Re: [apps-discuss] Fwd: New Version Notification for draft-shelby-exi-registration-01.txt

[Zach Shelby <[e-mail]>]

On Apr 13, 2012, at 12:34 AM, Ned Freed wrote:

>> So how would you handle the PERs and (schema-informed) EXIs of this world?
>> Should it be -exi, not +exi?
> 
>> (As in application/senml-exi?)
> 
> That's definitely an option - an informal sort of suffix. Another alternative
> would be to define some other formalism (it really cannot be "-" since other
> media types use "-" for other purposes). Looking at the ABNF of allowed
> characters, carat is allowed and I don't think it has ever been used. OTOH, how
> many things like EXI are ever going to show up?

This could be the right approach to look at. As I wrote in a previous mail there are a few types of serializations that require OOB dictionary information such as EXI schema-informed mode, ASN.1 PER, and Google Protocol Buffers.     

So application/senml^exi would need to result in some sort of registry entry where information on how to determine the OOB dictionary is easily available to anyone who runs across that media type. It seems to me that the media type registry already contains quite a few fields that could be used for that purpose -- for example "interoperability considerations". 

Zach

-- 
Zach Shelby, Chief Nerd, Sensinode Ltd.
http://www.sensinode.com
http://zachshelby.org  - My blog "On the Internet of Things"
http://6lowpan.net - My book "6LoWPAN: The Wireless Embedded Internet"
Mobile: [phone]