Internet DRAFT - draft-bormann-apparea-bpack
draft-bormann-apparea-bpack
APP area C. Bormann
Internet-Draft Universitaet Bremen TZI
Intended status: Informational February 25, 2013
Expires: August 29, 2013
The BinaryPack1pre2 JSON-like representation format
draft-bormann-apparea-bpack-01
Abstract
JSON (RFC 4627) is an extremely successful format for the
representation of structured information, supporting Boolean values,
numbers, strings, arrays, and tables. Recently, a number of
applications have started to look for binary representation formats
that solve a similar problem. In particular, constrained node
networks can benefit from such a binary representation format.
A very successful binary representation that is otherwise comparable
to JSON is MessagePack. Recently, a number of implementations have
modified or extended MessagePack such that it allows for
distinguishing UTF-8 strings from binary data. Further discussion on
the MessagePack repository has resulted in proposals how to integrate
such an addition back into the MessagePack community.
This draft, as an independent effort, documents one such format,
tentatively calling it BinaryPack1pre2 while the MessagePack
extension proposals make their way through the MessagePack community.
The current version -01 of this document is a snapshot that
demonstrates a general direction. The details may change in future
versions based on the development of the MessagePack specification.
Status of This Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
Bormann Expires August 29, 2013 [Page 1]
Internet-Draft binarypack1pre2 February 2013
This Internet-Draft will expire on August 29, 2013.
Copyright Notice
Copyright (c) 2013 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
(http://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.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1. Objectives . . . . . . . . . . . . . . . . . . . . . . . 3
1.2. Terminology . . . . . . . . . . . . . . . . . . . . . . . 4
1.3. Notation . . . . . . . . . . . . . . . . . . . . . . . . 4
2. The BinaryPack1pre2 Representation Format . . . . . . . . . . 5
2.1. Data Types . . . . . . . . . . . . . . . . . . . . . . . 5
2.2. Integers . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3. Floating Point Values . . . . . . . . . . . . . . . . . . 6
2.4. Special Values . . . . . . . . . . . . . . . . . . . . . 6
2.5. Binary: Opaque Byte Strings . . . . . . . . . . . . . . . 7
2.6. UTF-8 Strings . . . . . . . . . . . . . . . . . . . . . . 7
2.7. Arrays . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.8. Tables . . . . . . . . . . . . . . . . . . . . . . . . . 8
3. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1. JSON roundtripping . . . . . . . . . . . . . . . . . . . 9
4. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 9
5. Security Considerations . . . . . . . . . . . . . . . . . . . 9
6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 9
7. References . . . . . . . . . . . . . . . . . . . . . . . . . 10
7.1. Normative References . . . . . . . . . . . . . . . . . . 10
7.2. Informative References . . . . . . . . . . . . . . . . . 10
Appendix A. Unicode Considerations . . . . . . . . . . . . . . . 11
Appendix B. Potential future work . . . . . . . . . . . . . . . 12
B.1. Reserved Code Points . . . . . . . . . . . . . . . . . . 12
B.2. 16-bit floating point . . . . . . . . . . . . . . . . . . 12
B.3. DateTime . . . . . . . . . . . . . . . . . . . . . . . . 12
B.4. Prefixing extensions . . . . . . . . . . . . . . . . . . 13
B.5. Extension Points . . . . . . . . . . . . . . . . . . . . 13
Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 13
Bormann Expires August 29, 2013 [Page 2]
Internet-Draft binarypack1pre2 February 2013
1. Introduction
(To be written - for now please see the Abstract.)
A description of the MessagePack binary representation format can be
found in [msgpack]. A recent proposal for an update, still under
discussion, is in [msgpack-update].
One of the early proposals implementing separate types for byte
strings and UTF-8 strings was called BinaryPack. An implementation
of BinaryPack is available in [binarypack]. (An extension similar in
spirit, but different in details, was made for the [msgpack-js] and
[msgpack-js-browser] projects.)
