From: pavenis AT lanet DOT lv
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>, djgpp-workers AT delorie DOT com
Date: Wed, 14 Jun 2000 15:52:27 +0200
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: Warning patch for exceptn.S
Message-ID: <3947AA3B.21327.5284DC@localhost>
References: <39478E68 DOT 31020 DOT B60EA5 AT localhost>
In-reply-to: <Pine.SUN.3.91.1000614162344.841B-100000@is>
X-mailer: Pegasus Mail for Win32 (v3.12c)
Reply-To: djgpp-workers AT delorie DOT com

On 14 Jun 2000, at 16:26, Eli Zaretskii wrote:

> 
> On Wed, 14 Jun 2000 pavenis AT lanet DOT lv wrote:
> 
> > It's worse. AS from binutils-2.9.1 refuses to accept exceptn.S after 
> > patch. The same is with
> > 	.byte 0x2E
> > 	ljmp *foo
> > 
> > So we have an example where binutils-2.[89].* and 2.10 prereleases
> > are incompatible.
> 
> Darn!
> 
> > Only (only partial) cure I see is not to use  -Werror
> > when compiling .S files
> 
> Is there perhaps some preprocessor magic that should help us out?  I'm 
> thinking about keeping the old code for older Binutils, the new one for 
> v2.10 and later, and having some #ifdef or assembly-level directive to 
> choose between them.  Is this possible?

So we should detect version of GNU assembler somewhere. After that 
we could add some definition to make the actual version of AS 
available to CPP.

Perhaps it could be rather easily done with autoconf, but I don't think 
it's nice idea to use autoconf. 

Maybe we could also remove -Werror when compiling .S files. So we'll 
get warnings with current sources but at least no bigger problems.

Andris

PS. I put my build of DJGPP binary archive of binutils-2.10 release 
candidate 3 at   http://www.lanet.lv/~pavenis/djgpp/bnu210rc3.zip
if You are interested to do some testing and don't want to build it 
Yourself. As usually it's built under Linux (by cross-compiling)