On Wed, Sep 19, 2012 at 12:32:18PM -0600, Eric Blake wrote:
On 09/19/2012 12:08 PM, Laine Stump wrote:
This new virsh command uses the new virNetworkUpdate() API to modify an existing network definition, and optionally have those modifications take effect immediately without restarting the network.
+ +static const vshCmdOptDef opts_network_update[] = { + {"network", VSH_OT_DATA, VSH_OFLAG_REQ, N_("network name or uuid")}, + {"command", VSH_OT_DATA, VSH_OFLAG_REQ, + N_("type of update (add, delete, or modify)")},
s/add/add-first, add-last/
+ {"section", VSH_OT_DATA, VSH_OFLAG_REQ, + N_("which section of network configuration to update")}, + {"xml", VSH_OT_BOOL, 0, N_("xml is specified directly on commandline")}, + {"file", VSH_OT_BOOL, 0, N_("file containing xml is specified on commandline")}, + {"xmldata", VSH_OT_DATA, VSH_OFLAG_REQ, + N_("complete xml element (or name of file containing xml) to add/modify, " + "or to be matched for search")},
Interesting choice to make --xml and --file be boolean flags, and '[--xmldata] data' be the string that becomes either the file name or the xml content. I might have done just two optional VSH_OT_DATA arguments for --xml and --file and then manually checked that exactly one of the two was supplied, instead of using three arguments. But what you did works, so no need to change it.
Yes, this is total overkill. When I suggested --xmldata as an option my intent was that you'd have the following usage virsh net-update /path/to/xmlfile virsh net-update --xmldata "some XML data inline" IMHO, requiring --xml and --file in addition is unfriendly to the user too Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|