Xref: news2.mv.net comp.os.msdos.djgpp:5189
From: brucef AT central DOT co DOT nz (Bruce Foley)
Newsgroups: comp.os.msdos.djgpp
Subject: Question from a new user
Date: Thu, 20 Jun 1996 07:15:26 GMT
Organization: Internet Company of New Zealand
Lines: 33
Message-ID: <4qb17n$2gf@status.gen.nz>
NNTP-Posting-Host: brucef.central.co.nz
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

As a new user of this cool new compiler, I have struck a
couple of issues that I would appreciate comments
on from the more experienced users amoung us.

I have just started converting some core routines from
my old library over to DJGPP.  The first thing I wanted to
do was convert a routine that does vertical retrace
syncronization, so I could get something happening on
the screen!!!!

I recoded it using at&t style syntax, but kept the logic
flow identical.  When I did a test run, it ran slow!!!
I checked my code against the original, and everything
looked 100%.  The problem turned out to be with the
loopnz instruction.  This now decrements ecx (of course)
not cx, like I'm used to.  by doing an xor on it before hand,
it worked perfectly.  However, I tried (perhaps somewhat
hopefully) to code it as loopnzw (why not) and it even
compiled.  However, it still decremented ecx.
Can we make loop instructions work on cx?

The next thing is that I mistakenly typed:
gcc -o module.c module.s
The bloody thing deleted my source code!!!!!

I know the above was a silly mistake (brought on by my
egerness to do a test run)  but isn't that a harsh punishment?
I can't help but feel that the compiler should treat anything
that might potentially be source code with absolute sanctity.

Regards, Bruce.