> The problem is that if you set the submitted value to null, the > validation will not be triggered. But this whole issue is something we > need to get over again, I just got rid of this else part cause > checkboxes and radiobuttons were using their own decoding anyways; and > for normal inputs the else branch is not necessary.
That's actually the behavior required in the spec (AFAIK). If the submitted value is null, no validation is performed. See this snip by Craig from the JSF forum: [quote] The expert group considered this decision carefully, and concluded that the current behavior was the best -- if there is no input data, skip any specified validations -- because of the likelihood for things like null pointer exceptons in validators that are not prepared for this scenario. [/quote] If you want the value to be required, you can still get validation by setting the required attribute (you just can't use a custom validator.) > hope that fixes some of the issues. Yes it seems to fix some issues related to dataTable and sorting (see my comments on MYFACES-228) but there are still some other (unrelated) issues to discuss as far as that goes. Also, not only does this fix some issues but it seems to be the mandatory behavior according to the spec ;-) Someone please correct me if I'm wrong on that. > Martin sean
