This prevents failures like "libgcc_s.so.1 must be installed for
pthread_cancel to work" that occur because Glibc assumes libgcc_s.so.1
to be in Glibc's libdir.
This solution is pretty hacky, because the libgcc_s.so.1 from
bootstrap-tools might be too old. So if we update GCC, programs might
end up using an outdated libgcc_s.so.1. Ideally, we would build
libgcc_s.so.1 *before* Glibc, which might not be impossible...
Fixes#3548.
The old driver wasn't updated since August 2013, and e.g. won't compile
with glibc-2.20. Distros like Arch or Ubuntu are using newer one for a long time.
Without this, ImageMagick's configure script will generate code
specific to the machine building the package. This code may then fail
on other CPU types.
http://hydra.nixos.org/build/16564129
The static curl program is gone, replaced by curl inside of the
bootstrap tools tarball. Also, we generate a .tar.xz archive rather
than .cpio.bz2, making the download smaller. The separate
{sh,cpio,mkdir,ln,bzip2} programs have been replaced by a single
busybox program.
This starts with "make allnoconfig" rather than "make defconfig",
making it easier to turn on only the needed features.
Also, fix broken .config generation (the presence of "#" lines
confused parseconfig).
Commit a28940d9d5 changed the SQLite
build to use CFLAGS instead of NIX_CFLAGS_COMPILE, but that's really
bad because it clobbers the default -O2 flag. So all this time we had
an unoptimized SQLite build. (This is one of the reasons why
NIX_CFLAGS_COMPILE exists - messing with CFLAGS is almost never a good
idea.)
Apparently it is possible for the config to be compiled correctly but
for ‘hint’ still not know where the packages are: it seems to ignore the
GHC in PATH and does whatever.
- test/test_support.py module is used by many libraries which backport python
3.0 functionality to 2.6/2.7
- enable subprocess32 tests (now passing for python 2.6/2.7)
