When I get through my hurdles (I just sent another one) I will make an
example for bug-checking.
Actually, after I did 2 or 3, the constraint errors went away so it wasn't
so bad. But it still may be an issue.
Thanks for getting back to me.
On Wed, Mar 26, 2014 at 3:04 PM, Hardy Ferentschik <hardy(a)hibernate.org>wrote:
a couple of more remarks on top of Emmanuel's.
First of all, as Emmanuel is saying method and constructor validation
should behave the same
on Glassfish and Wildfly. The Bean Validation specification (part of EE 7)
says that method and
constructor validation occurs per default as soon as Bean Validation
constraint annotations are
used on methods or constructors. You find the relevant specification
section here -
To answer your question on how to disable validation. You can add a
validation.xml file disabling
<?xml version="1.0" encoding="UTF-8"?>
Last but not least, open an issue with test case, so that we can check
whether there is a bug or not.
On 26 Jan 2014, at 16:40, Emmanuel Bernard <emmanuel(a)hibernate.org> wrote:
> Does your offending method actually override another method?
> Because that would definitely be an error on your part.
> Glassfish uses Hibernate Validator for the validation engine so I don't
> think your code is at fault here. Unless they have a broken integration
> and nothing is validated.
> But it could be due to a bug on how Weld does subclassing for proxies and
> copy constraint annotations but I thought we had that base covered
> The best would be to open an issue with a reproducible minimal test
> case. From there we could either fix the offending bug and if there is
> none see how we could report several errors in one go.
> On Wed 2014-03-26 10:16, Mike Hostetler wrote:
>> I'm probably one of the many that is moving away from Glassfish to
>> I have done some trivial changes but now Weld is giving validation
>> like the following on some of our EJB3 beans:
>> javax.validation.ConstraintDeclarationException: HV000151: A method
>> overriding another method must not alter the parameter constraint
>> The worse thing is that only see one at a time: so I fix one, do the
>> build-redeploy dance, and then see the next one. Most of the problems is
>> that in the overriding class we use javax.validation.constraints.NotNull
>> annotation on the parameter, which is what Weld doesn't seem to like.
>> Obviously these worked in Glassfish.
>> My first thought is: how can I shut off Weld validation? If that is not
>> possible (which I'm willing to accept), then how can I see these
>> errors all at once instead of one at a time during deployment?
>> Mike Hostetler
>> wildfly-dev mailing list
> wildfly-dev mailing list