1.1. Objectives
(TBD, but this is a rough first approach:)
The objectives of the present specification, roughly in decreasing
order of importance, are:
o Representing a reasonable set of basic data types and structures
using binary encoding. "Reasonable" here is largely influenced by
the capabilities of JSON, with the single addition of adding raw
byte strings. The structures supported are limited to trees; no
loops or lattice-style graphs.
o Being implementable in a very small amount of code, thus being
applicable to constrained nodes [I-D.ietf-lwig-terminology], even
of class 1. (Complexity goal.) As a corollary: Being close to
contemporary machine representations of data (e.g., not requiring
binary-to-decimal conversion).
o Being applicable to schema-less use. For schema-informed binary
encoding, a number of approaches are already available in the
IETF, including XDR [RFC4506]. (However, schema-informed use of
the present specification, such as for a marshaling scheme for an
RPC IDL, is not at all excluded. Any IDL for this is out of scope
for this specification.)
o Being reasonably compact. "Reasonable" here is bounded by JSON as
an upper bound in size, and by implementation complexity
maintaining a lower bound. The use of general compression schemes
violates both of the complexity goals.
o Being reasonably frugal in CPU usage. (The other complexity
goal.) This is relevant both for constrained nodes and for
potential usage in high-volume applications.
Bormann Expires August 29, 2013 [Page 3]
Internet-Draft binarypack1pre2 February 2013
o Supporting a reasonable level of round-tripping with JSON, as long
as the data represented are within the capabilities of JSON.
Defining a unidirectional mapping towards JSON for all types of
data.
1.2. Terminology
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in [RFC2119].
The term "byte" is used in its now customary sense as a synonym for
"octet".
All multi-byte integers in this protocol are interpreted in network
byte order.
Where arithmetic is used, this specification uses the notation
familiar from the programming language C, except that the operator
"**" stands for exponentiation.
1.3. Notation
This specification uses a trivial notation for code bytes and the
bitfields in them the meaning of which should be mostly obvious.
More formally speaking, the meaning of the notation is:
Potential values for the code bytes themselves are expressed by
templates that represent 8-bit most-significant-bit-first binary
numbers (without any special prefix), where 0 stands for 0, 1 for 1,
and variable segments in these code byte templates are indicated by
sequences of the same letter such as kkkkkkk or ssss, the length of
which indicates the length of the variable segment in bits.
In the notation of values derived from the code bytes, 0b is used as
a prefix for expressing binary numbers in most-significant-bit first
notation (akin to the use of 0x for most-significant-digit-first
hexadecimal numbers in the C programming language). Where the above-
mentioned sequences of letters are then referenced in such a binary
number in the text, the intention is that the value from these
bitfields in the actual code byte be inserted.
Example: The code byte template
101nssss
stands for a byte that starts (most-significant-bit-first) with the
bits 1, 0, and 1, and continues with five variable bits, the first of
Bormann Expires August 29, 2013 [Page 4]
Internet-Draft binarypack1pre2 February 2013
which is referenced as "n" and the next four are referenced as
"ssss". Based on this code byte template, a reference to
0b0ssss000
means a binary number composed from a zero bit, the four bits that
are in the "ssss" field (for 101nssss, the four least significant
bits) in the actual byte encountered, kept in the same order, and
three more zero bits.
Also, 0xhh stands for the hexadecimal value hh, and 1B, 2B, 4B, 8B,
nB stand for 1, 2, 4, 8, or n bytes of data following; (1B) etc.
stand for the numerical value of these bytes as an integer
interpreted in network byte order; nD stands for n data objects, each
in turn in BinaryPack1pre2 representation format.
