Date: Tue, 13 Jan 2009 23:56:40 +0000 From: Pegasus Mc Cleaft <ken@mthelicon.com> To: freebsd-current@freebsd.org Cc: Eitan Adler <eitanadlerlist@gmail.com>, Brooks Davis <brooks@freebsd.org>, Michel Talon <talon@lpthe.jussieu.fr>, Stephen Montgomery-Smith <stephen@math.missouri.edu> Subject: Re: Alternatives to gcc (was Re: gcc 4.3: when will it become standard compiler?) Message-ID: <200901132356.40820.ken@mthelicon.com> In-Reply-To: <496D1ED6.4090202@FreeBSD.org> References: <20090113044111.134EC1CC0B@ptavv.es.net> <20090113222023.GA51810@lor.one-eyed-alien.net> <496D1ED6.4090202@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tuesday 13 January 2009 23:08:06 Maxim Sobolev wrote: > Brooks Davis wrote: > > On Tue, Jan 13, 2009 at 05:04:21PM -0500, Eitan Adler wrote: > >>> Smells like FUD to me. In all of my reading, I have never seen such a > >>> claim. There may be some GPLv3 issues, but I seriously doubt this is > >>> one. > >> > >> Which leads to my next question: why not upgrade? > > > > Given the number of FreeBSD using companies who are completely banned the > > presence of GPLv3 source from their sites, improvements would have to > > be extremely compelling and there would have to be a straight forward > > way to produce snapshots of the src tree with out any GPLv3 components > > as well as a simple way to build said source tree with a non-GPLv3 > > compiler. > > Crazy idea perhaps, but can we make gcc 4.3 (as well as other GPLv3 > components) an opt-in, just like we used to have crypto parts in the > good old days when US was trying to limit export of this technology? > Then can make both camps happy. Yes, it probably means that more efforts > would be required to maintain it and keep code compatible with both > versions, but since our current GPLv2 compiler is pretty much frozen it > should not be much of the hassle as long as the initial work to support > both versions have been done. > > -Maxim At the moment you can already compile gcc 4.3 from the ports tree, however things like binutils only seems to exist in the ports as a cross compiling tool. How hard would it be to add binutils as a port and make the gcc 4.x ports dependent on it? This way you can install gcc 4.3 with the assembler and linker that play nice together during the build? At the moment, I have had to make binutils from a gnu downloaded source and then make gcc 4.3 with a silly make, IE: make AS=/usr/local/bin/as .......... Unless the makers of gcc 4.3 changed this, I had all sorts of compiling problems when a port was using g++ due to headers that were not included by default (that g++ 4.2 did). I would imagine there would be a lot of rework to do in the ports tree making it happy. Once you get a happy and stable 4.3 (and later) installed on your machine, its just a simple change in the make.conf to choose what compiler you want to use. Would this approach get around the need to have 4.3 installed as a BSD default? Peg
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200901132356.40820.ken>