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, 13 Jun 2001 16:01:10 +0300
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: GCC-3.0 related problem with src/libc/stubs/stubXXXX.S
Message-ID: <3B278E46.6870.A077C9@localhost>
References: <3B27870A DOT 168 DOT 843261 AT localhost>
In-reply-to: <Pine.SUN.3.91.1010613153321.17867E-100000@is>
X-mailer: Pegasus Mail for Win32 (v3.12c)
Reply-To: djgpp-workers AT delorie DOT com

On 13 Jun 2001, at 15:39, Eli Zaretskii wrote:

> 
> On Wed, 13 Jun 2001 pavenis AT lanet DOT lv wrote:
> 
> > > Does that mean that -remap and -imacros don't work with "gcc -traditional"?
> > > 
> > > If that is indeed so, I guess we have no other choice but to modify the 
> > > stubs as you suggested originally.  Tweaking specs for this is too gross, 
> > > IMHO, and might also cause compatibility problems in the long run.
> > 
> > I don't expect special problems from such modification of specs (I 
> > simply copied condition from rule how to choose which of cpp0
> > or tradcpp0 to use). 
> > 
> > gcc -traditional will fail unconditionally unless I exclude both
> > -remap and -imacros when tradcpp0 is being used. We'll unable to 
> > compile too much things with gcc -traditional even when we'll
> > remove these options with tradcpp0.
> 
> If the problem with -traditional is not limited to building the library, 
> then changing specs _is_ the solution, because we will have to make that 
> change anyway, to let users use -traditional.

-traditional was unusable with gcc-2.95.3 (simple hello world style 
program which includes stdio.h and outputs a message with printf
doesn't compile). gcc-3.0 doesn't change anything here. System 
headers (like stdio.h) are incompatible with -traditional also under 
Linux
> 
> However, I wonder: since what version of GCC did -traditional stop 
> working with -imacros?  Is this something new in GCC 3.0?

Verified gcc-2.95.3 doesn't feed cpp_spec to tradcpp0. gcc-3.0 does.
So for gcc-2.95.3 -imacros and -remap were excluded for tradcpp0 

> Also, this seems to suggest that "gcc -traditional" will not pick up the 
> __DJGPP__ and __DJGPP_MINOR__ symbols, and also will not be able to 
> automatically find some of the C++ headers which need -remap, and that 
> this side effect cannot be cured by changes in specs.  Is that true?

True.

Andris