From:	pavenis AT lanet DOT lv
To:	CBFalconer <cbfalconer AT yahoo DOT com>, djgpp AT delorie DOT com
Date:	Tue, 4 Jun 2002 14:25:48 +0300
MIME-Version:	1.0
Subject:	Re: GCC 3.1 and DJGPP - observations.
Message-ID:	<3CFCCDEC.29537.AC5AA5@localhost>
In-reply-to:	<3CFC9165.47E43C09@yahoo.com>
X-mailer:	Pegasus Mail for Windows (v4.01)
Reply-To:	djgpp AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	djgpp AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

On 4 Jun 2002 at 10:22, CBFalconer wrote:

> Having gotten this installed, and having removed GCC 2.953, I have
> recompiled a few things unchanged, and am observing the following:
> 
> 1) It is showing up failure to #include <string.h> here and there
> when using strlen and strcmp.  Good.

gcc-2.95.X permitted to use strlen(), strcmp() and some other functions
without declaration. gcc-3.0.X and gcc-3.1 no more permits that.
 
> 2) The executables are a little bit smaller, compiled with
> gstabs+.  However after passing through strip the executables are
> somewhat larger.  Don't know why.

My experience shows that both -gcoff (read -g) and -gstabs+ are
often near to unusable with  many C++ programs which uses 
STL. -gdwarf-2 seems to behave better (at least with GDB-5.2).

So maybe it's time to switch to -gdwarf-2 as the default soon?

The size of executable with debug info seriously increases when 
-gdwarf-2 is being used.

> 3) All versions were compiled with O2.  Looking at the generated
> code with "objdump -dS" it appears as if the optimization is quite
> different.  Many lumps seem to be relegated to "final" code
> outside the purvey of the source.  This may be related to 2)

About warnings generated by gcc-3.1: I often use '-W -Wall -Werror'
in my code and usually this forced to do some changes to use a new versions,
but not too much.

Andris

- Raw text -