On 08/20/2012 06:46 AM, Osier Yang wrote:
On 2012年08月20日 14:17, Laine Stump wrote:
> Currently there is no practical difference between this and using the
> old virNetworkDefinXML API, it just allows using the new API to make
> sure it's working.
>
> Soon there will be a flag defined for the new function (to request
> changes in re-definitions of existing+active networks take effect
> immediately), and virsh will add support for that flag.
> ---
> tools/virsh-network.c | 12 +++++++++++-
> 1 file changed, 11 insertions(+), 1 deletion(-)
>
> diff --git a/tools/virsh-network.c b/tools/virsh-network.c
> index 49ec34f..414ca86 100644
> --- a/tools/virsh-network.c
> +++ b/tools/virsh-network.c
> @@ -181,6 +181,7 @@ cmdNetworkDefine(vshControl *ctl, const vshCmd *cmd)
> const char *from = NULL;
> bool ret = true;
> char *buffer;
> + unsigned flags = 0;
>
> if (!vshConnectionUsability(ctl, ctl->conn))
> return false;
> @@ -191,7 +192,16 @@ cmdNetworkDefine(vshControl *ctl, const vshCmd
> *cmd)
> if (virFileReadAll(from, VIRSH_MAX_XML_FILE,&buffer)< 0)
> return false;
>
> - network = virNetworkDefineXML(ctl->conn, buffer);
> + network = virNetworkDefineXMLFlags(ctl->conn, buffer, flags);
> +
> + if (network == NULL&& last_error->code == VIR_ERR_NO_SUPPORT) {
It should reset the error before trying the old API:
vshResetLibvirtError();
Yep, I missed that. Thanks!
> + /* fall back to older function that doesn't take flags. note
> + * that if we want to use flags in the future, we'll have to
> + * error out here when flags != 0.
> + */
> + network = virNetworkDefineXML(ctl->conn, buffer);
> + }
> +
> VIR_FREE(buffer);
>
> if (network != NULL) {
ACK with the nit fixed (with wondering why not use new API
in virsh when the flags is/are supported though).
Just a way of assuring that everything is in place and working. :-)