idnits 2.17.1 draft-bormann-coap-misc-23.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack a Security Considerations section. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 744 has weird spacing: '... code mid...' -- The document date (March 12, 2013) is 4063 days in the past. Is this intentional? -- Found something which looks like a code comment -- if you have code sections in the document, please surround them with '' and '' lines. Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'RFCXXXX' is mentioned on line 1459, but not defined -- Looks like a reference, but probably isn't: '0' on line 1100 == Unused Reference: 'CoRE201' is defined on line 376, but no explicit reference was found in the text == Outdated reference: A later version (-18) exists of draft-ietf-core-coap-13 == Outdated reference: A later version (-16) exists of draft-ietf-core-observe-08 == Outdated reference: A later version (-26) exists of draft-ietf-httpbis-p1-messaging-22 == Outdated reference: A later version (-26) exists of draft-ietf-httpbis-p4-conditional-22 == Outdated reference: A later version (-26) exists of draft-ietf-httpbis-p6-cache-22 ** Obsolete normative reference: RFC 2616 (Obsoleted by RFC 7230, RFC 7231, RFC 7232, RFC 7233, RFC 7234, RFC 7235) ** Obsolete normative reference: RFC 5226 (Obsoleted by RFC 8126) Summary: 3 errors (**), 0 flaws (~~), 9 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 CoRE Working Group C. Bormann 3 Internet-Draft K. Hartke 4 Intended status: Informational Universitaet Bremen TZI 5 Expires: September 13, 2013 March 12, 2013 7 Miscellaneous additions to CoAP 8 draft-bormann-coap-misc-23 10 Abstract 12 This short I-D makes a number of partially interrelated proposals how 13 to solve certain problems in the CoRE WG's main protocol, the 14 Constrained Application Protocol (CoAP). The current version has 15 been resubmitted to keep information about these proposals available; 16 the proposals are not all fleshed out at this point in time. 18 Status of This Memo 20 This Internet-Draft is submitted in full conformance with the 21 provisions of BCP 78 and BCP 79. 23 Internet-Drafts are working documents of the Internet Engineering 24 Task Force (IETF). Note that other groups may also distribute 25 working documents as Internet-Drafts. The list of current Internet- 26 Drafts is at http://datatracker.ietf.org/drafts/current/. 28 Internet-Drafts are draft documents valid for a maximum of six months 29 and may be updated, replaced, or obsoleted by other documents at any 30 time. It is inappropriate to use Internet-Drafts as reference 31 material or to cite them other than as "work in progress." 33 This Internet-Draft will expire on September 13, 2013. 35 Copyright Notice 37 Copyright (c) 2013 IETF Trust and the persons identified as the 38 document authors. All rights reserved. 40 This document is subject to BCP 78 and the IETF Trust's Legal 41 Provisions Relating to IETF Documents 42 (http://trustee.ietf.org/license-info) in effect on the date of 43 publication of this document. Please review these documents 44 carefully, as they describe your rights and restrictions with respect 45 to this document. Code Components extracted from this document must 46 include Simplified BSD License text as described in Section 4.e of 47 the Trust Legal Provisions and are provided without warranty as 48 described in the Simplified BSD License. 50 Table of Contents 52 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 53 2. Observing Resources in CoAP . . . . . . . . . . . . . . . . . 4 54 3. The Base-Uri Option . . . . . . . . . . . . . . . . . . . . . 6 55 4. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 7 56 5. References . . . . . . . . . . . . . . . . . . . . . . . . . 7 57 5.1. Normative References . . . . . . . . . . . . . . . . . . 7 58 5.2. Informative References . . . . . . . . . . . . . . . . . 8 59 Appendix A. The Nursery (Things that still need to ripen a bit) 9 60 A.1. Envelope Options . . . . . . . . . . . . . . . . . . . . 9 61 A.2. Payload-Length Option . . . . . . . . . . . . . . . . . . 10 62 A.3. URI Authorities with Binary Adresses . . . . . . . . . . 10 63 A.4. Length-aware number encoding (o256) . . . . . . . . . . . 12 64 A.5. SMS encoding . . . . . . . . . . . . . . . . . . . . . . 14 65 A.5.1. ASCII-optimized SMS encoding . . . . . . . . . . . . 15 66 A.6. CONNECT . . . . . . . . . . . . . . . . . . . . . . . . . 17 67 A.6.1. Requesting a Tunnel with CONNECT . . . . . . . . . . 18 68 A.6.2. Using a CONNECT Tunnel . . . . . . . . . . . . . . . 18 69 A.6.3. Closing down a CONNECT Tunnel . . . . . . . . . . . . 19 70 Appendix B. The Museum (Things we did, but maybe not exactly 71 this way) . . . . . . . . . . . . . . . . . . . . . 19 72 B.1. Getting rid of artificial limitations . . . . . . . . . . 19 73 B.1.1. Beyond 270 bytes in a single option . . . . . . . . . 20 74 B.1.2. Beyond 15 options . . . . . . . . . . . . . . . . . . 21 75 B.1.3. Implementing the option delimiter for 15 or more 76 options . . . . . . . . . . . . . . . . . . . . . . . 23 77 B.1.4. Option Length encoding beyond 270 bytes . . . . . . . 24 78 B.2. Registered Option . . . . . . . . . . . . . . . . . . . . 27 79 B.2.1. A Separate Suboption Number Space . . . . . . . . . . 27 80 B.2.2. Opening Up the Option Number Space . . . . . . . . . 28 81 B.3. Enabling Protocol Evolution . . . . . . . . . . . . . . . 32 82 B.3.1. Potential new option number allocation . . . . . . . 33 83 B.4. Patience, Leisure, and Pledge . . . . . . . . . . . . . . 35 84 B.4.1. Patience . . . . . . . . . . . . . . . . . . . . . . 35 85 B.4.2. Leisure . . . . . . . . . . . . . . . . . . . . . . . 36 86 B.4.3. Pledge . . . . . . . . . . . . . . . . . . . . . . . 36 87 B.4.4. Option Formats . . . . . . . . . . . . . . . . . . . 37 88 Appendix C. The Cemetery (Things we won't do) . . . . . . . . . 37 89 C.1. Example envelope option: solving #230 . . . . . . . . . . 37 90 C.2. Example envelope option: proxy-elective options . . . . . 38 91 C.3. Stateful URI compression . . . . . . . . . . . . . . . . 39 92 Appendix D. Experimental Options . . . . . . . . . . . . . . . . 40 93 D.1. Options indicating absolute time . . . . . . . . . . . . 40 94 D.2. Representing Durations . . . . . . . . . . . . . . . . . 42 95 D.3. Rationale . . . . . . . . . . . . . . . . . . . . . . . . 43 96 D.4. Pseudo-Floating Point . . . . . . . . . . . . . . . . . . 43 97 D.5. A Duration Type for CoAP . . . . . . . . . . . . . . . . 44 98 D.6. CONTOUR (CoAP Non-trivial Option Useful Representation) . 51 99 D.6.1. Specification of the PlanB Encoding . . . . . . . . . 51 100 D.6.2. Optional Features . . . . . . . . . . . . . . . . . . 55 101 D.6.3. Discussion . . . . . . . . . . . . . . . . . . . . . 58 102 D.6.4. Examples . . . . . . . . . . . . . . . . . . . . . . 59 103 D.6.5. Acknowledgements . . . . . . . . . . . . . . . . . . 59 104 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 59 106 1. Introduction 108 The CoRE WG is tasked with standardizing an Application Protocol for 109 Constrained Networks/Nodes, CoAP [I-D.ietf-core-coap]. This protocol 110 is intended to provide RESTful [REST] services not unlike HTTP 111 [RFC2616], while reducing the complexity of implementation as well as 112 the size of packets exchanged in order to make these services useful 113 in a highly constrained network of themselves highly constrained 114 nodes. 116 This objective requires restraint in a number of sometimes 117 conflicting ways: 119 o reducing implementation complexity in order to minimize code size, 121 o reducing message sizes in order to minimize the number of 122 fragments needed for each message (in turn to maximize the 123 probability of delivery of the message), the amount of 124 transmission power needed and the loading of the limited-bandwidth 125 channel, 127 o reducing requirements on the environment such as stable storage, 128 good sources of randomness or user interaction capabilities. 130 This draft attempts to address a number of problems not yet 131 adequately solved in [I-D.ietf-core-coap]. The solutions proposed to 132 these problems are somewhat interrelated and are therefore presented 133 in one draft. As of the current version of the draft, the main body 134 is almost empty, since few significant problems remain with CoAP or 135 its satellite specifications. 137 The appendix contains the "CoAP cemetery" (Appendix C, possibly later 138 to move into its own draft), documenting roads that the WG decided 139 not to take, in order to spare readers from reinventing them in vain. 140 There is also a "CoAP museum" (Appendix B), which documents previous 141 forms of proposals part of which did make it into the main documents 142 in one form or another. Finally, the "CoAP nursery" (Appendix A) 143 contains half- to fully-baked proposals that might become interesting 144 as the basis for future extensions. 146 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 147 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 148 document are to be interpreted as described in [RFC2119]. 150 The term "byte" is used in its now customary sense as a synonym for 151 "octet". 153 2. Observing Resources in CoAP 155 (Co-Author for this section: Matthias Kovatsch) 157 There are two open issues related to -observe 158 [I-D.ietf-core-observe]: 160 o mixing freshness and observation lifetime, and 162 o non-cacheable resources. 164 To solve the first issue, we think that -observe should be clarified 165 as follows: 167 A server sends at least some notifications as confirmable messages. 168 Each confirmable notification is an opportunity for the server to 169 check if the client is still there. If the client acknowledges the 170 notification, it is assumed to be well and alive and still interested 171 in the resource. If it rejects the message with a reset message or 172 if it doesn't respond, it is assumed not longer to be interested and 173 is removed from the list of observers. So an observation 174 relationship can potentially go on forever, if the client 175 acknowledges each confirmable notification. If the server doesn't 176 send a notification for a while and wants to check if the client is 177 still there, it may send a confirmable notification with the current 178 resource state to check that. 180 So there is no mixing of freshness and lifetime going on. 182 The other issue is a bit less trivial to solve. The problem is that 183 normal CoAP and -observe actually have very different freshness 184 models: 186 Normally, when a client wants to know the current state of a 187 resource, it retrieves a representation, uses it and stores it in its 188 cache. Later, when it wants to know the current state again, it can 189 either use the stored representation provided that it's still fresh, 190 or retrieve a new representation, use it and store it in its cache. 192 If a server knows when the state of the resource will change the next 193 time, it can set the Max-Age of the representation to an accurate 194 time span. So the change of the resource state will coincide with 195 the expiration of the freshness of the representation stored in the 196 client's cache (ignoring network latency). 198 But if the resource changes its state unpredictably at any time, the 199 server can set the Max-Age only to an estimate. If the state then 200 actually changes before the freshness expires, the client wrongly 201 believes it has fresh information. Conversely, if the freshness 202 expires and the client wants to know the current state, the client 203 wrongly believes it has to make a new request although the 204 representation is actually still fresh (this is defused by ETag 205 validation). 