2. The BinaryPack1pre2 Representation Format
2.1. Data Types
The BinaryPack1pre2 representation format is able to represent the
following data types:
o Integers (represented in signed and unsigned forms)
o Floating point values (in IEEE 754 32-bit and 64-bit forms)
o special values nil, false, true
o opaque ("raw") byte strings, or "binary strings"
o UTF-8 strings
o arrays, which can contain any combination of data types
o tables (often called maps, hashes, dictionaries; objects in JSON),
which contain pairs, key and value, which may in turn be of any
data type
This list is mostly faithful to JSON [RFC4627], which however does
not distinguish integer from floating point number types. Based on
recent discussions on the use of binary representation formats, the
present specification distinguishes UTF-8 strings from opaque binary
strings. (Interestingly, such a separation was already done in the
binaryjs implementation of a "95 % MessagePack" format [binarypack],
so the author of the present specification started out by just lazily
copying that; more recent input taken from the msgpack developers
[msgpack-update] is the technical basis for the current proposal.)
Bormann Expires August 29, 2013 [Page 5]
Internet-Draft binarypack1pre2 February 2013
2.2. Integers
BinaryPack1pre2 provides a number of representations for integer
values, assuming that these occur often. The encoder is free to
choose any of these representations that is able to represent the
desired value.
+----------+--------------+----------------------------+
| Bits | Value | Description |
+----------+--------------+----------------------------+
| 0nnnnnnn | 0bnnnnnnn | Positive Integer (0..127) |
| | | |
| 111nnnnn | 0bnnnnn - 32 | Negative Integer (-32..-1) |
| | | |
| 0xcc 1B | 1B as uint | Unsigned Integer |
| | | |
| 0xcd 2B | 2B as uint | Unsigned Integer |
| | | |
| 0xce 4B | 4B as uint | Unsigned Integer |
| | | |
| 0xcf 8B | 8B as uint | Unsigned Integer |
| | | |
| 0xd0 1B | 1B as sint | Signed Integer |
| | | |
| 0xd1 2B | 2B as sint | Signed Integer |
| | | |
| 0xd2 4B | 4B as sint | Signed Integer |
| | | |
| 0xd3 8B | 8B as sint | Signed Integer |
+----------+--------------+----------------------------+
2.3. Floating Point Values
BinaryPack1pre2 provides 32-bit and 64-bit IEEE 754 values. (See
also Appendix B.2.)
+---------+-----------------------+-------------+
| Bits | Value | Description |
+---------+-----------------------+-------------+
| 0xca 4B | 4B as 32-bit IEEE 754 | Float |
| | | |
| 0xcb 8B | 8B as 64-bit IEEE 754 | Double |
+---------+-----------------------+-------------+
2.4. Special Values
Bormann Expires August 29, 2013 [Page 6]
Internet-Draft binarypack1pre2 February 2013
Similar to the special literals "false null true" in JSON,
BinaryPack1pre2 provides three special values:
+------+-------+---------------+
| Bits | Value | Description |
+------+-------+---------------+
| 0xc0 | nil | null, nothing |
| | | |
| 0xc2 | false | Boolean false |
| | | |
| 0xc3 | true | Boolean true |
+------+-------+---------------+
2.5. Binary: Opaque Byte Strings
(Note that the specific codepoint allocations in this section are
very much up for discussion. It can also be argued that we should be
spending some of the remaining reserved codepoints for short byte
strings.)
+------------+----------+----------------------------------+
| Bits | Value | Description |
+------------+----------+----------------------------------+
| 0xd5 1B nB | n = (1B) | byte string (0..(2**8-1) bytes) |
| | | |
| 0xd6 2B nB | n = (2B) | byte string (0..(2**16-1) bytes) |
| | | |
| 0xd7 4B nB | n = (4B) | byte string (0..(2**32-1) bytes) |
+------------+----------+----------------------------------+
2.6. UTF-8 Strings
+-------------+-------------+-----------------------------------+
| Bits | Value | Description |
+-------------+-------------+-----------------------------------+
| 101nnnnn nB | n = 0bnnnnn | Short UTF-8 string (0..31 bytes) |
| | | |
| 0xd9 1B nB | n = (1B) | UTF-8 string (0..(2**8-1) bytes) |
| | | |
| 0xda 2B nB | n = (2B) | UTF-8 string (0..(2**16-1) bytes) |
| | | |
| 0xdb 4B nB | n = (4B) | UTF-8 string (0..(2**32-1) bytes) |
+-------------+-------------+-----------------------------------+
Bormann Expires August 29, 2013 [Page 7]
Internet-Draft binarypack1pre2 February 2013
The strings transported MUST be UTF-8 strings [RFC3629]. (The
general assumption is that these UTF-8 strings are in Network Unicode
form [RFC5198], see Appendix A for some more discussion.)
