It looks like it's the right way to go. We will also need to retrofit
the bean-level case. In that situation, we will need to inherit the
inIterable, key and index values of the node being suppressed.
https://gist.github.com/emmanuelbernard/f73bff117cff8195e41f
I will try and do a write up for the spec.
https://hibernate.onjira.com/browse/BVAL-410
Emmanuel
On Tue 2013-02-19 13:00, Gunnar Morling wrote:
Hi,
We added the method addParameterNode() to the node builder API, allowing
for the creation of custom subpaths in cross-parameter constraint
validators.
Currently such a path will look like this (the first two nodes are the
default ones, the remain ones are custom created nodes):
METHOD -> CROSS_PARAMETER -> PARAMETER -> ...
I'm wondering whether it wouldn't be better to get rid of the
cross-parameter node in this case:
METHOD -> PARAMETER -> ...
E.g. when looking at the following example:
@OldAndNewPasswordDifferent
public void resetPassword(String new, String old) { ... }
Then I think the following path might be what constraint authors want:
MethodNode("resetPassword") -> ParameterNode("old", arg1)
As it seems, the RI does the same when custom nodes are added in
class-level constraint validators (i.e. no node for the class-level
constraint will be part of the path). Shortly looking in the spec, I
couldn't find an example mandating this behavior, though.
Any thoughts?
--Gunnar
_______________________________________________
beanvalidation-dev mailing list
beanvalidation-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/beanvalidation-dev