207 -observe doesn't have these kinds of problems: the server does not 208 have to predict when the resource will change its state the next 209 time. It just sends a notification when it does. The new 210 representation invalidates the old representation stored in the 211 client's cache. So the client always has a fresh representation that 212 it can use when it wants to know the current resource state without 213 ever having to make a request. An explicit Max-Age is not needed for 214 determining freshness. 216 But -observe has a different set of problems: 218 The first problem is that the resource may change its state more 219 often than there is bandwidth available or the client can handle. 220 Thus, -observe cannot make any guarantee that a client will see every 221 state change. The solution is that -observe guarantees that the 222 client will eventually see the latest state change, and follows a 223 best effort approach to enable the client to see as many state 224 changes as possible. 226 The second problem is that, when a notification doesn't arrive for a 227 while, the client does not know if the resource did not change its 228 state or if the server lost its state and forgot that the client is 229 interested in the resource. We propose the following solution: With 230 each notification that the server sends, it makes a promise to send 231 another notification, and that it will send this next notification at 232 latest after a certain time span. This time span is included with 233 each notification. So when no notification arrives for a while and 234 the time span has not expired yet, the client assumes that the 235 resource did not change its state. If the time span has expired, no 236 notification has arrived and the client wants to know the current 237 state of the resource, it has to make a new request. 239 The third problem is that, when an intermediary is observing a 240 resource and wants to create a response from a representation stored 241 in its cache, it needs to specify a Max-Age. But the intermediary 242 cannot predict when it will receive the next notification, because 243 the next notification can arrive at any time. Unlike the origin 244 server, it also doesn't have the application-specific knowledge that 245 the origin server has. We propose the following solution: With each 246 notification a server sends, it includes a value that an intermediary 247 should use to calculate the Max-Age. 249 To summarize: 251 o A notification doesn't have a Max-Age; it's fresh until the next 252 notification arrives. A notification is the promise for another 253 notification that will arrive at latest after Next-Notification- 254 At-Latest. This value is included with every notification. The 255 promise includes that the server attempts to transmit a 256 notification to the client for the promised time span, even if the 257 client does not seem to respond, e.g., due to a temporary network 258 outage. 260 o A notification also contains another value, called Max-Age-Hint. 261 This value is used by a cache to calculate a Max-Age for the 262 representation if needed. In a cache, the Max-Age-Hint of a 263 representation is counted down like Max-Age. When it reaches 264 zero, however, the representation can be still used to satisfy 265 requests, but is non-cacheable (i.e., Max-Age is 0). The Max-Age- 266 Hint must be less than or equal to Next-Notification-At-Latest. 268 We see two possible ways to encode Next-Notification-At-Latest and 269 Max-Age-Hint in a message: 271 o The first way is to require the values of Next-Notification-At- 272 Latest and Max-Age-Hint to be the same, although they are 273 conceptually unrelated. Then, a single option in the message can 274 be used to hold both values. 276 o The second way is to include two options, one for Next- 277 Notification-At-Latest and one for Max-Age-Hint. Since Next- 278 Notification-At-Latest is less than or equal to Max-Age-Hint, the 279 first option should indicates Max-Age-Hint, and the second option 280 Next-Notification-At-Latest minus Max-Age-Hint with a default 281 value of 0. 283 3. The Base-Uri Option 285 A proxy that forwards a response with embedded URIs may need to 286 indicate a base URI relative to which the embedded URIs need to be 287 interpreted that is different from the original request URI. E.g., 288 when the proxy forwarded the request to a multicast address, it may 289 need to indicate which specific server sent the response. A similar 290 requirement is the need to provide a request URI relative to which 291 the Location-* options can be interpreted. 293 The Base-Uri Option can be used in a response to provide this 294 information. It is structured like the Proxy-Uri option, but it is 295 elective and safe to forward (whether it is a cache-key is 296 irrelevant, as it is a response option only). 298 +--------+----------+-----------+ 299 | Number | Name | Reference | 300 +--------+----------+-----------+ 301 | TBD | Base-Uri | [RFCXXXX] | 302 +--------+----------+-----------+ 304 4. Acknowledgements 306 This work was partially funded by the Klaus Tschira Foundation and by 307 Intel Corporation. 309 Of course, much of the content of this draft is the result of 310 discussions with the [I-D.ietf-core-coap] authors. 312 Patience and Leisure were influenced by a mailing list discussion 313 with Esko Dijk, Kepeng Li, and Salvatore Loreto - thanks! 315 5. References 317 5.1. Normative References 319 [I-D.ietf-core-coap] 320 Shelby, Z., Hartke, K., Bormann, C., and B. Frank, 321 "Constrained Application Protocol (CoAP)", draft-ietf- 322 core-coap-13 (work in progress), December 2012. 324 [I-D.ietf-core-observe] 325 Hartke, K., "Observing Resources in CoAP", draft-ietf- 326 core-observe-08 (work in progress), February 2013. 328 [I-D.ietf-httpbis-p1-messaging] 329 Fielding, R. and J. Reschke, "Hypertext Transfer Protocol 330 (HTTP/1.1): Message Syntax and Routing", draft-ietf- 331 httpbis-p1-messaging-22 (work in progress), February 2013. 333 [I-D.ietf-httpbis-p4-conditional] 334 Fielding, R. and J. Reschke, "Hypertext Transfer Protocol 335 (HTTP/1.1): Conditional Requests", draft-ietf- 336 httpbis-p4-conditional-22 (work in progress), February 337 2013. 339 [I-D.ietf-httpbis-p6-cache] 340 Fielding, R., Nottingham, M., and J. Reschke, "Hypertext 341 Transfer Protocol (HTTP/1.1): Caching", draft-ietf- 342 httpbis-p6-cache-22 (work in progress), February 2013. 344 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 345 Requirement Levels", BCP 14, RFC 2119, March 1997. 347 [RFC2616] Fielding, R., Gettys, J., Mogul, J., Frystyk, H., 348 Masinter, L., Leach, P., and T. Berners-Lee, "Hypertext 349 Transfer Protocol -- HTTP/1.1", RFC 2616, June 1999. 351 [RFC3339] Klyne, G., Ed. and C. Newman, "Date and Time on the 352 Internet: Timestamps", RFC 3339, July 2002. 354 [RFC3629] Yergeau, F., "UTF-8, a transformation format of ISO 355 10646", STD 63, RFC 3629, November 2003. 357 [RFC4648] Josefsson, S., "The Base16, Base32, and Base64 Data 358 Encodings", RFC 4648, October 2006. 360 [RFC5198] Klensin, J. and M. Padlipsky, "Unicode Format for Network 361 Interchange", RFC 5198, March 2008. 363 [RFC5226] Narten, T. and H. Alvestrand, "Guidelines for Writing an 364 IANA Considerations Section in RFCs", BCP 26, RFC 5226, 365 May 2008. 367 [RFC5905] Mills, D., Martin, J., Burbank, J., and W. Kasch, "Network 368 Time Protocol Version 4: Protocol and Algorithms 369 Specification", RFC 5905, June 2010. 371 [RFC6256] Eddy, W. and E. Davies, "Using Self-Delimiting Numeric 372 Values in Protocols", RFC 6256, May 2011. 374 5.2. Informative References 376 [CoRE201] , "Clarify use of retransmission window for duplicate 377 detection", CoRE ticket #201, 2012, 378 . 380 [CoRE214] , "Adopt vendor-defined option into core-coap", CoRE 381 ticket #214, 2012, 382 . 384 [CoRE230] , "Multiple Location options need to be processed as a 385 unit", CoRE ticket #230, 2012, 386 . 388 [CoRE241] , "Proxy Safe & Cache Key indication for options", CoRE 389 ticket #241, 2012, 390 . 392 [REST] Fielding, R., "Architectural Styles and the Design of 393 Network-based Software Architectures", 2000. 395 [RFC1924] Elz, R., "A Compact Representation of IPv6 Addresses", RFC 396 1924, April 1996. 398 [RFC2817] Khare, R. and S. Lawrence, "Upgrading to TLS Within HTTP/ 399 1.1", RFC 2817, May 2000. 401 [RFC6648] Saint-Andre, P., Crocker, D., and M. Nottingham, 402 "Deprecating the "X-" Prefix and Similar Constructs in 403 Application Protocols", BCP 178, RFC 6648, June 2012. 405 Appendix A. The Nursery (Things that still need to ripen a bit) 407 A.1. Envelope Options 409 As of [I-D.ietf-core-coap], options can take one of four types, two 410 of which are mostly identical: 412 o uint: A non-negative integer which is represented in network byte 413 order using a variable number of bytes (see [I-D.ietf-core-coap] 414 Appendix A); 416 o string: a sequence of bytes that is nominally a Net-Unicode string 417 [RFC5198]; 419 o opaque: a sequence of bytes. 421 o empty (not explicitly identified as a fourth type in 422 [I-D.ietf-core-coap]). 424 It turns out some options would benefit from some internal structure. 425 Also, it may be a good idea to be able to bundle multiple options 426 into one, in order to ensure consistency for a set of elective 427 options that need to be processed all or nothing (i.e., the option 428 becomes critical as soon as another option out of the set is 429 processed, too). 431 In this section, we introduce a fifth CoAP option type: Envelope 432 options. 434 An envelope option is a sequence of bytes that looks and is 435 interpreted exactly like a CoAP sequence of options. Instead of an 436 option count or an end-of-option marker, the sequence of options is 437 terminated by the end of the envelope option. 439 The nested options (options inside the envelope option) may come from 440 the same number space as the top-level CoAP options, or the envelope 441 option may define its own number space - this choice needs to be 442 defined for each envelope option. 444 If the top-level number space is used, the envelope option typically 445 will restrict the set of options that actually can be used in the 446 envelope. In particular, it is unlikely that an envelope option will 447 allow itself inside the envelope (this would be a recursive option). 449 Envelope options are a general, but simple mechanism. Some of its 450 potential uses are illustrated by two examples in the cemetery: 451 Appendix C.1 and Appendix C.2. (Each of these examples has its own 452 merits and demerits, which led us to decide not to pursue either of 453 them right now, but this should be discussed separately from the 454 concept of Envelope options employed in the examples.) 456 A.2. Payload-Length Option 458 Not all transport mappings may provide an unambiguous length of the 459 CoAP message. For UDP, it may also be desirable to pack more than 460 one CoAP message into one UDP payload (aggregation); in that case, 461 for all but the last message there needs to be a way to delimit the 462 payload of that message. 464 This can be solved using a new option, the Payload-Length option. If 465 this option is present, the value of this option is an unsigned 466 integer giving the length of the payload of the message (note that 467 this integer can be zero for a zero-length payload, which can in turn 468 be represented by a zero-length option value). (In the UDP 469 aggregation case, what would have been in the payload of this message 470 after "payload-length" bytes is then actually one or more additional 471 messages.) 