IR test suite question regarding parse time errors

• From: Brian P. Davis <bpdavis@lucent.com>
• Date: Tue, 10 Feb 2004 10:51:02 -0500
• To: <www-voice@w3.org>
• Message-ID: <001501c3efed$b0076c70$75545c87@ascc.lucent.com>

Hi,

There are several tests in the VoiceXML 2.0 Implementation Report test suite
that involve one document calling a second document that contains some type
of invalid voicexml markup.  In these tests, it is considered a pass when a
bad fetch is caught in the first (calling) document.  For example,
http://www.w3.org/Voice/2003/ir/1135/1135.txml.  

These tests imply that invalid markup (like missing attributes or
conflicting attributes) should be detected when the document is initially
loaded, rather than when the interpretation of offending element begins.

What is the basis in the specification for these tests?  Does the spec
require this behavior?  I have not been able to find a reference, but I feel
like I must be missing it.

Any clarification/assistance that can be provided would be much appreciated.

Regards,
Brian

Received on Tuesday, 10 February 2004 10:54:01 UTC