Hi
On Wed, Aug 26, 2015 at 2:03 PM, Daniel P. Berrange <berra...@redhat.com> wrote:
> +ObjectProperty *
> +object_class_property_add(ObjectClass *klass,
> + const char *name,
> + const char *type,
> + ObjectPropertyAccessor *get,
> + ObjectPropertyAccessor *set,
> + ObjectPropertyRelease *release,
> + void *opaque,
> + Error **errp)
> +{
> + ObjectProperty *prop;
> + size_t name_len = strlen(name);
> +
> + if (name_len >= 3 && !memcmp(name + name_len - 3, "[*]", 4)) {
> + int i;
> + ObjectProperty *ret;
> + char *name_no_array = g_strdup(name);
> +
I question the need for dynamic/array property name registered in
classes. What would be more useful is an array property instead. It
would help to introspect classes for dynamic "children[*]" case.
object_property_add_child() could verify/check against the class
declaration, and grow the instance properties list (like it does now,
but it would be only for instances of children[] items). On
introspection of classes, the class "children[*]" property would be
visible, but would be hidden when introspecting the instance, and you
wouldn't be able to lookup that "array" property.
It seems relatively straightforward to deal with the link<> case, by
storing the offset of the "child" pointer. This seems fine if limited
to a single link<> (it should probably check the prop is not of the
name[*] style already), ex:
https://gist.github.com/elmarco/905241b683fb9c5f2a08
Your patches looks good to me in general but object_property_del()
should be fixed, since the prop find may belong to the class.
cheers
--
Marc-André Lureau
