RFC 8710: Multipart Content-Format for the Constrained Application Protocol (CoAP)

Status of This Memo

This is an Internet Standards Track document.¶

This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 7841.¶

Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at https://www.rfc-editor.org/info/rfc8710.¶

Copyright Notice

Copyright (c) 2020 IETF Trust and the persons identified as the document authors. All rights reserved.¶

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.¶

1. Introduction

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 with a CoAP Content-Format identifier [RFC7252]) into a single representation, with minimal framing overhead.¶

This simple and efficient binary framing mechanism can be employed to create application-specific message bodies that build on multiple already existing media types.¶

As the name of the media type suggests, application/multipart-core was inspired by the multipart media types initially defined in the original set of MIME specifications [RFC2046] and later. However, while those needed to focus on the syntactic aspects of integrating multiple representations into one email, transfer protocols providing full data transparency such as CoAP as well as readily available encoding formats such as the Concise Binary Object Representation (CBOR) [RFC7049] shift the focus towards the intended use of the combined representations. In this respect, the basic intent of the application/multipart-core media type is like that of multipart/mixed (Section 5.1.3 of [RFC2046]); however, the semantics are relaxed to allow for both ordered and unordered collections of media types.¶

The detailed semantics of the representations are refined by the context established by the application in the accompanying request parameters, e.g., the resource URI and any further options (header fields), but three usage scenarios are envisioned:¶

In one case, the individual representations in an application/multipart-core message body occur in a sequence, which may be employed by an application where such a sequence is natural, e.g., for a number of audio snippets in various formats to be played out in that sequence or search results returned in order of relevance.¶

In another case, an application may be more interested in a bag of representations (which are distinguished by their Content-Format identifiers), such as an audio snippet and a text representation accompanying it. In such a case, the sequence in which these occur may not be relevant to the application. This specification adds the option of substituting a null value for the representation of an optional part, which indicates that the part is not present.¶

A third common situation only has a single representation in the sequence, and the sender selects just one of a set of formats possible for this situation. This kind of union "type" of formats may also make the presence of the actual representation optional, the omission of which leads to a zero-length array.¶

Where these rules are not sufficient, an application might still use the general format defined here but register a new media type and an associated Content-Format identifier to associate the representation with these more specific semantics instead of using the application/multipart-core media type.¶

Also, future specifications might want to define rough equivalents for other multipart media types with specific semantics not covered by the present specification, such as multipart/alternative (Section 5.1.4 of [RFC2046]), where several alternative representations are provided in the message body, but only one of those is to be selected by the recipient for its use (this is less likely to be useful in a constrained environment that has facilities for pre-flight discovery).¶

2. Multipart Content-Format Encoding

A representation of media type application/multipart-core contains a collection of zero or more representations, each along with their respective Content-Format.¶

The collection is encoded as a CBOR [RFC7049] array with an even number of elements. Counting from zero, the odd-numbered elements are a byte string containing a representation or the value "null" (if an optional part is indicated as not given). The (even-numbered) element preceding each of these is an unsigned integer specifying the Content-Format ID of the representation following it.¶

For example, a collection containing two representations, one with Content-Format ID 42 and one with Content-Format ID 0, looks like this in CBOR diagnostic notation:¶

    [42, h'0123456789abcdef', 0, h'3031323334']

  

For illustration, the structure of an application/multipart-core representation can be described by the Concise Data Definition Language (CDDL) [RFC8610] specification in Figure 1:¶

    multipart-core = [* multipart-part]
multipart-part = (type: uint .size 2, part: bytes / null)

  

This format is intended as a strict specification: an implementation MUST stop processing the representation if there is a CBOR well-formedness error, a deviation from the structure defined above, or any residual data left after processing the CBOR data item. (This generally means the representation is not processed at all unless some streaming processing has already happened.)¶

3. Usage Example: Observing Resources

This section illustrates a less obvious example for using application/multipart-core: combining it with observing a resource [RFC7641] to handle pending results.¶

When a client registers to observe a resource for which no representation is available yet, the server may send one or more 2.05 (Content) notifications that indicate the lack of an actual representation. Later on, when one becomes available, the server will send the first actual 2.05 (Content) or 2.03 (Valid) notification. A diagram depicting possible resulting sequences of notifications, identified by their respective response code, is shown in Figure 2.¶

          __________       __________       __________
     |          |     |          |     |          |
---->|   2.05   |---->|  2.05 /  |---->|  4.xx /  |
     |  Pending |     |   2.03   |     |   5.xx   |
     |__________|     |__________|     |__________|
        ^   \ \          ^    \           ^
         \__/  \          \___/          /
                \_______________________/

  

The specification of the Observe option requires that all notifications carry the same Content-Format. The application/multipart-core media type can be used to provide that Content-Format, e.g., by carrying an empty list of representations in the case marked as "Pending" in Figure 2 and carrying a single representation specified as the target Content-Format in the case in the middle of the figure.¶

4. Implementation Hints

This section describes the serialization for readers that may be new to CBOR. It does not contain any new information.¶

An application/multipart-core representation carrying no representations is represented by an empty CBOR array, which is serialized as a single byte with the value 0x80.¶

An application/multipart-core representation carrying a single representation is represented by a two-element CBOR array, which is serialized as 0x82 followed by the two elements. The first element is an unsigned integer for the Content-Format value, which is represented as described in Table 1. The second element is the object as a byte string, which is represented as a length as described in Table 2 followed by the bytes of the object.¶

For example, a single text/plain object (Content-Format 0) of value "Hello World" (11 characters) would be serialized as follows:¶

    0x82 0x00 0x4b H e l l o 0x20 W o r l d

  

In effect, the serialization for a single object is done by prefixing the object with information that there is one object (here: 0x82), information about its Content-Format (here: 0x00), and information regarding its length (here: 0x4b).¶

For more than one representation included in an application/multipart-core representation, the head of the CBOR array is adjusted (0x84 for two representations, 0x86 for three, etc.), and the sequences of Content-Format and embedded representations follow.¶

For instance, the example from Section 2 would be serialized as follows:¶

    0x84 (*) 0x182A 0x48 0x0123456789ABCDEF (+) 0x00 0x45 0x3031323334

  

where (*) marks the start of the information about the first representation (Content-Format 42, byte string length 8), and (+) marks the start of the second representation (Content-Format 0, byte string length 5).¶

5. IANA Considerations

6. Security Considerations

The security considerations of [RFC7049] apply. In particular, resource exhaustion attacks may employ large values for the byte string size fields or employ deeply nested structures of recursively embedded application/multipart-core representations.¶

Acknowledgements

Most of the text in this document is from earlier contributions by two of the authors, Thomas Fossati and Klaus Hartke. This earlier work was reorganized in this document based on the requirements in [EST-COAPS] and discussions with Michael Richardson, Panos Kampanis, and Peter van der Stok.¶

Authors' Addresses