Re: [core] MULTIPART-CT: A few editorial comments

[Carsten Bormann <[e-mail]>]

Hi Christer,

thank you for alerting us to this oversight.
(And thank you for the critical reading in the first place!)

> Q1:
> 
> The Introduction says talks about “request or response body’.
> 
> However, when looking in RFC 7252, I don’t find ‘body’ anywhere.

Right.  Body is an RFC 7959 term (which was adapted from HTTP’s “Message Body”, obviously).

I didn’t really want to muddy the picture by explaining again that CoAP block-wise transfers can cut up a single body into multiple payloads (as per RFC 7959).  The current text already might be misunderstood to say that this media type is CoAP specific, which it is not at all.

We didn’t go to the trouble of having a formal terminology import section (apart from the usual BCP14 boilerplate in 1.1).  We could add that text, reference 7959 and 7230 as well, and make this abundantly clear.  Or we could leave it as it is.

> I assume that ‘body’ refers to what RFC 7252 calls ‘payload’. If so, I suggest aligning the terminology.
> 
> (For some reason section 10.2.3 of RFC 7252 does use ‘message-body’, without any explanation or reference, but I guess that is just an error)

That is about the message-body on the HTTP side, so it is exactly as it should be.

(Actually, 1.2 of 7252 says 
 »This specification requires readers to be familiar with all the terms
  and concepts that are discussed in [RFC2616], including "resource",
  "representation", "cache", and "fresh".  (Having been completed
  before the updated set of HTTP RFCs, RFC 7230 to RFC 7235, became
  available, this specification specifically references the predecessor
  version -- RFC 2616.)  In addition, this specification defines the
  following terminology: « […]

So this term is already defined in RFC 7252, albeit implicitly only.)

> Q2:
> 
> Related to Q1, the text says:
> 
>   “can be used to combine representations of zero or more different media types into a single message”

Actually, it now says

  This memo defines application/multipart-core, an application-
  independent media-type that can be used to combine representations of
  zero or more different media types, each along with a CoAP Content-
  Format identifier, into a single representation, with minimal framing
  overhead.  This combined representation may then be carried in a
  single message, such as a CoAP [RFC7252] request or response body.

(Where “message” might be a whole block-wise transfer.)

Is that better?

> I assume it means ‘single message payload’? Because, you are not combing other parts (Codes, Tokens etc) of the message, are you?

To carry that in a single message, you would carry it in the message’s payload (or, theoretically, in an option, but we only do that in OSCORE so far).

> I think it would be more clear if you said ‘combine multiple representations, each using the same or different media types, into a single message payload’.

The new text above more clearly separates the representation combining from the ability to send it in a single message then.

> 
> ---
> 
> Q3: The draft uses both ‘media type’ and ‘media-type’.

Yeah, that’s a bit funny.
I actually believe we should have used media-type throughout the history of the term, as this is an artificial construct that only loosely relates to media types (types of media).
But maybe this draft is not the place to correct this mistake which has been in place since 1996.

> RFC 7252 uses ‘media type’ (there is one instance of ‘media-type’, though),

That one instance is in the construction “media-type ranges”, which is the normal way to compose composites in English.

> so maybe align with that?

I’m sure the RFC editor will force us to do that anyway, but I have made the change to “media type” (with a space) in the editors’ copy in case we get to submit another version.  Waiting for CI to show this at
https://core-wg.github.io/multipart-ct/draft-ietf-core-multipart-ct.html

(And the RFC editor will make us expand abbreviations and all that, too.)

Grüße, Carsten