On 6/13/06, Richard Fish <[EMAIL PROTECTED]> wrote:
Nope. I checked just for "FLAGS" and got nothing. I also checked for things
like "ldl" and "O1" to see if I could find fragments. No joy.
I found the ebuilds and grepped in there too. I found where it downgrades
optimizer flags to 1, but otherwise no joy.
I used find. It didn't.
I sure agree that it's confusing. My own hope is that it's something like
a broken package source -- especially configure itself, the script that's
reporting the error. If it were causing the error too the fix would be obvious.
How can I force a reload?
++ kevin
On 6/13/06, Kevin O'Gorman <[EMAIL PROTECTED]> wrote:
> On 6/13/06, Richard Fish <[EMAIL PROTECTED]> wrote:
>
> > On 6/13/06, Kevin O'Gorman <[EMAIL PROTECTED]> wrote:
> > > checking for gcc... gcc
> > > checking whether the C compiler (gcc -O2 -march=pentium4
> > > -fomit-frame-pointer -pipe -mfpmath=sse -msse2 -mmmx -fPIC -Wl,-O1
> > > -ldl-Wl,-O1) works... no
> >
> > Looks like CFLAGS and LDFLAGS have gotten combined somehow.
> >
> > What does emerge --info show? Also post the output of "env | grep -e
> > CFLAGS -e CXXFLAGS -e LDFLAGS".
>
>
> That one comes up empty. There are no FLAGs in my environment.
Well this is a bit confusing, because the command that configure uses
for the compiler test is:
${CC-cc} -o conftest${ac_exeext} $CFLAGS $CPPFLAGS $LDFLAGS
conftest.$ac_ext $LIBS
The ebuild only adds the -fPIC option, and the "-Wl,-O1 -ldl-Wl,-O1"
stuff doesn't show up anywhere in the glib sources. So the only place
that could come from is the environment...
We didn't check CPPFLAGS, do you have that set by chance?
Nope. I checked just for "FLAGS" and got nothing. I also checked for things
like "ldl" and "O1" to see if I could find fragments. No joy.
optimizer flags to 1, but otherwise no joy.
What does "locate config.site" report? (or "find / -name config.site"
if you don't use slocate). It should come back with nothing...
I used find. It didn't.
Just for comparison, when I build glib-1.2, I get:
checking whether the C compiler (gcc -march=pentium4 -Os
-fomit-frame-pointer -pipe -fweb -fPIC ) works... yes
and my CFLAGS are set to
CFLAGS="-march=pentium4 -Os -fomit-frame-pointer -pipe -fweb"
So you can see the ebuild adds -fPIC option, and nothing else should
be sneaking in...
-Richard
I sure agree that it's confusing. My own hope is that it's something like
a broken package source -- especially configure itself, the script that's
reporting the error. If it were causing the error too the fix would be obvious.
How can I force a reload?
++ kevin
--
Kevin O'Gorman, PhD
