On 06/04/2012 04:29 AM, Michal Privoznik wrote:
On 04.06.2012 05:37, Wen Congyang wrote:
> ---
> tools/virsh.c | 170 ++++++++++++++++++++++++++++----------------------------
> 1 files changed, 85 insertions(+), 85 deletions(-)
>
> diff --git a/tools/virsh.c b/tools/virsh.c
> index a934c13..0e41d00 100644
> --- a/tools/virsh.c
> +++ b/tools/virsh.c
> @@ -3502,16 +3502,16 @@ cmdSaveImageEdit(vshControl *ctl, const vshCmd *cmd)
> if (vshCommandOptString(cmd, "file", &file) <= 0)
> return false;
>
> - #define EDIT_GET_XML \
> - virDomainSaveImageGetXMLDesc(ctl->conn, file, getxml_flags)
> +#define EDIT_GET_XML \
> + virDomainSaveImageGetXMLDesc(ctl->conn, file, getxml_flags)
Since these are not usual #define and #include (for example of what I
mean usual see head of virsh.c) but rather a part of function I think
they should be indented like other statements inside the function.
Therefore we need slightly different approach:
https://www.redhat.com/archives/libvir-list/2012-June/msg00055.html
Maybe others should step in and share their opinions.
I actually prefer Wen's solution, and think it is better to be
consistent with # processing even if it looks odd in function context.
ACK to Wen's version.
--
Eric Blake eblake(a)redhat.com +1-919-301-3266
Libvirt virtualization library
http://libvirt.org