On 10/23/2014 06:58 AM, Daniel P. Berrange wrote:
Create a new libvirt-domain.h file to hold the public
API definitions for the virDomain type. This header
file is not self-contained, so applications will not directly
include it. They will contain to #include <libvirt/libvirt.h>
---
cfg.mk | 2 +-
docs/apibuild.py | 1 +
include/libvirt/Makefile.am | 1 +
include/libvirt/libvirt-domain.h | 3496 +++++++++++++++++++++++++++++++++++
include/libvirt/libvirt.h.in | 3731 ++------------------------------------
5 files changed, 3632 insertions(+), 3599 deletions(-)
create mode 100644 include/libvirt/libvirt-domain.h
You ended up changing a lot of '#ifdef' to '# ifdef' to keep cppi happy,
which in turn made my filter comparison a little more verbose. Worse,
your diff claims that stuff related to VIR_NODEINFO_MAXCPUS appeared to
move around (so you weren't using the patience algorithm). But overall,
once I recomputed your diff with --patience and added a step of 's/#
/#/' to my filter, it was once again an easy review that it is
straightforward code motion.
ACK.
Here's the same diffstat when generating it under --patience (notice how
libvirt.h.in has a lot less churn once you turn on the better algorithm):
cfg.mk | 2 +-
docs/apibuild.py | 1 +
include/libvirt/Makefile.am | 1 +
include/libvirt/libvirt-domain.h | 3496
++++++++++++++++++++++++++++++++++++++
include/libvirt/libvirt.h.in | 3467
+------------------------------------
5 files changed, 3500 insertions(+), 3467 deletions(-)
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library
http://libvirt.org