On Thu, Oct 20, 2016 at 14:22:54 -0500, Eric Blake wrote:
On 10/20/2016 10:24 AM, Peter Krempa wrote:
> From: Andrea Bolognani <abologna(a)redhat.com>
>
> Sync up to the released version of qemu.
>
> Signed-off-by: Peter Krempa <pkrempa(a)redhat.com>
> ---
> .../qemucapabilitiesdata/caps_2.6.0.ppc64le.replies | 21 +++++++++++----------
> tests/qemucapabilitiesdata/caps_2.6.0.ppc64le.xml | 4 +++-
> 2 files changed, 14 insertions(+), 11 deletions(-)
>
I find these files very fragile to edit. It would be nice if we could
Thats why we don't edit them any more. tests/qemucapsprobe generates
them from a qemu binary.
Not keeping them in order will promote manual edits which will result in
horrible diffs in case when anyone would actually try to generate them
again.
rework them to be independent of the libvirt 'id':nnn line
(so that
added or removed commands from libvirt's end don't resequence every
reply in the file) as well as correlate it with strategic comments to
Pavel wrote a script that renumbers those. It's not commited yet but
quite useful if you are going to do a manual edit.
the command that libvirt sent (to make it more obvious what the reply
is
to). But that's not a task for this series.
I thought about doing this, but by Jirka's addition of qemucapsprobe it
doesn't make much sense anu more.
Fairly mechanical, so I'm going to trust that this and similar
patches
are correct, based on the fact that the testsuite still passes 'make
check', without actually digging out a particular qemu version to do a
100% cross-check.
The only variable here is how the given qemu version was compiled at the
time when qemucapsprobe was invoked. Other than that it's not even
mechanical, but programatic.