2.7. Arrays
+-------------+------------+------------------------------------+
| Bits | Value | Description |
+-------------+------------+------------------------------------+
| 1001nnnn nD | n = 0bnnnn | Short array (0..15 data elements) |
| | | |
| 0xdc 2B nD | n = (2B) | array (0..(2**16-1) data elements) |
| | | |
| 0xdd 4B nD | n = (4B) | array (0..(2**32-1) data elements) |
+-------------+------------+------------------------------------+
2.8. Tables
+-------------+----------------+---------------------------------+
| Bits | Value | Description |
+-------------+----------------+---------------------------------+
| 1000nnnn nD | n = 2 * 0bnnnn | Short table (0..15 data pairs) |
| | | |
| 0xde 2B nD | n = 2 * (2B) | table (0..(2**16-1) data pairs) |
| | | |
| 0xdf 4B nD | n = 2 * (4B) | table (0..(2**32-1) data pairs) |
+-------------+----------------+---------------------------------+
The sequence of n elements is a sequence of pairs of data objects,
each pair represented as one data object representing the key
followed by the data object representing its associated value.
3. Discussion
This draft tries to be faithful to the successful MessagePack
[msgpack] format, including an recent extension proposal that enables
the distinction between opaque binary byte strings and UTF-8 byte
strings [msgpack-update].
Little analysis has been made whether a slightly different bit
allocation (e.g., using up fewer of the code combination for single-
byte integers) would be advantageous. However, the gains from a
different allocation are likely to be limited except for pathological
cases. (The main benefit achievable may be to have more codepoints
reserved for future expansion.)
Bormann Expires August 29, 2013 [Page 8]
Internet-Draft binarypack1pre2 February 2013
A short floating point (e.g., based on the 16-bit IEEE 754 floating
point value) might be a useful additional representation format.
Adding decimal floating point values probably is not so useful,
except where high fidelity to JSON is desired.
Some additional data types might be useful for some protocols, e.g.
UUIDs [RFC4122], date/time. See also Appendix B. This would further
increase the distance from JSON that BinaryPack1pre2 creates by
distinguishing opaque and UTF-8 strings.
3.1. JSON roundtripping
BinaryPack1pre2 enables mostly lossless translation to JSON. JSON
[RFC4627]. JSON roundtripping, however, is not necessarily the
primary design goal of BinaryPack1pre2, but it is a consideration.
In the translation of BinaryPack1pre2 to JSON, opaque byte strings
SHOULD be converted to equivalent base64url [RFC4648] UTF-8 strings.
Without a schema, it is hard to do the inverse consistently, as
base64url encoded byte strings are not specially marked up in JSON.
When translating BinaryPack1pre2 floating point values to JSON, the
usual problem of converting binary fractions to decimal
representation arises. In the other direction, the choice of a
floating point format may be hard to do properly. Clearly, any
number that can be transformed from a 64-bit IEEE 754 number to a
32-bit IEEE 754 number without loss of information can be represented
as the latter. Without schema information, it may be hard to find
other cases where the precision maybe is not that important.
4. IANA Considerations
Once this has received some discussion, we will understand how
exactly to register Internet media types for this.
The potential extension mechanisms discussed in Appendix B may need
an IANA registry.
5. Security Considerations
(Nothing but generic warnings about correctly implementing protocol
encoders/decoders so far; this section will certainly grow as
additional security considerations become known.)