473 A.3. URI Authorities with Binary Adresses 474 One problem with the way URI authorities are represented in the URI 475 syntax is that the authority part can be very bulky if it encodes an 476 IPv6 address in ASCII. 478 Proposal: Provide an option "Uri-Authority-Binary" that can be an 479 even number of bytes between 2 and 18 except 12 or 14. 481 o If the number of bytes is 2, the destination IP address of the 482 packet transporting the CoAP message is implied. 484 o If the number of bytes is 4 or 6, the first four bytes of the 485 option value are an IPv4 address in binary. 487 o If the number of bytes is 8 or 10, the first eight bytes are the 488 lower 64 bits of an IPv6 address; the upper eight bytes are 489 implied from the destination address of the packet transporting 490 the CoAP message. 492 o If the number of bytes is 16 or 18, the first 16 bytes are an IPv6 493 address. 495 o If two more bytes remain, this is a port number (as always in 496 network byte order). 498 The resulting authority is (conceptually translated into ASCII and) 499 used in place of an Uri-Authority option, or inserted into a Proxy- 500 Uri. Examples: 502 +--------------+------------------+---------+-----------------------+ 503 | Proxy-Uri | Uri-Authority- | Uri- | URI | 504 | | Binary | Path | | 505 +--------------+------------------+---------+-----------------------+ 506 | (none) | (none) | (none) | "/" | 507 | | | | | 508 | (none) | (none) | 'temp' | "/temp" | 509 | | | | | 510 | (none) | 2 bytes: 61616 | 'temp' | "coap://[DA]:61616/te | 511 | | | | mp" | 512 | | | | | 513 | (none) | 16 bytes: | temp | "coap://[2000::1]/tem | 514 | | 2000::1 | | p" | 515 | | | | | 516 | 'http://' | 10 bytes: | (none) | "http://[DA::123:45]: | 517 | | ::123:45 + 616 | | 616" | 518 | | | | | 519 | 'http:///tem | 18 bytes: | (none) | "http://[2000::1]:616 | 520 | p' | 2000::1 + 616 | | /temp" | 521 +--------------+------------------+---------+-----------------------+ 523 A.4. Length-aware number encoding (o256) 525 The number encoding defined in Appendix A of [I-D.ietf-core-coap] has 526 one significant flaw: Every number has an infinite number of 527 representations, which can be derived by adding leading zero bytes. 528 This runs against the principle of minimizing unnecessary choice. 529 The resulting uncertainty in encoding ultimately leads to unnecessary 530 interoperability failures. (It also wastes a small fraction of the 531 encoding space, i.e., it wastes bytes.) 533 We could solve the first, but not the second, by outlawing leading 534 zeroes, but then we have to cope with error cases caused by illegal 535 values, another source of interoperability problems. 537 The number encoding "o256" defined in this section avoids this flaw. 538 The suggestion is not to replace CoAP's "uint" encoding wholesale 539 (CoAP is already too widely implemented for such a change), but to 540 consider this format for new options. 542 The basic requirements for such an encoding are: 544 o numbers are encoded as a sequence of zero or more bytes 546 o each number has exactly one encoding 547 o for a < b, encoding-size(a) <= encoding-size(b) -\u002D i.e., with 548 larger numbers, the encoding only gets larger, never smaller 549 again. 551 o within each encoding size (0 bytes, 1 byte, etc.), lexicographical 552 ordering of the bytes is the same as numeric ordering 554 Obviously, there is only one encoding that satisfies all these 555 requirements. As illustrated by Figure 1, this is unambiguously 556 derived by 558 1. enumerating all possible byte sequences, ordered by length and 559 within the same length in lexicographic ordering, and, 561 2. assigning sequential cardinals. 563 0x'' -> 0 564 0x'00' -> 1 565 0x'01' -> 2 566 0x'02' -> 3 567 ... 568 0x'fe' -> 255 569 0x'ff' -> 256 570 0x'0000' -> 257 571 0x'0001' -> 258 572 ... 573 0x'fefd' -> 65534 574 0x'fefe' -> 65535 575 0x'feff' -> 65536 576 ... 577 0x'ffff' -> 65792 578 0x'000000' -> 65793 579 0x'000001' -> 65794 581 Figure 1: Enumerating byte sequences by length and then lexicographic 582 order 584 This results in an exceedingly simple algorithm: each byte is 585 interpreted in the base-256 place-value system, but stands for a 586 number between 1 and 256 instead of 0 to 255. We therefore call this 587 encoding "o256" (one-to-256). 0 is always encoded in zero bytes; 1 588 to 256 is one byte, 257 (0x101) to 65792 (0x10100) is two bytes, 589 65793 (0x10101) to 16843008 (0x1010100) is three bytes, etc. 591 To further illustrate the algorithmic simplicity, pseudocode for 592 encoding and decoding is given in Figure 2 and Figure 3, respectively 593 (in the encoder, "prepend" stands for adding a byte at the _leading_ 594 edge, the requirement for which is a result of the network byte 595 order). Note that this differs only in a single subtraction/addition 596 (resp.) of one from the canonical algorithm for Appendix A uints. 598 while num > 0 599 num -= 1 600 prepend(num & 0xFF) 601 num >>= 8 602 end 604 Figure 2: o256 encoder (pseudocode) 606 num = 0 607 each_byte do |b| 608 num <<= 8 609 num += b + 1 610 end 612 Figure 3: o256 decoder (pseudocode) 614 On a more philosophical note, it can be observed that o256 solves the 615 inverse problem of Self-Delimiting Numeric Values (SDNV) [RFC6256]: 616 SDNV encodes variable-length numbers together with their length 617 (allowing decoding without knowing their length in advance, deriving 618 delimiting information from the number encoding). o256 encodes 619 variable-length numbers when there is a way to separately convey the 620 length (as in CoAP options), encoding (and later deriving) a small 621 part of the numeric value into/from that size information. 623 A.5. SMS encoding 625 For use in SMS applications, CoAP messages can be transferred using 626 SMS binary mode. However, there is operational experience showing 627 that some environments cannot successfully send a binary mode SMS. 629 For transferring SMS in character mode (7-bit characters), 630 base64-encoding [RFC4648] is an obvious choice. 3 bytes of message 631 (24 bits) turn into 4 characters, which cosume 28 bits. The overall 632 overhead is approximately 17 %; the maximum message size is 120 bytes 633 (160 SMS characters). 635 If a more compact encoding is desired, base85 encoding can be 636 employed (however, probably not the version defined in [RFC1924] 637 -\u002D instead, the version used in tools such as btoa and PDF 638 should be chosen). However, this requires division operations. 639 Also, the base85 character set includes several characters that 640 cannot be transferred in a single 7-bit unit in SMS and/or are known 641 to cause operational problems. A modified base85 character set can 642 be defined to solve the latter problem. 4 bytes of message (32 bits) 643 turn into 5 characters, which consume 35 bits. The overall overhead 644 is approximately 9.3 %; the resulting maximum message size is 128 645 bytes (160 SMS characters). 647 Base64 and base85 do not make use of the fact that much CoAP data 648 will be ASCII-based. Therefore, we define the following experimental 649 SMS encoding. 651 A.5.1. ASCII-optimized SMS encoding 653 Not all 128 theoretically possible SMS characters are operationally 654 free of problems. We therefore define: 656 Shunned code characters: @ sign, as it maps to 0x00 658 LF and CR signs (0x0A, 0x0D) 660 uppercase C cedilla (0x09), as it is often mistranslated in 661 gateways 663 ESC (0x1B), as it is used in certain character combinations only 665 Some ASCII characters cannot be transferred in the base SMS character 666 set, as their code positions are taken by non-ASCII characters. 667 These are simply encoded with their ASCII code positions, e.g., an 668 underscore becomes a section mark (even though underscore has a 669 different code position in the SMS character set). 671 Equivalently translated input bytes: $, @, [, \, ], ^, _, `, {, |, 672 }, ~, DEL 674 In other words, bytes 0x20 to 0x7F are encoded into the same code 675 positions in the 7-bit character set. 677 Out of the remaining code characters, the following SMS characters 678 are available for encoding: 680 Non-equivalently translated (NET) code characters: 0x01 to 0x08, (8 681 characters) 683 0x0B, 0x0C, (2 characters) 685 0x0E to 0x1A, (13 characters) 687 0x1C to 0x1F, (4 characters) 689 Of the 27 NET code characters, 18 are taken as prefix characters (see 690 below), and 8 are defined as directly translated characters: 692 Directly translated bytes: Equivalently translated input bytes are 693 represented as themselves 695 0x00 to 0x07 are represented as 0x01 to 0x08 697 This leaves 0x08 to 0x1F and 0x80 to 0xFF. Of these, the bytes 0x80 698 to 0x87 and 0xA0 to 0xFF are represented as the bytes 0x00 to 0x07 699 (represented by characters 0x01 to 0x08) and 0x20 to 0x7F, with a 700 prefix of 1 (see below). The characters 0x08 to 0x1F are represented 701 as the characters 0x28 to 0x3F with a prefix of 2 (see below). The 702 characters 0x88 to 0x9F are represented as the characters 0x48 to 703 0x5F with a prefix of 2 (see below). (Characters 0x01 to 0x08, 0x20 704 to 0x27, 0x40 to 0x47, and 0x60 to 0x7f with a prefix of 2 are 705 reserved for future extensions, which could be used for some 706 backreferencing or run-length compression.) 708 Bytes that do not need a prefix (directly translated bytes) are sent 709 as is. Any byte that does need a prefix (i.e., 1 or 2) is preceded 710 by a prefix character, which provides a prefix for this and the 711 following two bytes as follows: 713 +------+-----+---+------+-----+ 714 | 0x0B | 100 | | 0x15 | 200 | 715 +------+-----+---+------+-----+ 716 | 0x0C | 101 | | 0x16 | 201 | 717 | | | | | | 718 | 0x0E | 102 | | 0x17 | 202 | 719 | | | | | | 720 | 0x0F | 110 | | 0x18 | 210 | 721 | | | | | | 722 | 0x10 | 111 | | 0x19 | 211 | 723 | | | | | | 724 | 0x11 | 112 | | 0x1A | 212 | 725 | | | | | | 726 | 0x12 | 120 | | 0x1C | 220 | 727 | | | | | | 728 | 0x13 | 121 | | 0x1D | 221 | 729 | | | | | | 730 | 0x14 | 122 | | 0x1E | 222 | 731 +------+-----+---+------+-----+ 733 (This leaves one non-shunned character, 0x1F, for future extension.) 735 The coding overhead of this encoding for random bytes is similar to 736 Base85, without the need for a division/multiplication. For bytes 737 that are mostly ASCII characters, the overhead can easily become 738 negative. (Conversely, for bytes that are more likely to be non- 739 ASCII than in a random sequence of bytes, the overhead becomes 740 greater.) 742 So, for instance, for the CoAP message in Figure 4: 744 ver tt code mid 745 1 ack 2.05 17033 746 content_type 40 747 token sometok 748 3c 2f 3e 3b 74 69 74 6c 65 3d 22 47 65 6e 65 72 |;title="Gener| 749 61 6c 20 49 6e 66 6f 22 3b 63 74 3d 30 2c 3c 2f |al Info";ct=0,;if="clock"| 751 3b 72 74 3d 22 54 69 63 6b 73 22 3b 74 69 74 6c |;rt="Ticks";titl| 752 65 3d 22 49 6e 74 65 72 6e 61 6c 20 43 6c 6f 63 |e="Internal Cloc| 753 6b 22 3b 63 74 3d 30 2c 3c 2f 61 73 79 6e 63 3e |k";ct=0,| 754 3b 63 74 3d 30 |;ct=0 | 756 Figure 4: CoAP response message as captured and decoded 758 The 116 byte unencoded message is shown as ASCII characters in Figure 759 5 (\xDD stands for the byte with the hex digits DD): 761 bEB\x89\x11(\xA7sometok;title="General Info";ct=0, 762 ;if="clock";rt="Ticks";title="Internal Clock";ct=0,;ct=0 764 Figure 5: CoAP response message shown as unencoded characters 766 The equivalent SMS encoding is shown as equivalent-coded SMS 767 characters in Figure 6 (7 bits per character, \x12 is a 220 prefix 768 and \x0B is a 100 prefix, the rest is shown in equivalent encoding), 769 adding two characters of prefix overhead, for a total length of 118 770 7-bit characters or 104 (103.25 plus padding) bytes: 772 bEB\x12I1(\x0B'sometok;title="General Info";ct=0, 773 ;if="clock";rt="Ticks";title="Internal Clock";ct=0,;ct=0 775 Figure 6: CoAP response message shown as SMS-encoded characters 777 A.6. CONNECT 779 [RFC2817] defines the HTTP CONNECT method to establish a TCP tunnel 780 through a proxy so that end-to-end TLS connections can be made 781 through the proxy. Recently, a requirement for similar functionality 782 has been discussed for CoAP. This section defines a straw-man 783 CONNECT method and related methods and response codes for CoAP. 785 (IANA considerations for this section TBD.) 787 A.6.1. Requesting a Tunnel with CONNECT 789 CONNECT is allocated as a new method code in the "CoAP Method Codes" 790 registry. When a client makes a CONNECT request to an intermediary, 791 the intermediary evaluates the Uri-Host, Uri-Port, and/or the 792 authority part of the Proxy-Uri Options in a way that is defined by 793 the security policy of the intermediary. If the security policy 794 allows the allocation of a tunnel based on these parameters, the 795 method returns an empty payload and a response code of 2.30 Tunnel 796 Established. Other possible response codes include 4.03 Forbidden. 