarts;	.val	.;	.scl	-1;	.endef
	.data
	.text

and then the compilation terminates with the error

Assembler: yukfoo.c
	aline 16	: syntax error

The compiler is invoked with "cc -c -O yukfoo.c" on the included source.
The assembly source was produced with "cc -c -O -S yukfoo.c".

--- end included text ---
(Note:  the source code was from Larry Wall's "warp" game...)
Reply received:
--- more included text ---
I've reported the problem to our 386 group and they have 
reproduced the problem. We're checking to see if it's still
a problem in the 2.0 release of V/386 rel 3.

Date:  Wed, 29 Jul 87 10:40:18 PDT
Subject: C bug & Build disk

C bug:

It turns out that the *= operator is not supported.
This is what's been causing you all the grief.
The work-around should be pretty simple.
--- end inclusion ---

The fix worked; however, I wouldn't call non-support of a standard operator
just a bug.  "It aint' C" :-)

          \scott
-- 
Scott Hazen Mueller   City of Turlock    901 S. Walnut Rd.  Turlock CA 95380
(209) 668-5590        lll-crg!csustan!helium!scott   uunet!zorch!helium!scott