On 11.02.2014 10:54, Daniel P. Berrange wrote:
On Mon, Feb 10, 2014 at 07:52:34PM +0100, Michal Privoznik wrote:
@@ -3583,6 +3639,48 @@ validate: } }
+ /* finally we can call the 'plugged' hook script if any */ + if (virHookPresent(VIR_HOOK_DRIVER_NETWORK)) { + /* the XML construction is a bit complicated here, + * as we want to pass both domain XML and network XML */ + virBuffer buf = VIR_BUFFER_INITIALIZER; + char *xml, *net_xml, *dom_xml; + int hookret; + + net_xml = virNetworkDefFormat(netdef, 0); + dom_xml = virDomainDefFormat(dom, 0); + + if (!net_xml || !dom_xml) { + VIR_FREE(net_xml); + VIR_FREE(dom_xml); + goto error; + } + + virBufferAdd(&buf, net_xml, -1); + virBufferAdd(&buf, dom_xml, -1);
This isn't very easy for applications to consume. With all the other things you can just pass stdin straight to your XML parser and process it. When you concatenate 2 XML docs this way it becomes much harder, since you have to split the two docs which means parsing them without an XML parser. Usually you'd want to place the 2 docs within a parent XML doc so the overall result is still a single wellformed XML doc.
Okay, but how should the XML look then? <hookData> <network/> <domain/> </hookData> I'm basically asking about the root element name. Then, if we do this for plug & unplug - should we follow the same scheme for network start & stop? Without the <domain/>. You know, once we require hook script to take network xml from /hookData/network we can do that for other cases too, so the XPath to select the network doesn't change. Or is that an overkill, because the hook script can easily do: if [ "$2" == "plugged" -o "$2" == "unplugged" ]; then XPATH="/hookData/network" else XPATH="/network" fi Michal