798 It may be the case that the intermediary itself can only reach the 799 requested origin server through another intermediary. In this case, 800 the first intermediary SHOULD make a CONNECT request of that next 801 intermediary, requesting a tunnel to the authority. A proxy MUST NOT 802 respond with any 2.xx status code unless it has either a direct or 803 tunnel connection established to the authority. 805 An origin server which receives a CONNECT request for itself MAY 806 respond with a 2.xx status code to indicate that a tunnel is 807 established to itself. 809 Code 2.30 "Tunnel Established" is allocated as a new response code in 810 the "CoAP Response Codes" registry. 812 A.6.2. Using a CONNECT Tunnel 814 Any successful (2.xx) response to a CONNECT request indicates that 815 the intermediary has established a tunnel to the requested host and 816 port. The tunnel is bound to the requesting end-point and the Token 817 supplied in the request (as always, the default Token is admissible). 818 The tunnel can be used by the client by making a DATAGRAM request. 820 DATAGRAM is allocated as a new method code in the "CoAP Method Codes" 821 registry. When a client makes a DATAGRAM request to an intermediary, 822 the intermediary looks up the tunnel bound to the client end-point 823 and Token supplied in the DATAGRAM request (no other Options are 824 permitted). If a tunnel is found and the intermediary's security 825 policy permits, the intermediary forwards the payload of the DATAGRAM 826 request as the UDP payload towards the host and port established for 827 the tunnel. No response is defined for this request (note that the 828 request can be given as a CON or NON request; for CON, there will be 829 an ACK on the message layer if the tunnel exists). 831 The security policy on the intermediary may restrict the allowable 832 payloads based on its security policy, possibly considering host and 833 port. An inadmissible payload SHOULD cause a 4.03 Forbidden response 834 with a diagnostic message as payload. 836 The UDP payload of any datagram received from the tunnel and admitted 837 by the security policy is forwarded to the client as the payload of a 838 2.31 "Datagram Received" response. The response does not carry any 839 Option except for Token, which identifies the tunnel towards the 840 client. 842 Code 2.31 "Datagram Received" is allocated as a new response code in 843 the "CoAP Response Codes" registry. 845 An origin server that has established a tunnel to itself processes 846 the CoAP payloads of related DATAGRAM requests as it would process an 847 incoming UDP payload, and forwards what would be outgoing UDP 848 payloads in 2.31 "Datagram Received" responses. 850 A.6.3. Closing down a CONNECT Tunnel 852 A 2.31 "Datagram Received" response may be replied to with a RST, 853 which closes down the tunnel. Similarly, the Token used in the 854 tunnel may be reused by the client for a different purpose, which 855 also closes down the tunnel. 857 Appendix B. The Museum (Things we did, but maybe not exactly this way) 859 B.1. Getting rid of artificial limitations 861 _Artificial limitations_ are limitations of a protocol or system that 862 are not rooted in limitations of actual capabilities, but in 863 arbitrary design decisions. Proper system design tries to avoid 864 artificial limitations, as these tend to cause complexity in systems 865 that need to work with these limitations. 867 E.g., the original UNIX filesystem had an artificial limitation of 868 the length of a path name component to 14 bytes. This led to a 869 cascade of workarounds in programs that manipulate file names: E.g., 870 systematically replacing a ".el" extension in a filename with a 871 ".elc" for the compiled file might exceed the limit, so all ".el" 872 files were suddenly limited to 13-byte filenames. 874 Note that, today, there still is a limitation in most file system 875 implementations, typically at 255. This just happens to be high 876 enough to rarely be of real-world concern; we will refer to this case 877 as a "painless" artificial limitation. 879 CoAP-08 had two highly recognizable artificial limitations in its 880 protocol encoding 882 o The number of options in a single message is limited to 15 max. 884 o The length of an option is limited to 270 max. 886 It has been argued that the latter limitation causes few problems, 887 just as the 255-byte path name component limitation in filenames 888 today causes few problems. Appendix B.1.1 provided a design to 889 extend this; as a precaution to future extensions of this kind, the 890 current encoding for length 270 (eight ones in the extension byte) 891 could be marked as reserved today. Since, Matthias Kovatsch has 892 proposed a simpler scheme that seems to gain favor in the WG, see 893 Appendix B.1.4. 895 The former limitation has been solved in CoAP-09. A historical 896 discussion of other approaches for going beyond 15 options is in 897 Appendix B.1.2. Appendix B.1.3 discusses implementation. 899 B.1.1. Beyond 270 bytes in a single option 901 The authors would argue that 270 as the maximum length of an option 902 is already beyond the "painless" threshold. 904 If that is not the consensus of the WG, the scheme can easily be 905 extended as in Figure 7: 907 for 15..269: 908 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 909 | Option Delta | 1 1 1 1 | Length - 15 | 910 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 911 | Option Value ... 912 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 914 for 270..65805: 915 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 916 | Option Delta | 1 1 1 1 | 1 1 1 1 1 1 1 1 | 917 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 918 | Length - 270 (in network byte order) | 919 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 920 | Option Value ... 921 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 923 Figure 7: Ridiculously Long Option Header 925 The infinite number of obvious variations on this scheme are left as 926 an exercise to the reader. 928 Again, as a precaution to future extensions, the current encoding for 929 length 270 (eight ones in the extension byte) could be marked as 930 reserved today. 932 B.1.2. Beyond 15 options 934 (This section keeps discussion that is no longer needed as we have 935 agreed to do what is documented in Appendix B.1.3). 937 The limit of 15 options is motivated by the fixed four-bit field "OC" 938 that is used for indicating the number of options in the fixed-length 939 CoAP header (Figure 8). 941 0 1 2 3 942 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 943 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 944 |Ver| T | OC | Code | Message ID | 945 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 946 | Options (if any) ... 947 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 948 | Payload (if any) ... 949 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 951 Figure 8: Four-byte fixed header in a CoAP Message 953 Note that there is another fixed four-bit field in CoAP: the option 954 length (Figure 9 - note that this figure is not to the same scale as 955 the previous figure): 957 0 1 2 3 4 5 6 7 958 +---+---+---+---+---+---+---+---+ 959 | Option Delta | Length | for 0..14 960 +---+---+---+---+---+---+---+---+ 961 | Option Value ... 962 +---+---+---+---+---+---+---+---+ 964 Figure 9: Short Option Header 966 Since 15 is inacceptable for a maximum option length, the all-ones 967 value (15) was taken out of the set of allowable values for the short 968 header, and a long header was introduced that allows the insertion of 969 an extension byte (Figure 10): 971 for 15..270: 972 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 973 | Option Delta | 1 1 1 1 | Length - 15 | 974 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 975 | Option Value ... 976 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 978 Figure 10: Long Option Header 980 We might want to use the same technique for the CoAP header as well. 981 There are two obvious places where the extension byte could be 982 placed: 984 1. right after the byte carrying the OC field, so the structure is 985 the same as for the option header; 987 2. right after the fixed-size CoAP header. 989 Both solutions lose the fixed-size-ness of the CoAP header. 991 Solution 1 has the disadvantage that the CoAP header is also changing 992 in structure: The extension byte is wedged between the first and the 993 second byte of the CoAP header. This is unfortunate, as the number 994 of options only comes into play when the option processing begins, so 995 it is more natural to use solution 2 (Figure 11): 997 0 1 2 3 998 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 999 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1000 |Ver| T | 15 | Code | Message ID | 1001 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1002 | OC - 15 | Options ... 1003 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1004 | Payload (if any) ... 1005 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1007 Figure 11: Extended header for CoAP Messages with 15+ options 1009 This would allow for up to 270 options in a CoAP message, which is 1010 very likely way beyond the "painless" threshold. 1012 B.1.2.1. Implementation considerations 1014 For a message decoder, this extension creates relatively little pain, 1015 as the number of options only becomes interesting when the encoding 1016 turns to the options part of the message, which is then simply lead 1017 in by the extension byte if the four-bit field is 15. 1019 For a message encoder, this extension is not so rosy. If the encoder 1020 is constructing the message serially, it may not know in advance 1021 whether the number of options will exceed 14. None of the following 1022 implementation strategies is particularly savory, but all of them do 1023 work: 1025 1. Encode the options serially under the assumption that the number 1026 of options will be 14 or less. When the 15th option needs to be 1027 encoded, abort the option encoding, and restart it from scratch 1028 one byte further to the left. 1030 2. Similar to 1, except that the bytes already encoded are all moved 1031 one byte to right, the extension byte is inserted, and the option 1032 encoding process is continued. 1034 3. The encoder always leaves space for the extension byte (at least 1035 if it can't prove the number will be less thatn 14). If the 1036 extension byte is not needed, an Option 0 with length 0 is 1037 encoded instead (i.e., one byte is wasted - this option is 1038 elective and will be ignored by the receiver). 