nixos/nixpkgs.git, branch staging.patchShebangs

While patching shebangs in source, keep the modification date

2015-01-08T09:25:04+00:00

Otherwise this triggers automake/autoconf/m4 dependencies that
are redundant because of the trivial change.

Introduce patchShebangsPhase ran in preConfigurePhases

2015-01-02T01:58:32+00:00

(cherry picked from commit 91c7e8747af1fdc2a70cd98594ccbb12a5c6902d)
Signed-off-by: Domen Kožar

determinism: change some fixed timestamp to != (time_t)0

2014-12-30T16:03:39+00:00

vcunat removed the unrelated glib change.
Conflicts:
	pkgs/development/libraries/glib/default.nix
	pkgs/os-specific/linux/kernel/generic.nix
	pkgs/os-specific/linux/kernel/manual-config.nix

makeInitrd: use deterministic compression

2014-12-30T16:03:15+00:00

perl-modules: do not create perllocal.pod, for determinism

2014-12-30T16:03:15+00:00

Re-Revert merge #5505: util-linux: use /run/current-system/sw/bin/*

2014-12-30T09:53:41+00:00

This reverts commit 6222093eecf1f314f82deafa3162c2ad923e66d3.

Revert "merge #5505: use /run/current-system/sw/bin/*"

2014-12-30T09:51:36+00:00

I didn't realize it causes a nontrivial rebuild (through systemd).
I'll re-push to staging.

This reverts commit 8cf4ef7ab747b3e3d0b1319de538257281355b33.

merge #5505: use /run/current-system/sw/bin/*

2014-12-30T09:31:20+00:00

A working /sbin/shutdown (usually provided by systemd) is required by
"rtcwake -m off". Creating a circular dependency on systemd for this
just isn't worth the trouble, so take the straightforward way out.

One could easily make the argument that rtcwake wants to shut down the
*currently running system*, and that the correct API for that *is* in
fact /run/current-system. And it makes a very tempting sort of sense.

@vcunat moved to postPatch and added #FIXME.

cyrus_sasl: Enable kerberos

2014-12-30T09:04:16+00:00

krb5: 1.12.2 -> 1.13

2014-12-30T09:03:29+00:00