6. Acknowledgements
MessagePack was developed and promoted by Sadayuki Furuhashi
("frsyuki").
Bormann Expires August 29, 2013 [Page 9]
Internet-Draft binarypack1pre2 February 2013
BinaryPack is a minor derivation of MessagePack that was developed by
Eric Zhang for the binaryjs project. A similar, but different
extension was made by Tim Caswell for his [msgpack-js] and
[msgpack-js-browser] projects.
The author of the present specification deserves absolutely no
credits whatsoever for any of this.
7. References
7.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC3629] Yergeau, F., "UTF-8, a transformation format of ISO
10646", STD 63, RFC 3629, November 2003.
[RFC4648] Josefsson, S., "The Base16, Base32, and Base64 Data
Encodings", RFC 4648, October 2006.
[RFC5198] Klensin, J. and M. Padlipsky, "Unicode Format for Network
Interchange", RFC 5198, March 2008.
[RFC5905] Mills, D., Martin, J., Burbank, J., and W. Kasch, "Network
Time Protocol Version 4: Protocol and Algorithms
Specification", RFC 5905, June 2010.
7.2. Informative References
[I-D.ietf-lwig-terminology]
Bormann, C. and M. Ersue, "Terminology for Constrained
Node Networks", draft-ietf-lwig-terminology-00 (work in
progress), February 2013.
[N4246R2] Lunde, K., "Stabilizing CJK Compatibility Ideographs
through the use of Standardized Variants", ISO/IEC JTC1/
SC2/WG2 N4246R2, March 2012, <ftp://std.dkuug.dk/JTC1/sc2/
wg2/docs/n4246.pdf>.
[RFC3339] Klyne, G., Ed. and C. Newman, "Date and Time on the
Internet: Timestamps", RFC 3339, July 2002.
[RFC4122] Leach, P., Mealling, M., and R. Salz, "A Universally
Unique IDentifier (UUID) URN Namespace", RFC 4122, July
2005.
Bormann Expires August 29, 2013 [Page 10]
Internet-Draft binarypack1pre2 February 2013
[RFC4506] Eisler, M., "XDR: External Data Representation Standard",
STD 67, RFC 4506, May 2006.
[RFC4627] Crockford, D., "The application/json Media Type for
JavaScript Object Notation (JSON)", RFC 4627, July 2006.
[binarypack]
Zhang, E., "BinaryPack for Javascript browsers", 2012,
<https://github.com/binaryjs/js-binarypack>.
[msgpack-js-browser]
Caswell, T., "msgpack for the browser", 2012, <https://
github.com/creationix/msgpack-js-browser>.
[msgpack-js]
Caswell, T., "msgpack for node", 2012, <https://github.com
/creationix/msgpack-js>.
[msgpack-update]
Furuhashi, S., "msgpack-update-proposal1.md", February
2012, <https://gist.github.com/frsyuki/5022569>.
[msgpack] Ohta, K. and S. Colebourne, "MessagePack format
specification", 2011, <http://wiki.msgpack.org/display/
MSGPACK/Format+specification>.
Appendix A. Unicode Considerations
(TBD. Some initial guidelines at [msgpack-update]. This section
should make clear that:)
o At the BinaryPack1pre2 encoding/decoding layer, implementations
are never concerned about Unicode normalization.
o Internet usage of Unicode is governed by [RFC5198]. The present
specification will not try to second-guess the evolution of this
standards-track document.
o [RFC5198] states that >>Before transmission, all character
sequences SHOULD be normalized according to Unicode normalization
form "NFC"<<. There may be some need to interpret this "SHOULD"
in the context of the present specification, as follows.
o There is a very strong expectation that applications making use of
BinaryPack1pre2 will lean towards using Unicode in NFC form, as
opposed to NFD. In other words, receivers may expect data in the
maximally composed form, as opposed to decomposed form.
