Date: Mon, 11 Apr 2005 21:44:39 +0200 From: Roman Neuhauser <neuhauser@sigpipe.cz> To: Scott Long <scottl@samsco.org> Cc: doc@freebsd.org Subject: Re: mozilla's install hanging on amd64 Message-ID: <20050411194439.GA98521@isis.sigpipe.cz> In-Reply-To: <425AA728.7050809@samsco.org> References: <200504102137.j3ALbm0h079084@corbulon.video-collage.com> <200504111212.25215.mi%2Bmx@aldan.algebra.com> <425AA274.2090604@samsco.org> <200504111229.39072.mi%2Bmx@aldan.algebra.com> <425AA728.7050809@samsco.org>
next in thread | previous in thread | raw e-mail | index | archive | help
# scottl@samsco.org / 2005-04-11 10:34:48 -0600:
> Mikhail Teterin wrote:
> >My whole point, Scott, is that there is nothing special about "my gcc
> >flags". NOTHING. Setting CPUTYPE to match one's processor flavor is a
> >long-documented way to do things -- fact. Such setting results in the
> >corresponding -march=XXX automatically -- fact. All ports must handle
> >this. Those that can not are broken. Plan and simple. There is nothing to
> >argue: `-O -pipe -march=opteron' must work.
(...)
> Is my car broken because it can't go 300km/hr? A Porshe can, so
> shouldn't mine be able to? My point here is that there is a distinction
> between the mozilla port being completely broken, and being broken in
> non-standard or specific configurations. Saying that it's BROKEN
> implies that it's 100% unusable, and that gives a false summary of the
> problem. I built it from scratch with a very stock, unmodified system
> and it works as well for me as I would expect. In my eyes, it's not
> broken. But I don't play with custom gcc options.
>
> I'll readily concede that there are a lot of ports that are authored
> in an i386-specific way and only work on amd64 by sheer luck.
> Understnad that FreeBSD tends to also be at the leading edge of compiler
> development. While we are using GCC 3.4, many other BSD and Linux
> variants are still using 3.1, 3.2, or even 2.95. So it's quite possible
> that the newer compiler has either bugs or stricter languange
> compliance, and there aren't enough eyes yet to discover and fix these
> problems. Shouting to everyone that mozilla is BROKEN doesn't help this
> very much.
I think this thread shows that there is a need for a written policy
on -march / -mcpu / -mtune gcc options in ports. Heated discussions
like this one could be avoided if the Porters Handbook included an
article or chapter on this.
Whether that policy was "-march= must work" or "-march= is
unsupported" or "ports should filter out -march= if they break with
it, patches for -march-related bugs welcome" (most reasonable
IMNSHO) would be a matter of consensus, but would be a boon to have.
MFT set to ports@ and doc@.
--
How many Vietnam vets does it take to screw in a light bulb?
You don't know, man. You don't KNOW.
Cause you weren't THERE. http://bash.org/?255991
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050411194439.GA98521>