W3C home > Mailing lists > Public > www-forms@w3.org > July 2002

RE: Managing Validation Error Messages

From: Dan Dennedy <DDennedy@digitalbang.com>
Date: Wed, 31 Jul 2002 13:02:58 -0400
Message-ID: <11DC9B322580C843AE5A3D030D950B03031593@intrabang01.digitalbang.com>
To: <werner.donne@re.be>, <www-forms@w3.org>

> From: Werner Donné 
> Dan,
> I think it is the intention that validation reveals the exact 
> elements in the instance data that violate a rule. These 
> elements then must be mapped back to the corresponding form 
> control. The form contains all the necessary information to 
> do that. Then, according to section 4.3.12, an alert event is 
> sent to the form control. Its default processing is to 
> display an error message to the user. The alert element is 
> available for this. It is up to the XForms implementation to 
> decide how to present it to the user. If the implementation 
> is a component, this can be left to its client, who in 
> principle decides on the error reporting policy. The client 
> simply registers a DOM event handler on the form controls to do it.

Correct, and xforms:invalid is the relevant event, but it contains no context info. There is nothing to indicate *what* is wrong--only *where* it is wrong--which contol and thus which instance. Furthermore, there are both Schema rules/facets and xforms constraints. These ought to be made available in a unified manner.

> I therefore believe all the necessary mechanisms are 
> described in the specification. Personally, I would always 
> use the alert elements, because end-users will be in contact 
> with the forms. It makes no sense to confront them with 
> technical validation messages. No matter how sofisticated XML 

Well, my messages are not very technical, and I believe they are sensible. However, there is always disagreement on such matters.

> Schema may be, it will in general not be possible to derive 
> user-friendly messages from the post-schema-validation 
> infoset. The form author, on the other hand, knows what the 
> user is expected to enter in a certain form control. He can 
> formulate this with an end-user-oriented message.
This is what I was referring to. It seems the best message an author can do at the moment is to simply say the field is invalid and identify the rules in their own terms.
Received on Wednesday, 31 July 2002 13:03:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:36:06 UTC