On Thu, 2004-04-15 at 00:13, Sylvain Wallez wrote:
> Bruno Dumon wrote:
>
> >On Wed, 2004-04-14 at 11:40, Sylvain Wallez wrote:
> >
> >
> >>Hi all,
> >>
> >>A few weeks ago, I generalized validation to every widgets, and not only
> >>fields. The main visible result of this is that <fd:validation> is now a
> >>direct child of <fd:field> (or other widget-defining element), and no
> >>more a child of <fd:datatype>.
> >>
> >>Validation as a child of <fd:datatype> is still supported as a legacy
> >>behaviour, but I would like to remove it to drive people towards the new
> >>generalized validation. My plan to ease the migration is to raise a
> >>meaningful exception ("fd:validation has moved") whenever we encounter a
> >>fd:validation inside a fd:datatype.
> >>
> >>
> >
> >Just been looking a bit at this...
> >
> >For the validation rules associated with datatypes, it is checked that
> >the validation rule is compatible with the datatype (i.e. can handle the
> >kind of objects that the datatype represents). Does such a check still
> >happen? Or will this give errors at runtime?
> >
> >
>
> Damn, missed that one...
>
> The problem with the generalized validation is that it is erm...
> generalized. And as such validation rules can be applied to any kind of
> widget, including those that don't have a value.
>
> A solution is to generalize this check also, by introducing a
> ValidationRule.canValidate(WidgetDefinition) which, in the case of typed
> validators, will get the widget definition's datatype to perform today's
> check.
sounds good.
While we're at it, I propose we also move the package
o.a.c.forms.datatype.validationruleimpl to o.a.c.forms.validators
--
Bruno Dumon http://outerthought.org/
Outerthought - Open Source, Java & XML Competence Support Center
[EMAIL PROTECTED] [EMAIL PROTECTED]
