From: Donn Miller <dmm125 AT bellatlantic DOT net>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Parse errors
Date: Fri, 23 Oct 1998 02:08:41 +0000
Organization: Bell Atlantic Internet Solutions
Lines: 47
Message-ID: <Pine.BSF.4.05.9810230201470.22326-100000@localhost>
References: <4af57b8d DOT 36227988 AT aol DOT com> <Pine DOT OSF DOT 4 DOT 05 DOT 9810122353150 DOT 5384-100000 AT sable DOT ox DOT ac DOT uk> <703q7p$bus$1 AT nnrp1 DOT dejanews DOT com>
NNTP-Posting-Host: client201-122-35.bellatlantic.net
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Sender: dmm125 AT localhost
In-Reply-To: <703q7p$bus$1@nnrp1.dejanews.com>
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

On Thu, 15 Oct 1998 silkwodj AT my-dejanews DOT com wrote:

> George, this is the kind of frustration I suggested I may have to post as
> well.
> 
> 
> I've been progressing pretty well, writing Nasm code and linking with DJGPP. 
> I worked on a modestly sizeable Borland conversion project.  After marching
> through a huge page of compiler error messages(newbie dumb stuff), and
> resolving one at a time I finally get down to a compiler execution line and
> output that looks like this:
> 
> C:\Work\pcx >gcc pcxview3.cc pcx.o myscrn1.o myscrn2.cc -o pcxview.exe
> pcxview3.cc:71: parse error at end of input.
> 
> Now my pcxview3.cc file has only 70 lines in it!  What's also baffling is the
> file pcx.o was compiled a couple of weeks ago, but now I can't compile it
> again. I get the same kind of error message for the end of its 276 lines of
> code.
> 
> I have assumed I'm missing a squiggly bracket or a semicolon or something,
> but I'm going up a wall!  I've brought the files into several editors and
> made sure there is nothing like a non-printing character at the end.  I've
> flipped the order of functions in the source to make sure it is not specific
> to one function at the end.  The parse error always appears at one line past
> the end of my code.

It could mean that you've got a piece of hardware like memory or cache
gone bad.  But with bad memory, you get parse errors in random places each
time.  Like if the compiler tells you you forgot a semicolon somewhere,
and there is one, you can be sure it's bad memory.

The reason bad memory causes parse errors:  first, your source code is
read into memory.  Then, when reading one char out from a bad memory
location, it will look like a random or null char instead of what it
really is.

I would check the memory by booting up under command line only mode, and
see what HIMEM.SYS tells you.  But don't go by the memory check at the
beginning (the BIOS memory check).  My BIOS rolled up the memory 4 times
each time I reset my machine and said all was OK.  However, when compiling
I got parse errors for no reason and HIMEM.SYS told me I had bad memory at
this one location.  I replaced the 32M ram with a fresh, all new 64M ram
and no more parse errors or HIMEM boot-time errors.

Donn