From fnaumann@mail.cs.uni-magdeburg.de Fri Nov 21 15:32:47 2003 Date: Fri, 21 Nov 2003 15:29:21 +0100 (CET) From: Frank Naumann X-X-Sender: fnaumann@wh58-508.st.uni-magdeburg.de To: mint@fishpool.com Subject: Re: [MiNT] 030-version of gcc In-Reply-To: <3FBE19BD.7040506@lutece.net> Message-ID: References: <3FBD2E32.90809@rhrk.uni-kl.de> <3FBD3425.8070405@lutece.net> <3FBE19BD.7040506@lutece.net> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Spam-Score: -4.9 (----) X-Spam-Report: ---- Start SpamAssassin results -4.9 points, 5.0 required; -4.9 BAYES_00 BODY: Bayesian spam probability is 0 to 1% [score: 0.0000] ---- End of SpamAssassin results X-Scan-Signature: ada762d350314a5239641ada00cc5f9d Delivered-To: mint@fishpool.com Delivered-To: mint@lists.fishpool.fi X-ecartis-version: Ecartis v1.0.0 Sender: mint-bounce@lists.fishpool.fi Errors-to: mint-bounce@lists.fishpool.fi X-original-sender: fnaumann@cs.uni-magdeburg.de Precedence: bulk List-help: List-unsubscribe: List-ID: X-List-ID: Hello! > > You oversee some important points. The 3.x series of the gcc is much > > slower and require much more memory than the 2.95.3. > > You may be right, but only for the compilation. The people who use the gcc already complain about the memory and the CPU usage. Not everybody have a 060 machine with tons of RAM. > The produced executables > are quicker and smaller. This is the significant progress, don't you > think so ? The performance gain of the compiled executables is smaller as you may think (this apply especially for C code; C++ profit more). The m68k backend don't changed much from 2.95.3 -> 3.x; on other architectures it look different for sure. > > Also the C++ > > ABI changed incompatible, you must rebuild and replace all C++ > > libraries at once. > > Is there many C++ programs specifically written for ATARI machines ? > I only know one, that is "Nethack". The other are adapted from other > platforms and should be rewritten. > > I don't practice C++, but for C there's no problem of portability > between 2.9x to 3.x. "math.h" should only be adapted a bit for > the inline code. Anyway, if C++ (not C) programs have problems, > it should be corrected, because it is a problem of non portability. > They are not cleanly written. Don't you think so ? You don't read my statement correctly (or I wrote it wrong). You can't link C++ objects from 2.95.3 with 3.x objects and vice versa. That means that all sparemint libraries that include C++ objects must be recompiled. If you provide the 2.95.x as fallback you also need all C++ including libraries as 2.95.x version. > I personally have a Hades 060, I'm an ATARI developer with no > such problems. All ATARI machines are not suitable for developments. > People who don't want GCC 3.x can still use 2.9x. But you can't > limitate the majority of developers because of a minority. For > the users, they should be pleased, because GCC 3.x is better. > 3.x produces better code, and can be better optimized. That's > a strong advantage ! The 3.x versions of the gcc are no such better as you describe here (for the m68k backend), sorry that's not the reality. Not to speak about the newly introduced bugs in gcc 3.x (they will be eliminated over the time). Ciao ...Frank -- ATARI FALCON 040 // MILAN 060 ----------------------------------------- http://www.cs.uni-magdeburg.de/~fnaumann/ e-Mail: fnaumann@freemint.de