1040 As a minimum, to enable strategy 3, the option 0 should be reserved 1041 at least for the case of length=0. 1043 B.1.2.2. What should we do now? 1045 As a minimum proposal for the next version of CoAP, the value 15 for 1046 OC should be marked as reserved today. 1048 B.1.2.3. Alternatives 1050 One alternative that has been discussed previously is to have an 1051 "Options" Option, which allows the carriage of multiple options in 1052 the belly of a single one. This could also be used to carry more 1053 than 15 options. However: 1055 o The conditional introduction of an Options option has 1056 implementation considerations that are likely to be more severe 1057 than the ones listed above; 1059 o since 270 bytes may not be enough for the encoding of _all_ 1060 options, the "Options" option would need to be repeatable. This 1061 creates many different ways to encode the same message, leading to 1062 combinatorial explosion in test cases for ensuring 1063 interoperability. 1065 B.1.2.4. Alternative: Going to a delimiter model 1067 Another alternative is to spend the additional byte not as an 1068 extended count, but as an option terminator. 1070 B.1.3. Implementing the option delimiter for 15 or more options 1071 Implementation note: As can be seen from the proof of concept code 1072 in Figure 12, the actual implementation cost for a decoder is 1073 around 4 lines of code (or about 8-10 machine code instructions). 1075 while numopt > 0 1076 nextbyte = ... get next byte 1078 if numopt == 15 # new 1079 break if nextbyte == 0xF0 # new 1080 else # new 1081 numopt -= 1 1082 end # new 1084 ... decode delta and length from nextbyte and handle them 1085 end 1087 Figure 12: Implementing the Option Terminator 1089 Similarly, creating the option terminator needs about four more lines 1090 (not marked "old" in the C code in Figure 13). 1092 b0 = 0x40 + (tt << 4); /* old */ 1093 buffer[0] = b0 + 15; /* guess first byte */ 1095 .... encode options .... /* old */ 1097 if (option_count >= 15 || first_fragment_already_shipped) 1098 buffer[pos++] = 0xF0; /* use delimiter */ 1099 else /* save a byte: */ 1100 buffer[0] = b0 + option_count; /* old: backpatch */ 1102 Figure 13: Creating the Option Terminator 1104 B.1.4. Option Length encoding beyond 270 bytes 1106 For option lengths beyond 270 bytes, we reserve the value 255 of an 1107 extension byte to mean "add 255, read another extension byte" Figure 1108 14. While this causes the length of the option header to grow 1109 linearly with the size of the option value, only 0.4 % of that size 1110 is used. With a focus on short options, this encoding is justified. 1112 for 15..269: 1113 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1114 | Option Delta | 1 1 1 1 | Length - 15 | 1115 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1116 | Option Value ... 1117 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1118 for 270..524: 1119 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1120 | Option Delta | 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1121 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1122 | Length - 270 | Option Value ... 1123 +---+---+---+---+---+---+---+---+ 1124 | 1125 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1127 for 525..779: 1128 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1129 | Option Delta | 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1130 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1131 | 1 1 1 1 1 1 1 1 | Length - 525 | 1132 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1133 | Option Value ... 1134 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1136 for 780..1034: 1137 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1138 | Option Delta | 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1139 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1140 | 1 1 1 1 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1141 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1142 | Length - 780 | Option Value ... 1143 +---+---+---+---+---+---+---+---+ 1144 | 1145 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1147 Figure 14: Options beyond 270 bytes 1149 Options that are longer than 1034 bytes MUST NOT be sent; an option 1150 that has 255 (all one bits) in the field called "Length - 780" MUST 1151 be rejected upon reception as an invalid option. 1153 In the process, the maximum length of all options that are currently 1154 set at 270 should now be set to a carefully chosen value. With the 1155 purely encoding-based limit gone, Uri-Proxy should now be restored to 1156 be a non-repeatable option. 1158 A first proposal for a new set of per-option length restrictions 1159 follows: 1161 +--------+---------------------+-----+------+--------+--------+ 1162 | number | name | min | max | type | repeat | 1163 +--------+---------------------+-----+------+--------+--------+ 1164 | 1 | content_type | 0 | 2 | uint | - | 1165 | | | | | | | 1166 | 2 | max_age | 0 | 4 | uint | - | 1167 | | | | | | | 1168 | 3 | proxy_uri | 1 | 1023 | string | - | 1169 | | | | | | | 1170 | 4 | etag | 1 | 8 | opaque | yes | 1171 | | | | | | | 1172 | 5 | uri_host | 1 | 255 | string | - | 1173 | | | | | | | 1174 | 6 | location_path | 0 | 255 | string | yes | 1175 | | | | | | | 1176 | 7 | uri_port | 0 | 2 | uint | - | 1177 | | | | | | | 1178 | 8 | location_query | 0 | 255 | string | yes | 1179 | | | | | | | 1180 | 9 | uri_path | 0 | 255 | string | yes | 1181 | | | | | | | 1182 | 10 | observe | 0 | 2 | uint | - | 1183 | | | | | | | 1184 | 11 | token | 1 | 8 | opaque | - | 1185 | | | | | | | 1186 | 12 | accept | 0 | 2 | uint | yes | 1187 | | | | | | | 1188 | 13 | if_match | 0 | 8 | opaque | yes | 1189 | | | | | | | 1190 | 14 | registered_elective | 1 | 1023 | opaque | yes | 1191 | | | | | | | 1192 | 15 | uri_query | 1 | 255 | string | yes | 1193 | | | | | | | 1194 | 17 | block2 | 0 | 3 | uint | - | 1195 | | | | | | | 1196 | 18 | size | 0 | 4 | uint | - | 1197 | | | | | | | 1198 | 19 | block1 | 0 | 3 | uint | - | 1199 | | | | | | | 1200 | 21 | if_none_match | 0 | 0 | empty | - | 1201 | | | | | | | 1202 | 25 | registered_critical | 1 | 1023 | opaque | yes | 1203 +--------+---------------------+-----+------+--------+--------+ 1205 (Option 14 with a length of 0 is a fencepost only.) 1207 B.2. Registered Option 1209 CoAP's option encoding is highly efficient, but works best with small 1210 option numbers that do not require much fenceposting. The CoAP 1211 Option Number Registry therefore has a relatively heavyweight 1212 registration requirement: "IETF Review" as described in [RFC5226]. 1214 However, there is also considerable benefit in a much looser registry 1215 policy, enabling a first-come-first-served policy for a relatively 1216 large option number space. 1218 Here, we discuss two solutions that enable such a registry. One is 1219 to define a separate mechanism for registered options, discussed in 1220 Appendix B.2.1. Alternatively, we could make it easier to use a 1221 larger main option number space, discussed in Appendix B.2.2. 1223 B.2.1. A Separate Suboption Number Space 1225 This alternative defines a separate space of suboption numbers, with 1226 an expert review [RFC5226] (or even first-come-first-served) 1227 registration policy. If expert review is selected for this registry, 1228 it would be with a relatively loose policy delegated to the expert. 1229 This draft proposes leaving the registered suboption numbers 0-127 to 1230 expert review with a policy that mainly focuses on the availability 1231 of a specification, and 128-16383 for first-come-first-served where 1232 essentially only a name is defined. 1234 The "registered" options are used in conjunction with this suboption 1235 number registry. They use two normal CoAP option numbers, one for 1236 options with elective semantics (Registered-Elective) and one for 1237 options with critical semantics (Registered-Critical). The suboption 1238 numbers are not separate, i.e. one registered suboption number might 1239 have some elective semantics and some other critical semantics (e.g., 1240 for the request and the response leg of an exchange). The option 1241 value starts with an SDNV [RFC6256] of the registered suboption 1242 number. (Note that there is no need for an implementation to 1243 understand SDNVs, it can treat the prefixes as opaque. One could 1244 consider the SDNVs as a suboption prefix allocation guideline for 1245 IANA as opposed to a number encoding.) 1247 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1248 |1 0 0 0 0 0 0 1|0 1 1 1 0 0 1 1| value... | 1249 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1250 \___SDNV of registered number___/ 1252 Figure 15: Example option value for registered option 1254 Note that a Registered Option cannot be empty, because there would be 1255 no space for the SDNV. Also, the empty option 14 is reserved for 1256 fenceposting ([I-D.ietf-core-coap], section 3.2). (Obviously, once a 1257 Registered-Elective Option is in use, there is never a need for a 1258 fence-post for option number 14.) 1260 The Registered-Elective and Registered-Critical Options are 1261 repeatable. 1263 +------+----------+---------------------+--------+--------+---------+ 1264 | No. | C/E | Name | Format | Length | Default | 1265 +------+----------+---------------------+--------+--------+---------+ 1266 | 14 | Elective | Registered-Elective | (see | 1-1023 | (none) | 1267 | | | | above) | B | | 1268 | | | | | | | 1269 | 25 | Critical | Registered-Critical | (see | 1-1023 | (none) | 1270 | | | | above) | B | | 1271 +------+----------+---------------------+--------+--------+---------+ 1273 This solves CoRE issue #214 [CoRE214]. (How many options we need 1274 will depend on the resolution of #241 [CoRE241].) 1276 B.2.2. Opening Up the Option Number Space 1278 The disadvantage of the registered-... options is that there is a 1279 significant syntactic difference between options making use of this 1280 space and the usual standard options. This creates a problem not 1281 unlike that decried in [RFC6648]. 1283 The alternative discussed in this section reduces the distance by 1284 opening up the main Option number space instead. 1286 There is still a significant incentive to use low-numbered Options. 1287 However, the proposal reduces the penalty for using a high-numbered 1288 Option to two or three bytes. More importantly, using a cluster of 1289 related high-numbered options only carries a total penalty of two or 1290 three bytes. 1292 The main reason high-numbered options are expensive to use and thus 1293 the total space is relatively limited is that the option delta 1294 mechanism only allows increasing the current option number by up to 1295 14 per one-byte fencepost. To use, e.g., Option number 1234 together 1296 with the usual set of low-numbered Options, one needs to insert 88 1297 fence-post bytes. This is prohibitive. 1299 Enabling first-come-first-served probably requires easily addressing 1300 a 16-bit option number space, with some potential increase later in 1301 the lifetime of the protocol (say, 10 to 15 years from now). 1303 To enable the use of large option numbers, one needs a way to advance 1304 the Option number in bigger steps than possible by the Option Delta. 1305 So we propose a new construct, the Long Jump construct, to move the 1306 Option number forward. 