[Html-widget] v1.08 released

[Carl Franks]

On 14/08/06, Carl Franks <fireartist at gmail.com> wrote:
> On 14/08/06, Adam Sjøgren <asjo at koldfront.dk> wrote:
> > On Mon, 14 Aug 2006 16:07:25 +0100, Carl wrote:
> >
> > > If for some reason someone creates a constraint like so:
> > > $w->constraint( In => "foo" )->in( @values );
> > > and @values is empty - the constraint will not mark "foo" as invalid,
> > > regardless of the submitted value of "foo".
> >
> > > Now that you bring it up, I wonder why it was necessary. I'll look
> > > through the mail archive and see what problem it was supposed to
> > > solve.
> >
> > I think it originated in this email: <7pu084munn.fsf at sille.nzcorp.net>
> > on the Catalyst mailinglist - in which I have an AllOrNone-constraint
> > that includes a Select-element. I would like it to be valid to not
> > choose anything in the drop-down and not fill out the rest of the
> > elements in the AllOrNone-constraint.
> >
> > So either I need a way to not have/remove the In-constraint on the
> > Select, or the In shouldn't mark empty as invalid - either way, or a
> > third, is fine with me.
>
> Hmm, ok. This fix doesn't even solve your original problem - as it's
> not taking into account whether the key was submitted - only whether
> $constraint->in() contains values.
>
> I also recall sri saying he wanted the implicit In constraint on
> Select elements to be scrapped - too much magic.
> That at least would fix your problem - though it would break backwards
> compatability for people relying on that constraint being silently
> added...
>
> I'm going to leave this until tomorrow, and hope sri's reading this
> and makes a call.
> I think the 2 choices are:
> remove the implicit In constraint, but add an accessor to keep the
> current behaviour,
> or
> add an accessor to override current behaviour by not adding the
> constraint (which was Adam's original suggestion!).
>
> Any opinions / other ideas?

Ok, I've removed the implicit constraints that were being added to
Select and RadioGroup elements.
I've also rolled-back the "In constraint no longer fails when the in()
list is empty" change, as it's redundant.

Carl