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

[Cullen Jennings <[e-mail]>]

On Apr 24, 2012, at 8:51 , Ned Freed wrote:

> 
>> On Apr 23, 2012, at 18:14 , Ned Freed wrote:
> 
>>>>> application/senml+json
>>>>> application/senml+xml
>>>>> application/senml-exi
> 
>> I think the idea that first two are OK but the third one needs to have a "-"
>> instead of a "+" is simply baffling. Can someone explain the logic to me ?
> 
> +exi implies a syntax that can be processed generically by systems knowing

Thanks - that answer does help. But what do you mean can be "processed generically"? If we mean can check the syntax is valid, sure, all can do that. If we mean understand the semantic content of the data without the schema information,  it seem pretty unlikely that senml should have the + for json or xml. 

I will point out the senml mandates that data inside has to identify the schema in use for that inside the exi data - now it's not a globally resolvable URI for the schema or something but it does make sense in the context of senml. You can process the senml exi without the schema at least in the context of determining it might be valid exi. 

But really the questions is what's is our definition of "generically processable" ?


> nothing specific about senml. But in order to decode schema-aware EXI, which is
> what is being used here, you have to know the schema for the specific version
> of senml that's in use. There is an field in the document itself that can be
> used to provide a pointer to the schema, but it's optional. In other words, in
> some cases the schema has to be known through out of band information, and that
> violates the rules for +suffix.
> 
>> I'll rest the urge to insert a rant here about the reason people don't bother
>> to register new thing and just use text/plain for everything …
> 
> In turn I'll resist the urge to rant about folks who jump into a long
> discussion without making any attempt to read the preceeding messages and
> expect others to do their work for them.

I've really had read the whole thread when I sent the previous email. I think you will agree with me that many people find it is hard to register a new type like this. I applaud you doing the draft-ietf-appsawg-media-type-regs draft and the work on draft-hansen-media-type-suffix-regs to try and make this easier.