1308 B.2.2.1. Long Jump construct 1310 The following construct can occur in front of any Option: 1312 0 1 2 3 4 5 6 7 1313 +---+---+---+---+---+---+---+---+ 1314 | 1 1 1 1 | 0 0 0 1 | 0xf1 (Delta = 15) 1315 +---+---+---+---+---+---+---+---+ 1317 0 1 2 3 4 5 6 7 1318 +---+---+---+---+---+---+---+---+ 1319 | 1 1 1 1 | 0 0 1 0 | 0xf2 1320 +---+---+---+---+---+---+---+---+ 1321 | Long Jump Value | (Delta/8)-2 1322 +---+---+---+---+---+---+---+---+ 1324 0 1 2 3 4 5 6 7 1325 +---+---+---+---+---+---+---+---+ 1326 | 1 1 1 1 | 0 0 1 1 | 0xf3 1327 +---+---+---+---+---+---+---+---+ 1328 | Long Jump Value, MSB | 1329 +---+---+---+---+---+---+---+---+ (Delta/8)-258 1330 | Long Jump Value, LSB | 1331 +---+---+---+---+---+---+---+---+ 1333 Figure 16: Long Jump Format 1335 This construct is not by itself an Option. It can occur in front of 1336 any Option to increase the current Option number that then goes into 1337 its Option number calculation. The increase is done in multiples of 1338 eight. More specifically, the actual addition to the current Option 1339 number is computed as follows: 1341 Delta = ((Long Jump Value) + N) * 8 1343 where N is 2 for the one-byte version and N is 258 for the two-byte 1344 version. 1346 A Long Jump MUST be followed by an actual Option, i.e., it MUST NOT 1347 be followed by another Long Jump or an end-of-options indicator. A 1348 message violating this MUST be rejected as malformed. 1350 Long Jumps do NOT count as Options in the Option Count field of the 1351 header (i.e., they cannot by themselves end the Option sequence). 1353 B.2.2.2. Discussion 1355 Adding a mechanism at this late stage creates concerns of backwards 1356 compatibility. A message sender never needs to implement long-jumps 1357 unless it wants to make use of a high-numbered option. So this 1358 mechanism can be added once a high-numbered option is added. A 1359 message receiver, though, would more or less unconditionally have to 1360 implement the mechanism, leading to unconditional additional 1361 complexity. There are good reasons to minimize this, as follows: 1363 o The increase in multiples of eight allows looking at an option and 1364 finding out whether it is critical or not even if the Long Jump 1365 value has just been skipped (as opposed to having been processed 1366 fully). (It also allows accessing up to approximately 2048 1367 options with a two-byte Long Jump.) This allows a basic 1368 implementation that does not implement any high-numbered options 1369 to simply ignore long jumps and any elective options behind them, 1370 while still properly reacting to critical options. 1372 o There is probably a good reason to disallow long-jumps that lead 1373 to an option number of 42 and less, enabling simple receivers to 1374 do the above simplification. 1376 o It might seem obvious to remove the fenceposting mechanism 1377 altogether in favor of long jumps. This is not advisable: 1378 Fenceposting already has zero implementation effort at the 1379 receiver, and the overhead at the sender is very limited (it is 1380 just a third kind of jump, at one byte per jump). Beyond 42, 1381 senders can ignore the existence of fenceposts if they want 1382 (possibly obviating the need for more complex base-14 arithmetic). 1384 There is no need for a finer granularity than 8, as the Option 1385 construct following can also specify a Delta of 0..14. (A 1386 granularity of 16 will require additional fenceposting where an 1387 option delta of 15 would happen to be required otherwise, which we 1388 have reserved. It can be argued that 16 is still the better choice, 1389 as fenceposting is already in the code path.) 1391 The Long Jump construct takes 0xf1 and 0xf2 from the space available 1392 for initial bytes of Options. (Note that we previously took 0xf0 to 1393 indicate end-of-options for OC=15.) 1394 Varying N with the length as defined above makes it unambiguous 1395 whether a one- or two-byte Long Jump is to be used. Setting N=2 for 1396 the one-byte version makes it clear that a Delta of 8 is to be 1397 handled the usual way (i.e., by Option Delta itself and/or 1398 fenceposting). If the delta is not small and not 7 modulo 8, there 1399 is still a choice between using the smaller multiple of 8 and a 1400 larger Delta in the actual Option or v.v., this biases the choice 1401 towards a larger Long Jump and a smaller following Delta, which is 1402 also easier to implement as it reduces the number of choice points. 1404 B.2.2.3. Example 1406 The following sequence of bytes would encode a Uri-Path Option of 1407 "foo" followed by Options 1357 (value "bar") and 1360 (value "baz"): 1409 93 65 6f 6f Option 9 (0 + 9, "foo") 1410 f1 a6 Long Jump by 1344 1411 43 62 61 72 Option 1357 (9 + 1344 + 4, "bar") 1412 33 62 61 7a Option 1360 (1357 + 3, "baz") 1414 Figure 17: Example using a Long Jump construct 1416 where f1 a6 is the long jump forward by (0xa6+2)*8=1344 option 1417 numbers. The total option count (OC) for the CoAP header is 3. Note 1418 that even if f1 a6 is skipped, the 1357 (which then appears as an 1419 Option number 13) is clearly visible as Critical. 1421 B.2.2.4. IANA considerations 1423 With the scheme proposed above, we could have three tiers of Option 1424 Numbers, differing in their allocation policy [RFC5226]: 1426 +---------------+-------------------------+ 1427 | Option Number | Policy | 1428 +---------------+-------------------------+ 1429 | 0..255 | Standards Action | 1430 | | | 1431 | 256..2047 | Designated Expert | 1432 | | | 1433 | 2048..65535 | First Come First Served | 1434 +---------------+-------------------------+ 1436 For the inventor of a new option, this would provide a small 1437 incentive to go through the designated expert for some minimal cross- 1438 checking in order to be able to use the two-byte long-jump. 1440 This draft adds option numbers to Table 2 of [I-D.ietf-core-coap]: 1442 +--------+---------------------+-----------+ 1443 | Number | Name | Reference | 1444 +--------+---------------------+-----------+ 1445 | 14 | Registered-Elective | [RFCXXXX] | 1446 | | | | 1447 | 25 | Registered-Critical | [RFCXXXX] | 1448 +--------+---------------------+-----------+ 1450 Table 1: New CoAP Option Numbers 1452 This draft adds a suboption registry, initially empty. 1454 +------------+-----------------------------+-----------+ 1455 | Number | Name | Reference | 1456 +------------+-----------------------------+-----------+ 1457 | 0..127 | (allocate on export review) | [RFCXXXX] | 1458 | | | | 1459 | 128..16383 | (allocate fcfs) | [RFCXXXX] | 1460 +------------+-----------------------------+-----------+ 1462 Table 2: CoAP Suboption Numbers 1464 B.3. Enabling Protocol Evolution 1466 To enable a protocol to evolve, it is critical that new capabilities 1467 can be introduced without requiring changes in components that don't 1468 really care about the capability. One such probem is exhibited by 1469 CoAP options: If a proxy does not understand an elective option in a 1470 request, it will not be able to forward it to the origin server, 1471 rendering the new option ineffectual. Worse, if a proxy does not 1472 understand a critical option in a request, it will not be able to 1473 operate on the request, rendering the new option damaging. 1475 As a conclusion to the Ticket #230 discussion in the June 4th interim 1476 call, we decided to solve the identification of options that a proxy 1477 can safely forward even if not understood (previously called Proxy- 1478 Elective). 1480 The proposal is to encode this information in the option number, just 1481 like the way the information that an option is critical is encoded 1482 now. This leads to two bits with semantics: the lowest bit continues 1483 to be the critical bit, and the next higher bit is now the "unsafe" 1484 bit (i.e., this option is not safe to forward unless understood by 1485 the proxy). 1487 Another consideration (for options that are not unsafe to forward) is 1488 whether the option should serve as a cache key in a request. HTTP 1489 has a vary header that indicates in the response which header fields 1490 were considered by the origin server to be cache keys. In order to 1491 avoid this complexity, we should be able to indicate this information 1492 right in the option number. However, reserving another bit is 1493 wasteful, in particular as there are few safe-to-forward options that 1494 are not cache-keys. 1496 Therefore, we propose the following bit allocation in an option 1497 number: 1499 xxx nnn UC 1501 Figure 18 1503 (where xxx is a variable length prefix, as option numbers are not 1504 bounded upwards). UC is the unsafe and critical bits. For U=0 only, 1505 if nnn is equal to 111 binary, the option does not serve as a cache 1506 key (for U=1, the proxy has to know the option to act on it, so there 1507 is no point in indicating whether it is a cache key). There is no 1508 semantic meaning of xxx. 1510 Note that clients and servers are generally not interested in this 1511 information. A proxy may use an equivalent of the following C code 1512 to derive the characteristics of an option number "onum": 1514 Critical = (onum & 1); 1515 UnSafe = (onum & 2); 1516 NoCache = ((onum & 0x1e) == 0x1c); 1518 Figure 19 1520 Discussion: This requires a renumbering of all options. 1522 This renumbering may also be considered as an opportunity to make 1523 the numbering straight again shortly before nailing down the 1524 protocol 1526 In particular, Content-Type is now probably better considered to 1527 be elective. 1529 B.3.1. Potential new option number allocation 1531 We want to give one example for a revised allocation of option 1532 numbers. Option numbers are given as decimal numbers, one each for 1533 xxx, nnn, and UC, with the UC values as follows 1535 +-----------+------------+------------------------------------+ 1536 | UC binary | UC decimal | meaning | 1537 +-----------+------------+------------------------------------+ 1538 | 00 | 0 | (safe, elective, 111=no-cache-key) | 1539 | | | | 1540 | 01 | 1 | (safe, critical, 111=no-cache-key) | 1541 | | | | 1542 | 10 | 2 | (unsafe, elective) | 1543 | | | | 1544 | 11 | 3 | (unsafe, critical) | 1545 +-----------+------------+------------------------------------+ 1547 The table is: 1549 +-----+-------+---------+-----------------------+-------------------+ 1550 | New | xx | Old | Name | Comment | 1551 | | nnn | | | | 1552 | | UC | | | | 1553 +-----+-------+---------+-----------------------+-------------------+ 1554 | 4 | 0 1 0 | 1 | Content-Type | category change | 1555 | | | | | (elective) | 1556 | | | | | | 1557 | 8 | 0 2 0 | 4 | ETag | | 1558 | | | | | | 1559 | 12 | 0 3 0 | 12 | Accept | | 1560 | | | | | | 1561 | 16 | 0 4 0 | 6 | Location-Path | | 1562 | | | | | | 1563 | 20 | 0 5 0 | 8 | Location-Query | | 1564 | | | | | | 1565 | 24 | 0 6 0 | - | (unused) | | 1566 | | | | | | 1567 | 28 | 0 7 0 | 18 | Size | needs nnn=111 | 1568 | | | | | | 1569 | 32 | 1 0 0 | 20/22 | Patience | | 1570 | | | | | | 1571 | 64 | 2 x 0 | - | Location-reserved | (nnn = 0..3, 4 | 1572 | | | | | reserved numbers) | 1573 | | | | | | 1574 | 1 | 0 0 1 | 13 | If-Match | | 1575 | | | | | | 1576 | 5 | 0 1 1 | 21 | If-None-Match | | 1577 | | | | | | 1578 | 2 | 0 0 2 | 2 | Max-Age | | 1579 | | | | | | 1580 | 6 | 0 1 2 | 10 | Observe | | 1581 | | | | | | 1582 | 10 | 0 2 2 | xx | Observe-2 | | 1583 | | | | | | 1584 | 14 | 0 3 2 | xx | (unused) | was fencepost | 1585 | | | | | | 1586 | 3 | 0 0 3 | 3 | Proxy-Uri | | 1587 | | | | | | 1588 | 7 | 0 1 3 | 5 | Uri-Host | | 1589 | | | | | | 1590 | 11 | 0 2 3 | 7 | Uri-Port | | 1591 | | | | | | 1592 | 15 | 0 3 3 | 9 | Uri-Path | | 1593 | | | | | | 1594 | 19 | 0 4 3 | 15 | Uri-Query | | 1595 | | | | | | 1596 | 23 | 0 5 3 | 11 | Token | | 1597 | | | | | | 1598 | 27 | 0 6 3 | 17 | Block2 | | 1599 | | | | | | 1600 | 31 | 0 7 3 | 19 | Block1 | yes, we can use | 1601 | | | | | nnn=111 with U=1 | 1602 +-----+-------+---------+-----------------------+-------------------+ 1604 B.4. Patience, Leisure, and Pledge 1606 A number of options might be useful for controlling the timing of 1607 interactions. 