On Thu, 2004-05-06 at 19:15, Sylvain Wallez wrote:
> Bruno Dumon wrote:
>
> >On Thu, 2004-05-06 at 16:23, [EMAIL PROTECTED] wrote:
> >
> >
> >>vgritsenko 2004/05/06 07:23:04
> >>
> >> Modified: . status.xml
> >> src/blocks/forms/samples sitemap.xmap welcome.xml
> >> src/blocks/forms/java/org/apache/cocoon/forms/flow/javascript
> >> Form.js
> >> src/blocks/forms/java/org/apache/cocoon/forms/flow/javascript/v2
> >> Form.js
> >> src/blocks/forms/java/org/apache/cocoon/forms/flow/javascript/v3
> >> Form.js
> >> src/blocks/forms/samples/forms sitemap.xmap
> >> Removed: src/blocks/forms/samples/flow customvalidationdemo.js
> >> src/blocks/forms/samples/forms customvalidationdemo_form.xml
> >> customvalidationdemo_template.xml
> >> Log:
> >> Remove flow level custom validators
> >> http://marc.theaimsgroup.com/?t=108091920700001&r=1&w=2
> >>
> >>
> >>
> >
> >(I highly appreciate the cleanup work, but...)
> >
> >Allthough the flow-level validator function was a hack, won't it be too
> >annoying for users who are relying on it to throw it out completely?
> >This will make the woody -> cforms move a bit harder...
> >
> >
>
> Cleaning up means so backwards incompatibilities, and I'm +1000 to
> remove this hack. What we can do however, is check if a "validator"
> property exists in a form object and fail hard with an exception if it
> exists. That way, users will be informed that this "feature" has been
> removed.
ok, sounds good.
>
> >Also, couldn't the example be updated to how it should be done now?
aha, Sylvain mentions another sample of custom validation below so that
should be fine.
> >
> >Lastly, the behaviour of the WidgetValidators are not yet a complete
> >replacement for the validator function, see this important BUT in
> >AbstractContainerWidget:
> >
> > public boolean validate() {
> > // Validate self only if child widgets are valid
> > //TODO: check if we should not change this to still validating
> >kids first
> > // BUT also validating the top level
> > if (widgets.validate()) {
> > return super.validate();
> > } else {
> > return false;
> > }
> > }
> >
> >Is there anyone who knows a reason why the parent shouldn't be validated
> >if kids fail? Seems to be too limitting to me.
> >
> >
>
> Ditto. A validator relying on other widget's values (be them children or
> not) must be consider these widgets do be potentially invalid. And there
> are some valid use cases in our "form1" sample: the "contacts" repeater
> checks uniqueness of contact names among rows, and this validation makes
> sense even if emails are invalid.
yep. Would be nice if this change could still make it before the code
freeze (I probably won't have time).
--
Bruno Dumon http://outerthought.org/
Outerthought - Open Source, Java & XML Competence Support Center
[EMAIL PROTECTED] [EMAIL PROTECTED]
