Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 18 Aug 2002 14:23:10 +0200
From:      Alexander Leidinger <Alexander@Leidinger.net>
To:        naddy@mips.inka.de (Christian Weisgerber)
Cc:        freebsd-ports@FreeBSD.ORG
Subject:   Re: ports/41710: [port]  lame update (fix CFLAGS)
Message-ID:  <20020818142310.7245a08b.Alexander@Leidinger.net>
In-Reply-To: <ajjihu$jo0$1@kemoauc.mips.inka.de>
References:  <200208161640.g7GGe3Dh028390@freefall.freebsd.org> <20020816191056.0a9a3b2c.Alexander@Leidinger.net> <ajjihu$jo0$1@kemoauc.mips.inka.de>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, 16 Aug 2002 19:07:42 +0000 (UTC) naddy@mips.inka.de (Christian
Weisgerber) wrote:

> > >  *I* want to decide if I build with "-pipe" or without.  And if I
> > 
> > I know only one bugreport with "-pipe", Irix doesn't seem to like
> > it.
> 
> To its credit, the script checks whether the compiler is gcc before
> adding this.

I know, I'm responsible for a large part of configure.in for LAME.

> > Besides this, every OS seems to accept it. Normaly it speeds up the
> > build.
> 
> I know somebody who will tell you not to bother with "-pipe" and
> instead to use a memory-backed /tmp.  Hey, maybe we could benchmark
> the effect of either by building the lame port... Oh, I guess we
> can't.

Not every OS has such a possibility, and LAME tries to work out of the
box on as much systems as possible.

> > So why do we need to patch it (nitpicking on "-pipe" seems to be
> > a bikeshed argument for me)?
> 
> I don't nitpick on "-pipe", you do.

I just want to know a showstopper argument: why do we need to remove
"-pipe" and "-Wall"? Depending on your answer I may want to change it in
the LAME CVS.

> > >  want to, then what's the point of "-pipe -pipe"?  "-Wall" doesn't
> > >  add anything outside development.
> > 
> > It isn't harmfull. We do we need to add a patch for it?
> 
> It obscures the important warnings.  Why does the configure script
> have to go out of its way to *add* it?  Who will fix the warnings

Because configure normaly sets -Wall (and other options) in the macro
AC_PROG_LIBTOOL and I refuse to use theses modified CFLAGS here. I
restore the initial CFLAGS after AC_PROG_LIBTOOL and only use options we
want. Running configure is supposed to configure everything to enable
the use of LAME for users and easy development for developers. We also
have a --enable-debug option for configure, but this is overkill for
normal development, so adding '-Wall' there isn't an option.

> if I dump the output of the build on him?

Generally: nobody.
LAME is Mark Taylors baby and he refuses to allow a lot of the changes
which are needed to get the source -Wall clean (there are a lot of
warnings related to not so strict use of types). There are some
developers which don't share his opinion, and those developers try to
write warns clean code where possible.

Feel free to point out major errors, I will forward them to the people
in charge for the particular subsystems.

> > >  Basically these happened to be nearby, so I removed them along
> > >  with the other cruft that killed the compile on -CURRENT/alpha.
> > 
> > Can you be more specific please?
> 
> The lame configure script used to add various optimization flags
> to CFLAGS, in particular -O3, which produces garish warnings on
> alpha
> 
> cc1: warning: 
>    ***
>    ***  The -O3 flag TRIGGERS KNOWN OPTIMIZER BUGS ON THIS PLATFORM
>    ***

For which versions of gcc is this the case? I want to add a workaround
to configure.

> There are probably thousands of ports in the tree that don't respect
> CFLAGS.  Haggling over each flag isn't going to help clean this
> mess up.

I don't want to do that, I just interested in LAME. This conversation
revealed a bug in the configure script (which I have fixed locally now)
and may also lead to some workarounds for gcc-alpha.

Bye,
Alexander.

-- 
   If Bill Gates had a dime for every time a Windows box crashed...
                ...Oh, wait a minute, he already does.

http://www.Leidinger.net                       Alexander @ Leidinger.net
  GPG fingerprint = C518 BC70 E67F 143F BE91  3365 79E2 9C60 B006 3FE7

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-ports" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020818142310.7245a08b.Alexander>