Fwd: Use of HTTP in VoiceXML3

[Dan Burnett-3]

CC'ing www-voice.

Begin forwarded message:

From: Dan Burnett <dburnett@...>

Date: May 4, 2009 6:07:24 AM EDT

To: Lisa Dusseault <Lisa.Dusseault@...>

Subject: Re: Use of HTTP in VoiceXML3

Hi Lisa,

Thanks for your comments.  We will review these and likely ask for your help with the more detailed review that you suggest.

Dan Burnett
Co-editor, VoiceXML 3


On Feb 25, 2009, at 3:13 PM, Lisa Dusseault wrote:

Hi,

As a part of reviewing http://tools.ietf.org/html/draft-ietf-mediactrl-vxml-04, I had occasion to look at the recommendations for VoiceXML 2.0 and 2.1.  I noticed that use of POST is quite underspecified!  Some examples:

- The Content-Type has to be inferred, as it's not required to be anything

- I assume 'next' is used to construct the URI to send the GET or POST to but this isn't explained or referenced

When HTTP is used to tunnel application messages, we're coming to understand a bunch of additional things must be said to ensure interoperability.  In a case like this:

- What are the appropriate HTTP success codes?  Failure codes?

- MUST clients follow redirects?  Or are redirects completely inappropriate for the server to return?

Finally, some requirements although perfectly clear in RFC2616 are missed more often than they're implemented in applications like this. For example:

- Servers MUST handle conditional headers

- Servers MUST handle chunked transfer-encoding

At some point, review by IETF applications review team or HTTP WG members would be terrific for VoiceXML 3, and I can help find volunteers.  I also noticed that VoiceXML 3 goes further in its specification of caching, and we should make sure that synchs up well with the work in the HTTPBIS working group on caching.

Thanks,

Lisa Dusseault

Applications Area Director, IETF.

--- Scanned by M+ Guardian Messaging Firewall ---

Messaging Architects sponsors The Spamhaus Project.