I second that this issue needs to be addressed, as it's quite tiresome to code work-arounds for this problem. While not the most elegant solution, the errors DOM emits when validating an XML document against a schema can be captured, parsed and processed at the script level to emulate the functionality you described. 

From an implementational standpoint, I agree with your idea in principle, though I believe that a few modifications are necessary:

Let's assume that a user is going to design an XML Schema to validate user-submitted form data, which will be validated with DOM (using XPath): this data will be grouped into a set of elements, and then validated. 

In order to best represent the errors for this scenario, it would be helpful for DOM to throw a few basic pieces information, including: The parent path of the faulty element, the name of the element, a description of the error and the error -code-. In place of an error -number-, using the W3C-defined error codes for the XML Schema Structures specification would provide a consistant means of reporting invalidities. You could then, as you demonstrated, use a simple switch statement to process the errors as needed.

Though I believe the above would be well suited to addressing the problem outlined by this bug, as mentioned in the above scenario, it would also be helpful to know only the elements that -failed- to validate. A method of "reverse-validating" a document would be quite helpful, in that it would simply report the elements that failed to validate. This could be done at the script level if the functionality was not provided natively by PHP.