[vorbis-dev] Re: [vorbis] bug in glibc 2.1.2 and older

[R.L. Horn]

On Mon, 26 Jun 2000, Monty wrote:

> Insufficient!  The bug is *not* in the compiler.  It is in the glibc headers.
> I can make bleededing edge GCC break very easily with the bad glibc.

It seems to me that you could get the glibc version with something like:

LIBC=`/sbin/ldconfig -p | grep "libc\.so\.6" | sed "s/.*=> //"`
LIBCVER=`ls -l $LIBC | sed "s/.*-> libc-//" | sed "s/\.so//"`

A subexpression substitution for the latter would be more elegant, but
either they're broken in my copy of sed or I don't know how to work them.  
The perl equivalent would obviously be:

ls -l $LIBC | perl -pe "s/.*-> libc-(.+)\.so/\1/"

In practice, you might end up with something like:

case $host in
*-*-linux*)
  OPT="-O20 -ffast-math -D_REENTRANT -fsigned-char"

  echo -n "checking glibc version... "
  LIBC=`/sbin/ldconfig -p | grep "libc\.so\.6" | sed "s/.*=> //"`
  LIBCVER=`ls -l $LIBC | sed "s/.*-> libc-//" | sed "s/\.so//"`
  echo $LIBCVER
  case $LIBCVER in
    2.0.* | 2.1.0 | 2.1.1 | 2.1.2)
      OPT="$OPT -D__NO_MATH_INLINES"
      echo "using -D__NO_MATH_INLINES"
  esac
  unset LIBC LIBCVER

  DEBUG=...

in configure.in.  Believe it or not, this does work (within extremely
narrow parameters, of course).

Yes, I know it's an ugly sort of hack, but with only __GLIBC__ and
__GLIBC_MINOR__ defined in the headers I'm not sure what else to do.

If any autoconf experts out there have a better idea I'd sure like to hear
it (heck, maybe there's a macro for doing this that everyone else already
knows about).

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/