On Wed, Jul 18, 2012 at 03:39:39PM +0200, Christophe Fergeau wrote:
For objects with a subtype 'type' attribute, when the
_new_from_xml
function was called, the 'type' attribute was forcefully set to the
right value rather than checking that the passed-in value matches
the type of the subclass we are trying to instantiate. This commit
changes this, and returns NULL when the value of the 'type' attribute
of the passed-in XML document does not match the expected type.
---
libvirt-gconfig/libvirt-gconfig-domain-address-pci.c | 5 ++++-
libvirt-gconfig/libvirt-gconfig-domain-address-usb.c | 5 ++++-
libvirt-gconfig/libvirt-gconfig-domain-chardev-source-pty.c | 5 ++++-
libvirt-gconfig/libvirt-gconfig-domain-chardev-source-spicevmc.c | 5 ++++-
libvirt-gconfig/libvirt-gconfig-domain-controller-usb.c | 5 ++++-
libvirt-gconfig/libvirt-gconfig-domain-graphics-sdl.c | 5 +++--
libvirt-gconfig/libvirt-gconfig-domain-graphics-spice.c | 5 +++--
libvirt-gconfig/libvirt-gconfig-domain-graphics-vnc.c | 5 +++--
libvirt-gconfig/libvirt-gconfig-domain-interface-bridge.c | 5 +++--
libvirt-gconfig/libvirt-gconfig-domain-interface-network.c | 5 +++--
libvirt-gconfig/libvirt-gconfig-domain-interface-user.c | 5 +++--
libvirt-gconfig/libvirt-gconfig-domain-timer-pit.c | 5 ++++-
libvirt-gconfig/libvirt-gconfig-domain-timer-rtc.c | 5 ++++-
13 files changed, 46 insertions(+), 19 deletions(-)
I've wondered about this before too and never got around to fixing it.
I'm concerned about whether all callers will check for possibility of
returning NULL. Most glib apps assume _new() calls can't return
NULL, since we abort() on OOM.
Should we instead be using g_return_if_fail() so you get a dianostic,
on the basis that this is programmer mistake ?
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 :|