Bormann Expires August 29, 2013 [Page 11]
Internet-Draft binarypack1pre2 February 2013
o The Normalization component of NFC may create problems in some
applications (e.g., see [N4246R2]). Before this is repaired in
some future version of Unicode, there is no expectation that all
applications generating BinaryPack1pre2 always perform the
canonical normalization where information loss would result.
o There is a strong expectation that BinaryPack1pre2 receivers be
resilient to the small variations in Unicode usage discussed here.
Appendix B. Potential future work
Two data types have been discussed for addition to BinaryPack1pre2.
B.1. Reserved Code Points
As of today, the following code points are reserved and could be used
for further extension, if required:
0xc1, 0xc4..0xc9, 0xd4, 0xd8
B.2. 16-bit floating point
16-bit floating points have become popular recently. BinaryPack1pre2
could enable the efficient transport of small floating point numbers
by adding a Half-precision floating point representation:
+---------+-----------------------+-------------+
| Bits | Value | Description |
+---------+-----------------------+-------------+
| 0xc9 2B | 2B as 16-bit IEEE 754 | Half |
| | | |
| 0xca 4B | 4B as 32-bit IEEE 754 | Float |
| | | |
| 0xcb 8B | 8B as 64-bit IEEE 754 | Double |
+---------+-----------------------+-------------+
B.3. DateTime
Many applications need the transport of Date/Time information. Some
need micro- or nanosecond resolution, some are more concerned about
significant range.
In the IETF, both NTP timestamps [RFC5905] and ISO8601 dates
[RFC3339] are popular. The former probably require short and long
versions to accommodate the different requirements in precision and
range. As a start, a 32.32 and a 64.64 NTP timestamp could be
defined. ISO8601 dates would need a length indicator and could
Bormann Expires August 29, 2013 [Page 12]
Internet-Draft binarypack1pre2 February 2013
therefore look close to the string8 form in BinaryPack1pre2. It is
worth limiting the set of choices based on some more input on what is
actually required.
B.4. Prefixing extensions
As the small number of remaining code points could be used up
quickly, some additions might preferably be expressed by a prefixing
scheme. E.g., if 0xc1 is picked for prefixing, the format
0xc1 0xnn 0xd5 0x08 ...
could be used for designating an 8-byte binary string (0xd5 0x08 ...)
as e.g. a date/time in 32.32 NTP timestamp format; the same value
for 0xnn could also be followed by a 16-byte binary string for a full
64.64 NTP timestamp and maybe even followed by an UTF-8 string for
GeneralizedTime _or_ an ISO8601 time, depending on which of these
formats are desirable. Implementations unaware of the semantics for
a specific value of 0xnn could still process the information as a
binary or UTF-8 string.
The number of extensions defined this way should be kept very small,
not only to preserve coding efficiency by making do with the single-
byte discriminator. The values for 0xnn would then be maintained in
an IANA registry, with a suitably careful allocation policy. This
needs further discussion.
B.5. Extension Points
More generally, evolution of a format always raises considerations
about compatibility. There are two directions of compatibility: -
Old data/old senders to new receivers (forward compatibility) and -
new data/new senders to old receivers (backward compatibility).
Further extension of the msgpack format currently always loses
backward compatibility, as there is no way for an older
implementation to find out the length that is consumed by a construct
that uses a new codepoint. In addition to a prefixing mechanism, the
BinaryPack1pre2 format could include deliberate extension points that
would at least allow an old receiver to decode future versions of the
BinaryPack1pre2 format without losing synchronization in the byte
stream, while possibly having to treat some of the information as
opaque.
Author's Address
Bormann Expires August 29, 2013 [Page 13]
Internet-Draft binarypack1pre2 February 2013
Carsten Bormann
Universitaet Bremen TZI
Postfach 330440
Bremen D-28359
Germany
Phone: +49-421-218-63921
Email: cabo@tzi.org
Bormann Expires August 29, 2013 [Page 14]
