For now, we don't NATIVE_SYSTEM_HEADER_DIR because it breaks the
build. However, it points to Glibc in the Nix store (not /usr/include)
so it's kind of okay.
Even if using profiledbootstrap. This was unsafe before 4.8, and
then the documentation was not fixed on time.
The documentation got fixed here:
c763997f34
But the actual code was already fixed here:
5d2fca09d5
So this is safe both for GCC 4.8 and GCC 4.9.
(Or at least restore it to the same level as brokenness as before the
staging merge. I.e. it depends on gcc-apple, which doesn't build on
standard 10.9 systems.)
Having a separate lib64 is bad because it requires special-casing in
lots of places.
Previously done in e4a11b4d92, which
apparently got lost going from 4.6 to 4.8.
Recent versions of Xcode don't install headers in /usr/include but
in a directory like
/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.9.sdk/usr/include
So use that instead, falling back to /usr/include in case of an older
version of Xcode.
Somewhere the no-sys-dirs.patch got disabled, so gcc was looking in
/usr/local/include and /usr/lib. Since I can't fix the patch easily,
I've borrowed the --sysroot trick from clang-wrapper. This causes
builtin paths to be prefixed with /var/empty
(e.g. /var/empty/usr/lib), which don't exist.
This includes a lot of fixes for cross-building to Windows and Mac OS X
and could possibly fix things even for non-cross-builds, like for
example OpenSSL on Windows.
The main reason for merging this in 14.04 already is that we already
have runInWindowsVM in master and it doesn't work until we actually
cross-build Cygwin's setup binary as the upstream version is a fast
moving target which gets _overwritten_ on every new release.
Conflicts:
pkgs/top-level/all-packages.nix
On MinGW, we're passing these programs to the configure script, but this
obviously won't work for non-autoconf-based projects.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Both branches have quite a lot in common, so it's time for a merge and
do the cleanups with respect to both implementations and also generalize
both implementations as much as possible.
This also closes#1876.
Conflicts:
pkgs/development/interpreters/lua-5/5.2.nix
pkgs/development/libraries/SDL/default.nix
pkgs/development/libraries/glew/default.nix
pkgs/top-level/all-packages.nix
Cross-compiling stuff against Mac OS X's CoreFoundation won't work
without ObjC support, and we don't want to compile commandline utilities
only, right?
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Let's finally hook everything into the existing cross-building
infrastructure. We're using --with-sysroot instead of --with-headers
here, because the XCode SDK contains references to /usr/lib.
I've tried to patch those references, but unfortunately (at least with
install_name_tool) it isn't possible to change those refernces in stub
dylibs.
So after bugging @tpoechtrager with annoying questions (thanks again), I
think my initial approach (patching the SDK itself and/or regenerating
the dylib stubs) was way to complicated so I ended up with this
implementation.
Also, I've added a condition to binutilsCross to use cctools if the libc
is set to libSystem. This might need some cleanups someday, mainly to
figure out how to properly bridge cctools and binutils.
So, as an example on how to cross-compile GNU Hello to Darwin, you can
use something like this:
(import <nixpkgs> {
crossSystem = {
config = "x86_64-apple-darwin13";
arch = "x86_64";
libc = "libSystem";
platform = {};
};
}).hello.crossDrv
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Mingw(32) is rather poorly maintaned and has quite a lot of bugs. And
because our Windows cross builds were also poorly maintained and most of
the cross-tests were broken as well, I'm just taking this step and try
to switch to mingw-w64 for everything "cross Windows".
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Specifically, we are trying to fix the following error seen on Hydra:
../../gcc-4.7.3/gcc/gengtype-lex.c:1:21: fatal error: bconfig.h: No such file or directory
The patch is taken from gcc's SVN revision 193691.
It's to separate from other changes coming from master.
Conflicts:
pkgs/development/libraries/glibc/2.18/common.nix (taking both changes)
pkgs/development/libraries/ncurses/5_4.nix (deleted)
Sometimes the build failes with:
In file included from ../../gcc-4.4.6/gcc/ada/seh_init.c:44:
../../gcc-4.4.6/gcc/system.h:418: error: conflicting types for 'strsignal'
/nix/store/6h129q168ahnl2nzw6azr239cba884ng-glibc-2.18/include/string.h:560: note: previous declaration of 'strsignal' was here
and sometimes it doesn't. Hopefully disabling parallel builds fixes
this.
http://hydra.nixos.org/build/7179481
Without it, gcc builds for softfloat, and the glibc doesn't have support for
softfloat (it ends up requiring some gnu-soft.h file). We'll have to test if
this fixes the build of gcc or not, though.
Conflicts:
pkgs/development/compilers/gcc/4.6/default.nix
pkgs/development/compilers/gcc/4.7/default.nix
The 4.7 had some weird parameters added in crossAttrs; I've removed
them, but I don't understand where they come from.
This is for consistency with terminology in stdenv (and the terms
"hostDrv" and "buildDrv" are not very intuitive, even if they're
consistent with GNU terminology).