1609 (This section also addresses core-coap ticket #177.) 1611 B.4.1. Patience 1613 A client may have a limited time period in which it can actually make 1614 use of the response for a request. Using the Patience option, it can 1615 provide an (elective) indication how much time it is willing to wait 1616 for the response from the server, giving the server license to ignore 1617 or reject the request if it cannot fulfill it in this period. 1619 If the server knows early that it cannot fulfill the request in the 1620 time requested, it MAY indicate this with a 5.04 "Timeout" response. 1621 For non-safe methods (such as PUT, POST, DELETE), the server SHOULD 1622 indicate whether it has fulfilled the request by either responding 1623 with 5.04 "Timeout" (and not further processing the request) or by 1624 processing the request normally. 1626 Note that the value of the Patience option should be chosen such that 1627 the client will be able to make use of the result even in the 1628 presence of the expected network delays for the request and the 1629 response. Similarly, when a proxy receives a request with a Patience 1630 option and cannot fulfill that request from its cache, it may want to 1631 adjust the value of the option before forwarding it to an upstream 1632 server. 1634 (TBD: The various cases that arise when combining Patience with 1635 Observe.) 1637 The Patience option is elective. Hence, a client MUST be prepared to 1638 receive a normal response even after the chosen Patience period (plus 1639 an allowance for network delays) has elapsed. 1641 B.4.2. Leisure 1643 Servers generally will compute an internal value that we will call 1644 Leisure, which indicates the period of time that will be used for 1645 responding to a request. A Patience option, if present, can be used 1646 as an upper bound for the Leisure. Leisure may be non-zero for 1647 congestion control reasons, in particular for responses to multicast 1648 requests. For these, the server should have a group size estimate G, 1649 a target rate R (which both should be chosen conservatively) and an 1650 estimated response size S; a rough lower bound for Leisure can then 1651 be computed as follows: 1653 lb_Leisure = S * G / R 1655 Figure 20: Computing a lower bound for the Leisure 1657 E.g., for a multicast request with link-local scope on an 2.4 GHz 1658 IEEE 802.15.4 (6LoWPAN) network, G could be (relatively 1659 conservatively) set to 100, S to 100 bytes, and the target rate to 8 1660 kbit/s = 1 kB/s. The resulting lower bound for the Leisure is 10 1661 seconds. 1663 To avoid response implosion, responses to multicast requests SHOULD 1664 be dithered within a Leisure period chosen by the server to fall 1665 between these two bounds. 1667 Currently, we don't foresee a need to signal a value for Leisure from 1668 client to server (beyond the signalling provided by Patience) or from 1669 server to client, but an appropriate Option might be added later. 1671 B.4.3. Pledge 1672 In a basic observation relationship [I-D.ietf-core-observe], the 1673 server makes a pledge to keep the client in the observation 1674 relationship for a resource at least until the max-age for the 1675 resource is reached. 1677 To save the client some effort in re-establishing observation 1678 relationships each time max-age is reached, the server MAY want to 1679 extend its pledge beyond the end of max-age by signalling in a 1680 response/notification an additional time period using the Pledge 1681 Option, in parallel to the Observe Option. 1683 The Pledge Option MUST NOT be used unless the server can make a 1684 reasonable promise not to lose the observation relationship in this 1685 time frame. 1687 Currently, we don't foresee a need to signal a value for Pledge from 1688 client to server, but an appropriate behavior might be added later 1689 for this option when sent in a request. 1691 B.4.4. Option Formats 1693 +-----+----------+----------+-----------------+--------+---------+ 1694 | No. | C/E | Name | Format | Length | Default | 1695 +-----+----------+----------+-----------------+--------+---------+ 1696 | 22 | Elective | Patience | Duration in mis | 1 B | (none) | 1697 | | | | | | | 1698 | 24 | Elective | Pledge | Duration in s | 1 B | 0 | 1699 +-----+----------+----------+-----------------+--------+---------+ 1701 All timing options use the Duration data type (see Appendix D.2), 1702 however Patience (and Leisure, if that ever becomes an option) uses a 1703 timebase of mibiseconds (mis = 1/1024 s) instead of seconds. (This 1704 reduces the range of the Duration from ~ 91 days to 128 minutes.) 1706 Implementation note: As there are no strong accuracy requirements on 1707 the clocks employed, making use of any existing time base of 1708 milliseconds is a valid implementation approach (2.4 % off). 1710 None of the options may be repeated. 1712 Appendix C. The Cemetery (Things we won't do) 1714 This annex documents roads that the WG decided not to take, in order 1715 to spare readers from reinventing them in vain. 1717 C.1. Example envelope option: solving #230 1718 Ticket #230 [CoRE230] points out a design flaw of 1719 [I-D.ietf-core-coap]: When we split the elective Location option of 1720 draft -01 into multiple elective options, we made it possible that an 1721 implementation might process some of these and ignore others, leading 1722 to an incorrect interpretation of the Location expressed by the 1723 server. 1725 There are several more or less savory solutions to #230. 1727 Each of the elective options that together make up the Location could 1728 be defined in such a way that it makes a requirement on the 1729 processing of the related option (essentially revoking their elective 1730 status once the option under consideration is actually processed). 1731 This falls flat as soon as another option is defined that would also 1732 become part of the Location: existing implementations would not know 1733 that the new option is also part of the cluster that is re- 1734 interpreted as critical. The potential future addition of Location- 1735 Host and Location-Port makes this a valid consideration. 1737 A better solution would be to define an elective Envelope Option 1738 called Location. Within a Location Option, the following top-level 1739 options might be allowed (now or in the future): 1741 o Uri-Host 1743 o Uri-Port 1745 o Uri-Path 1747 o Uri-Query 1749 This would unify the code for interpreting the top-level request 1750 options that indicate the request URI with the code that interprets 1751 the Location URI. 1753 The four options listed are all critical, while the envelope is 1754 elective. This gives exactly the desired semantics: If the envelope 1755 is processed at all (which is elective), the nested options are 1756 critical and all need to be processed. 1758 C.2. Example envelope option: proxy-elective options 1759 Another potential application of envelope options is motivated by the 1760 observation that new critical options might not be implemented by all 1761 proxies on the CoAP path to an origin server. So that this does not 1762 become an obstacle to introducing new critical options that are of 1763 interest only to client and origin server, the client might want to 1764 mark some critical options proxy-elective, i.e. elective for a proxy 1765 but still critical for the origin server. 1767 One way to do this would be an Envelope option, the Proxy-Elective 1768 Option. A client might bundle a number of critical options into a 1769 critical Proxy-Elective Option. A proxy that processes the message 1770 is obliged to process the envelope (or reject the message), where 1771 processing means passing on the nested options towards the origin 1772 server (preferably again within a Proxy-Elective option). It can 1773 pass on the nested options, even ones unknown to the proxy, knowing 1774 that the client is happy with proxies not processing all of them. 1776 (The assumption here is that the Proxy-Elective option becomes part 1777 of the base standard, so all but the most basic proxies would know 1778 how to handle it.) 1780 C.3. Stateful URI compression 1782 Is the approximately 25 % average saving achievable with Huffman- 1783 based URI compression schemes worth the complexity? Probably not, 1784 because much higher average savings can be achieved by introducing 1785 state. 1787 Henning Schulzrinne has proposed for a server to be able to supply a 1788 shortened URI once a resource has been requested using the full- 1789 length URI. Let's call such a shortened referent a _Temporary 1790 Resource Identifier_, _TeRI_ for short. This could be expressed by a 1791 response option as shown in Figure 21. 1793 0 1794 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 1795 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1796 | duration | TeRI... 1797 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1799 Figure 21: Option for offering a TeRI in a response 1801 The TeRI offer option indicates that the server promises to offer 1802 this resources under the TeRI given for at least the time given as 1803 the duration. Another TeRI offer can be made later to extend the 1804 duration. 1806 Once a TeRI for a URI is known (and still within its lifetime), the 1807 client can supply a TeRI instead of a URI in its requests. The same 1808 option format as an offer could be used to allow the client to 1809 indicate how long it believes the TeRI will still be valid (so that 1810 the server can decide when to update the lifetime duration). TeRIs 1811 in requests could be distinguished from URIs e.g. by using a 1812 different option number. 1814 Proposal: Add a TeRI option that can be used in CoAP requests and 1815 responses. 1817 Add a way to indicate a TeRI and its duration in a link-value. 1819 Do not add any form of stateless URI encoding. 1821 Benefits: Much higher reduction of message size than any stateless 1822 URI encoding could achieve. 1824 As the use of TeRIs is entirely optional, minimal complexity nodes 1825 can get by without implementing them. 1827 Drawbacks: Adds considerable state and complexity to the protocol. 1829 It turns out that real CoAP URIs are short enough that TeRIs are 1830 not needed. 1832 (Discuss the security implications of TeRIs.) 1834 Appendix D. Experimental Options 1836 This annex documents proposals that need significant additional 1837 discussion before they can become part of (or go back to) the main 1838 CoAP specification. They are not dead, but might die if there turns 1839 out to be no good way to solve the problem. 1841 D.1. Options indicating absolute time 1843 HTTP has a number of headers that may indicate absolute time: 1845 o "Date", defined in Section 14.18 in [RFC2616] (Section 9.3 in 1846 [I-D.ietf-httpbis-p1-messaging]), giving the absolute time a 1847 response was generated; 1849 o "Last-Modified", defined in Section 14.29 in [RFC2616], 1850 (Section 6.6 in [I-D.ietf-httpbis-p4-conditional], giving the 1851 absolute time of when the origin server believes the resource 1852 representation was last modified; 1854 o "If-Modified-Since", defined in Section 14.25 in [RFC2616], "If- 1855 Unmodified-Since", defined in Section 14.28 in [RFC2616], and "If- 1856 Range", defined in Section 14.27 in [RFC2616] can be used to 1857 supply absolute time to gate a conditional request; 1859 o "Expires", defined in Section 14.21 in [RFC2616] (Section 3.3 in 1860 [I-D.ietf-httpbis-p6-cache]), giving the absolute time after which 1861 a response is considered stale. 