VoiceXML 2.0 spec quesiton/errata

We have had a couple of issues with our browser (easy to fix, so this is an
informative email) that I think point to some things that should be
clarified or noted in the spec.  I thought I'd post them for anyone looking
at an errata or updated document (not new features, so I'm not referring to
2.1 or 3.0).
1)	In 5.3.2, Table 46, the "expr" attribute is defined as "The new
value of the variable." The examples make it obvious that it is an
ECMAScript expression that evaluates to the new value of the variable.  The
reason I got to looking closely is we had a philosophical disagreement on
what <assign name="foo" expr="" /> means.  The expr value is not valid
ECMAScript and although the intent is obvious the spec wording is not.  It
seems that more formally it should be expr="null". The "" seems to be in
common usage so the practical answer is easy (allowing it does not cause any
loss of generality). An easy spec clarification would be "A valid ECMAScript
expression evaluating to the new value or a null parameter which will be
evaluated to ECMAScript null." I think a similar correction in <var> is
appropriate as well.
2)	Not quite the same but a similar issue came up on <subdialog
fetchaudio="">.  The "fetchaudio" is defined to be a uri and there is no
formal definition of a uri allowing nothing as a value. It is not clear if
this is an error (which the spec is clear should be ignored) or a valid
"null" value. Since in either case the result is to continue and play
nothing, this one is definitely a small point, but worth clarifying for
future developers. I would recommend a statement like "If the fetchaudio
attribute is present with no value or an empty string, no audio is played
and no error is thrown".
For what it's woprth.
Ken Waln
V.P of Engineering and C.T.O.
Edify Corporation
(408) 982-2050

Received on Friday, 20 May 2005 16:39:34 UTC