The wrapper script works by creating an image of the distribution in the users home directory
(defaulting to $HOME/.config/df_linux). This is done because the program expects to write
state into it's own distribution.
Better ideas are welcome!
What further is needed is to allow users to configure the init. One way is to use nix-config
to apply patches to the init.txt file, though this would break sharing of the package between
different users.
svn path=/nixpkgs/trunk/; revision=24538
The configure script had an hard-coded impure path to the lipo binary. If that
binary was found -- and it was --, then the script decided to pass an invalid
"-arch i386" flag to libtool, which subsequently failed. There is probably
another way to pass that -arch flag, but I didn't look hard because it's
redundant anyway.
svn path=/nixpkgs/trunk/; revision=24530
newest I uploaded yesterday. These have gcc 4.5.1, that fulfills the glibc
2.12.1 requirement of having a compiler equal or greater to gcc 4.4 to build
proper x86 32-bit code (http://sourceware.org/bugzilla/show_bug.cgi?id=12123)
svn path=/nixpkgs/branches/stdenv-updates/; revision=24523
Building with gcc 4.4 as minimum was needed, and not this patch.
http://sourceware.org/bugzilla/show_bug.cgi?id=12123
I uploaded new i686-linux bootstrap-tools with gcc 4.5.1, but I'll not update
the i686 reference to those bootstrap-tools until they are reachable by http.
The bootstrap-tools we used until now had gcc 4.3, said not to be able to build
glibc 2.12.1.
Meanwhile other platforms can go on building this glibc already without the
patch.
svn path=/nixpkgs/branches/stdenv-updates/; revision=24522
Always include the --with-frozenpaths configure flag ( independently of tetex ).
Include the --with-rsvg configure flag when librsvg is not null.
svn path=/nixpkgs/trunk/; revision=24509
- our gcc should not bring libssp, because glibc provides __stack_chk_fail
already. libssp is only for some non-glibc systems.
The gcc configure script was not finding the header files of our glibc, so
it assumed it did not provide __stack_chk_fail. I wrote code in the builder
that patches the gcc/configure script properly for that.
As a consequence, the glibc does not need anymore the "nscd-ssp-linking"
patch, and we have a saner gcc (without libssp when using -fstack-protector)
- Instead of disabling the sse42 strstr() implementation in glibc due to the
bug http://sourceware.org/bugzilla/show_bug.cgi?id=12123, I provide a better
patch written after more reserach (that I submitted upstream already, and
it's pending review).
svn path=/nixpkgs/branches/stdenv-updates/; revision=24493
