On Wed, Oct 07, 2009 at 05:04:40PM -0400, Laine Stump wrote: > (Yes, you read the subject correctly - 3 of 2!) > > The minimal XML returned from ncf_if_xml_state() doesn't contain these > attributes, and it was preventing it from passing through the > parse/format step (making my patches from earlier today unusable). > > Because adding an "UNSPECIFIED" value to the enums for type and > startmode would have made it possible for someone to send xml with, > eg, type="unspecified" (which we don't want), I've followed the > convention already used for the dhcp "peerdns" attribute - if it isn't > specified, the value is -1. IMHO 'type' should always be a mandatory attribute. There's no reason why the live XML dump should not include it - we shouldn't make it optional merely because it hasn't been implemented fully yet. For anything which is an enum, the default value if not specified should always be 0. This is because when allocating a new object, everything will be memset to 0, so 0 should be a sensible default. When declaring the enum <-> string mapping, simply put "" as the value for that if you don't want it to be usable for incoming XML. When generating the XML output doc, you can just make it conditional on if (if->startmode), so the default value is skipped when formatting the XML Daniel -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :| -- Libvir-list mailing list Libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list