Nmap itself is excluded, just want the basic network tools.
ncat is a netcat implementation with support for fun things like SSL.
In the future we could add nping but since we already include iputils
that is not quite as important as ncat.
The nmap ebuild includes a compile fix posted here:
https://bugs.gentoo.org/show_bug.cgi?id=501136
Symlinks out of /etc to /usr should generally be relative, that way they
work regardless of whether you are based at / or dealing with a new
chroot you haven't entered yet, or a build root like /build/amd64-usr
Namely the absolute links break cros_sdk which copies skel into home.
Also switch to /usr/share/skel since other packages may need install
things to that directory in the future.
If c10n fails etcd should not start, wait until it is known if there is
a cluster config to use or not. Also now c10n may not write out a
cluster config on ec2 if there isn't a need for one. Make this case
non-fatal and start up etcd as a master node.
Includes a few other changes which are either bug fixes or doc updates:
Brandon Philips (4):
bump(README): increase the version of the README to 0.3.0
fix(README): move contact closer to contributing
feat(scripts): use zip for windows and darwin
fix(server/v1): don't fail put on new v1 key
Brian Waldon (3):
doc(CompareAndDelete): Add missing CAD docs
doc(CompareAndSwap): clarify prevIndex in CAS
chore(gofmt): gofmt compare_and_delete_command.go
Jan-Erik Rediger (1):
Change token example to use returned value only
Michael Marineau (1):
add(server/v1/tests): Port many of the v2 HTTP handler tests to v1
TANABE Ken-ichi (2):
fix(mod/lock): Use dedicated channel to shutdown goroutine properly
fix(mod/lock): Use CreatedIndex in the first node to watch
A broken e2fsprogs-libs binary package lacking compile_et has been
causing problems. I am entirely at a loss as to how this happened in the
first place and unfortunately the error is not revealed until much
later. Hopefully this crude test at the end of src_install will find the
error as it happens. Or at the very least the revision bump will force
everything to move past the one bad build.
One oddity here: /etc/nsswitch.conf is setup as a postinst command in
order to avoid conflicting with baselayout once it starts installing
nsswitch.conf instead. Later glibc won't provide nsswitch.conf at all.
We haven't been using it, the version we have is old, and the build
appears to be a little flaky. Just drop it, can always bump and re-add
later if someone wants it.
