Re: [hibernate-dev] ValidationMode

2013-03-04 Thread Steve Ebersole
persistence provider portability. On Mon 04 Mar 2013 09:57:29 AM CST, Emmanuel Bernard wrote: > Applying validation at both levels has a few benefits: > > - DDL means that constraints are also applied to non Java apps or even >third party apps not using Bean Validation > - applying the constra

Re: [hibernate-dev] ValidationMode

2013-03-04 Thread Emmanuel Bernard
Applying validation at both levels has a few benefits: - DDL means that constraints are also applied to non Java apps or even third party apps not using Bean Validation - applying the constraint during the CRUD events provides more contextual exception reports, better than what we can do by co

Re: [hibernate-dev] ValidationMode

2013-03-04 Thread Steve Ebersole
Partially. What it also says is too perform double checking (callback and db constraints) when BV is present. I guess the reason you want to allow both is because we do not (and really can not) convert all validations to db contraints. My point is just that Set seems total overkill when there

Re: [hibernate-dev] ValidationMode

2013-03-04 Thread Emmanuel Bernard
I don't follow you. It seems to make sense to me to have ddl and callback. It says validate entities on CRUD events, apply database generation in accordance to the constraints and raise an exception if BV is not present in the CP. On Mon 2013-03-04 7:36, Steve Ebersole wrote: > My concern is no

Re: [hibernate-dev] ValidationMode

2013-03-04 Thread Steve Ebersole
My concern is not the additional option. My concern is the allowance of multiple selections. The only combo that ever even conceivably makes sense is "ddl" and "callback" together. But I am questioning how sensible it is really to have those 2 together. IMHO not much sense at all. So IMO we

Re: [hibernate-dev] ValidationMode

2013-03-04 Thread Emmanuel Bernard
Our impl allow multiple choices to be able to add ddl as an additional behavior as it (was) not defined in the spec. And indeed, auto uses the degraded mode and callback forced BV to be used and raise an exception if it is not present. I can see apps deciding for one or the other. Emmanuel On Fr

Re: [hibernate-dev] ValidationMode

2013-02-22 Thread Steve Ebersole
See inline... On Fri 22 Feb 2013 02:38:32 PM CST, Hardy Ferentschik wrote: > > On 22 Jan 2013, at 9:17 PM, Steve Ebersole wrote: > >> Kind of. What I mean is that the HEM integration code is allowing Set >> . So our impl allows a value like "auto, ddl" or other >> combos. I am not able to gr

Re: [hibernate-dev] ValidationMode

2013-02-22 Thread Hardy Ferentschik
On 22 Jan 2013, at 9:17 PM, Steve Ebersole wrote: > Kind of. What I mean is that the HEM integration code is allowing Set > . So our impl allows a value like "auto, ddl" or other > combos. I am not able to grok why we allow multiples. I think this is a Hibernate specific "feature" to suppo

Re: [hibernate-dev] ValidationMode

2013-02-22 Thread Steve Ebersole
Kind of. What I mean is that the HEM integration code is allowing Set . So our impl allows a value like "auto, ddl" or other combos. I am not able to grok why we allow multiples. Also auto and callback are slightly different. callback causes an error if no bv provider is available whereas auto

Re: [hibernate-dev] ValidationMode

2013-02-22 Thread Hardy Ferentschik
On 22 Jan 2013, at 6:26 PM, Steve Ebersole wrote: > Any idea why the ORM/JPA Bean Validation integration code allows > multiple validation modes to be specified? Are you talking about "auto", "callback", "none"? Well, "callback" to enable validation, "none" to disable it. Why there is also "