{stdenv, fetchurl, kernelHeaders, installLocales ? true}: stdenv.mkDerivation { name = "glibc-2.3.3"; builder = ./builder.sh; src = fetchurl { url = http://catamaran.labs.cs.uu.nl/dist/tarballs/glibc-2.3.3.tar.bz2; md5 = "e825807b98042f807799ccc9dd96d31b"; }; linuxthreadsSrc = fetchurl { url = http://catamaran.labs.cs.uu.nl/dist/tarballs/glibc-linuxthreads-2.3.3.tar.bz2; md5 = "8149ea62922e75bd692bc3b92e5e766b"; }; patches = [ # This patch fixes the bug # http://sources.redhat.com/bugzilla/show_bug.cgi?id=312. Note # that this bug was marked as `WORKSFORME' with the comment to # just use glibc from CVS. This and the unholy Linuxthreads/NPTL # mess proves that glibc, together with the Linux kernel, # constitutes an AXIS OF EVIL wrt release management. Patch # obtained from # http://www.pengutronix.de/software/ptxdist/patches-cvs/glibc-2.3.2/generic/fixup.patch. ./fixup.patch # Likewise, this fixes the bug reported in # http://sources.redhat.com/ml/libc-alpha/2003-07/msg00117.html. # Die, glibc, die. ./no-unit-at-a-time.patch # This is a patch to make glibc compile under GCC 3.3. Presumably # later releases of glibc won't need this. # ./glibc-2.3.2-sscanf-1.patch # Fix for localedata loading bug. ./loadlocale.patch # Add a whole bunch of UTF-8 locales. Most of these seem to be in # the CVS mainline now. ./utf8-locales.patch ]; inherit kernelHeaders installLocales; }