W3C home > Mailing lists > Public > www-voice@w3.org > October to December 2007

RE: VXML containing SRGS grammars with semantic errors

From: Andrew Hunt <andrew.hunt@holly.com.au>
Date: Wed, 3 Oct 2007 16:51:08 -0700
Message-ID: <14F99AAB26DBE04E87B64C25258794820423F9D0@ehost005-2.exch005intermedia.net>
To: "Ian Sutherland" <ian.sutherland@oracle.com>, <www-voice@w3.org>
I don't think there is a requirement for VoiceXML document validation to
verify correctness within an inline grammar (excepting XML correctness).
The implied behaviour is to throw a error.badfetch.grammar.syntax error
(S5.2.5) when activating a grammar at a wait-state and according to the
scope of that state.  This reflects the typical implementation in which
grammars are handled to a specialist ASR component (Grammar Processor)
for verification, activation and recognition.  I agree that behaviour
could be more explicitly defined.
Andrew Hunt
Holly Connects


From: www-voice-request@w3.org [mailto:www-voice-request@w3.org] On
Behalf Of Ian Sutherland
Sent: Wednesday, 3 October 2007 9:22 AM
To: www-voice@w3.org
Subject: VXML containing SRGS grammars with semantic errors

Does the VoiceXML 2.0 standard define what the response of a VoiceXML
browser should be to an SRGS grammar that is syntactically well-formed
and valid, but which has a semantic error, such as a <ruleref> that
refers to a <rule> id that doesn't exist?  The SRGS IR tests contain
several grammars like this that are supposed to be rejected by the
grammar processor, but how is this supposed to manifest in VoiceXML if a
VoiceXML document contains such a grammar inline, or refers to such an
external grammar?

My cursory inspection of the recommendation didn't turn up any guidance
on this.  I'm particularly interested in what event, if any, is thrown,
in what document, and under what circumstances (i.e. only when the
grammar is activated?)?  Even if the recommendation doesn't say exactly
what should happen, it seems desirable that it at least define the range
of possible behaviors.
Received on Wednesday, 3 October 2007 23:50:55 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:07:40 UTC