1863 o The more obscure headers "Retry-After", defined in Section 14.37 1864 in [RFC2616], and "Warning", defined in section 14.46 in 1865 [RFC2616], also may employ absolute time. 1867 [I-D.ietf-core-coap] defines a single "Date" option, which however 1868 "indicates the creation time and date of a given resource 1869 representation", i.e., is closer to a "Last-Modified" HTTP header. 1870 HTTP's caching rules [I-D.ietf-httpbis-p6-cache] make use of both 1871 "Date" and "Last-Modified", combined with "Expires". The specific 1872 semantics required for CoAP needs further consideration. 1874 In addition to the definition of the semantics, an encoding for 1875 absolute times needs to be specified. 1877 In UNIX-related systems, it is customary to indicate absolute time as 1878 an integer number of seconds, after midnight UTC, January 1, 1970. 1879 Unless negative numbers are employed, this time format cannot 1880 represent time values prior to January 1, 1970, which probably is not 1881 required for the uses ob absolute time in CoAP. 1883 If a 32-bit integer is used and allowance is made for a sign-bit in a 1884 local implementation, the latest UTC time value that can be 1885 represented by the resulting 31 bit integer value is 03:14:07 on 1886 January 19, 2038. If the 32-bit integer is used as an unsigned 1887 value, the last date is 2106-02-07, 06:28:15. 1889 The reach can be extended by: - moving the epoch forward, e.g. by 40 1890 years (= 1262304000 seconds) to 2010-01-01. This makes it impossible 1891 to represent Last-Modified times in that past (such as could be 1892 gatewayed in from HTTP). - extending the number of bits, e.g. by 1893 one more byte, either always or as one of two formats, keeping the 1894 32-bit variant as well. 1896 Also, the resolution can be extended by expressing time in 1897 milliseconds etc., requiring even more bits (e.g., a 48-bit unsigned 1898 integer of milliseconds would last well after year 9999.) 1900 For experiments, an experimental "Date" option is defined with the 1901 semantics of HTTP's "Last-Modified". It can carry an unsigned 1902 integer of 32, 40, or 48 bits; 32- and 40-bit integers indicate the 1903 absolute time in seconds since 1970-01-01 00:00 UTC, while 48-bit 1904 integers indicate the absolute time in milliseconds since 1970-01-01 1905 00:00 UTC. 1907 However, that option is not really that useful until there is a "If- 1908 Modified-Since" option as well. 1910 (Also: Discuss nodes without clocks.) 1912 D.2. Representing Durations 1914 Various message types used in CoAP need the representation of 1915 *durations*, i.e. of the length of a timespan. In SI units, these 1916 are measured in seconds. CoAP durations represent integer numbers of 1917 seconds, but instead of representing these numbers as integers, a 1918 more compact single-byte pseudo-floating-point (pseudo-FP) 1919 representation is used (Figure 22). 1921 0 1 2 3 4 5 6 7 1922 +---+---+---+---+---+---+---+---+ 1923 | 0... value | 1924 +---+---+---+---+---+---+---+---+ 1926 +---+---+---+---+---+---+---+---+ 1927 | 1... mantissa | exponent | 1928 +---+---+---+---+---+---+---+---+ 1930 Figure 22: Duration in (8,4) pseudo-FP representation 1932 If the high bit is clear, the entire n-bit value (including the high 1933 bit) is the decoded value. If the high bit is set, the mantissa 1934 (including the high bit, with the exponent field cleared out but 1935 still present) is shifted left by the exponent to yield the decoded 1936 value. 1938 The (n,e)-pseudo-FP format can be decoded with a single line of code 1939 (plus a couple of constant definitions), as demonstrated in Figure 1940 23. 1942 #define N 8 1943 #define E 4 1944 #define HIBIT (1 << (N - 1)) 1945 #define EMASK ((1 << E) - 1) 1946 #define MMASK ((1 << N) - 1 - EMASK) 1948 #define DECODE_8_4(r) (r < HIBIT ? r : (r & MMASK) << (r & EMASK)) 1949 Figure 23: Decoding an (8,4) pseudo-FP value 1951 Note that a pseudo-FP encoder needs to consider rounding; different 1952 applications of durations may favor rounding up or rounding down the 1953 value encoded in the message. 1955 The highest pseudo-FP value, represented by an all-ones byte (0xFF), 1956 is reserved to indicate an indefinite duration. The next lower value 1957 (0xEF) is thus the highest representable value and is decoded as 1958 7340032 seconds, a little more than 12 weeks. 1960 D.3. Rationale 1962 Where CPU power and memory is abundant, a duration can almost always 1963 be adequately represented by a non-negative floating-point number 1964 representing that number of seconds. Historically, many APIs have 1965 also used an integer representation, which limits both the resolution 1966 (e.g., if the integer represents the duration in seconds) and often 1967 the range (integer machine types have range limits that may become 1968 relevant). UNIX's "time_t" (which is used for both absolute time and 1969 durations) originally was a signed 32-bit value of seconds, but was 1970 later complemented by an additional integer to add microsecond 1971 ("struct timeval") and then later nanosecond ("struct timespec") 1972 resolution. 1974 Three decisions need to be made for each application of the concept 1975 of duration: 1977 o the *resolution*. What rounding error is acceptable? 1979 o the *range*. What is the maximum duration that needs to be 1980 represented? 1982 o the *number of bits* that can be expended. 1984 Obviously, these decisions are interrelated. Typically, a large 1985 range needs a large number of bits, unless resolution is traded. For 1986 most applications, the actual requirement for resolution are limited 1987 for longer durations, but can be more acute for shorter durations. 1989 D.4. Pseudo-Floating Point 1991 Constrained systems typically avoid the use of floating-point (FP) 1992 values, as 1994 o simple CPUs often don't have support for floating-point datatypes 1995 o software floating-point libraries are expensive in code size and 1996 slow. 1998 In addition, floating-point datatypes used to be a significant 1999 element of market differentiation in CPU design; it has taken the 2000 industry a long time to agree on a standard floating point 2001 representation. 2003 These issues have led to protocols that try to constrain themselves 2004 to integer representation even where the ability of a floating point 2005 representation to trade range for resolution would be beneficial. 2007 The idea of introducing _pseudo-FP_ is to obtain the increased range 2008 provided by embedding an exponent, without necessarily getting stuck 2009 with hardware datatypes or inefficient software floating-point 2010 libraries. 2012 For the purposes of this draft, we define an (n,e)-pseudo-FP as a 2013 fixed-length value of n bits, e of which may be used for an exponent. 2014 Figure 22 illustrates an (8,4)-pseudo-FP value. 2016 If the high bit is clear, the entire n-bit value (including the high 2017 bit) is the decoded value. If the high bit is set, the mantissa 2018 (including the high bit, but with the exponent field cleared out) is 2019 shifted left by the exponent to yield the decoded value. 2021 The (n,e)-pseudo-FP format can be decoded with a single line of code 2022 (plus a couple of constant definition), as demonstrated in Figure 23. 2024 Only non-negative numbers can be represented by this format. It is 2025 designed to provide full integer resolution for values from 0 to 2026 2^(n-1)-1, i.e., 0 to 127 in the (8,4) case, and a mantissa of n-e 2027 bits from 2^(n-1) to (2^n-2^e)*2^(2^e-1), i.e., 128 to 7864320 in the 2028 (8,4) case. By choosing e carefully, resolution can be traded 2029 against range. 2031 Note that a pseudo-FP encoder needs to consider rounding; different 2032 applications of durations may favor rounding up or rounding down the 2033 value encoded in the message. This requires a little more than a 2034 single line of code (which is left as an exercise to the reader, as 2035 the most efficient expression depends on hardware details). 2037 D.5. A Duration Type for CoAP 2038 CoAP needs durations in a number of places. In [I-D.ietf-core-coap], 2039 durations occur in the option "Subscription-lifetime" as well as in 2040 the option "Max-age". (Note that the option "Date" is not a 2041 duration, but a point in time.) Other durations of this kind may be 2042 added later. 2044 Most durations relevant to CoAP are best expressed with a minimum 2045 resolution of one second. More detailed resolutions are unlikely to 2046 provide much benefit. 2048 The range of lifetimes and caching ages are probably best kept below 2049 the order of magnitude of months. An (8,4)-pseudo-FP has the maximum 2050 value of 7864320, which is about 91 days; this appears to be adequate 2051 for a subscription lifetime and probably even for a maximum cache 2052 age. Figure 24 shows the values that can be expressed. (If a larger 2053 range for the latter is indeed desired, an (8,5)-pseudo-FP could be 2054 used; this would last 15 milleniums, at the cost of having only 3 2055 bits of accuracy for values larger than 127 seconds.) 2057 Proposal: A single duration type is used throughout CoAP, based on 2058 an (8,4)-pseudo-FP giving a duration in seconds. 2060 Benefits: Implementations can use a single piece of code for 2061 managing all CoAP-related durations. 2063 In addition, length information never needs to be managed for 2064 durations that are embedded in other data structures: All 2065 durations are expressed by a single byte. 2067 It might be worthwhile to reserve one duration value, e.g. 0xFF, for 2068 an indefinite duration. 2070 Duration Seconds Encoded 2071 ----------- ---------- ------- 2072 00:00:00 0x00000000 0x00 2073 00:00:01 0x00000001 0x01 2074 00:00:02 0x00000002 0x02 2075 00:00:03 0x00000003 0x03 2076 00:00:04 0x00000004 0x04 2077 00:00:05 0x00000005 0x05 2078 00:00:06 0x00000006 0x06 2079 00:00:07 0x00000007 0x07 2080 00:00:08 0x00000008 0x08 2081 00:00:09 0x00000009 0x09 2082 00:00:10 0x0000000a 0x0a 2083 00:00:11 0x0000000b 0x0b 2084 00:00:12 0x0000000c 0x0c 2085 00:00:13 0x0000000d 0x0d 2086 00:00:14 0x0000000e 0x0e 2087 00:00:15 0x0000000f 0x0f 2088 00:00:16 0x00000010 0x10 2089 00:00:17 0x00000011 0x11 2090 00:00:18 0x00000012 0x12 2091 00:00:19 0x00000013 0x13 2092 00:00:20 0x00000014 0x14 2093 00:00:21 0x00000015 0x15 2094 00:00:22 0x00000016 0x16 2095 00:00:23 0x00000017 0x17 2096 00:00:24 0x00000018 0x18 2097 00:00:25 0x00000019 0x19 2098 00:00:26 0x0000001a 0x1a 2099 00:00:27 0x0000001b 0x1b 2100 00:00:28 0x0000001c 0x1c 2101 00:00:29 0x0000001d 0x1d 2102 00:00:30 0x0000001e 0x1e 2103 00:00:31 0x0000001f 0x1f 2104 00:00:32 0x00000020 0x20 2105 00:00:33 0x00000021 0x21 2106 00:00:34 0x00000022 0x22 2107 00:00:35 0x00000023 0x23 2108 00:00:36 0x00000024 0x24 2109 00:00:37 0x00000025 0x25 2110 00:00:38 0x00000026 0x26 2111 00:00:39 0x00000027 0x27 2112 00:00:40 0x00000028 0x28 2113 00:00:41 0x00000029 0x29 2114 00:00:42 0x0000002a 0x2a 2115 00:00:43 0x0000002b 0x2b 2116 00:00:44 0x0000002c 0x2c 2117 00:00:45 0x0000002d 0x2d 2118 00:00:46 0x0000002e 0x2e 2119 00:00:47 0x0000002f 0x2f 2120 00:00:48 0x00000030 0x30 2121 00:00:49 0x00000031 0x31 2122 00:00:50 0x00000032 0x32 2123 00:00:51 0x00000033 0x33 2124 00:00:52 0x00000034 0x34 2125 00:00:53 0x00000035 0x35 2126 00:00:54 0x00000036 0x36 2127 00:00:55 0x00000037 0x37 2128 00:00:56 0x00000038 0x38 2129 00:00:57 0x00000039 0x39 2130 00:00:58 0x0000003a 0x3a 2131 00:00:59 0x0000003b 0x3b 2132 00:01:00 0x0000003c 0x3c 2133 00:01:01 0x0000003d 0x3d 2134 00:01:02 0x0000003e 0x3e 2135 00:01:03 0x0000003f 0x3f 2136 00:01:04 0x00000040 0x40 2137 00:01:05 0x00000041 0x41 2138 00:01:06 0x00000042 0x42 2139 00:01:07 0x00000043 0x43 2140 00:01:08 0x00000044 0x44 2141 00:01:09 0x00000045 0x45 2142 00:01:10 0x00000046 0x46 2143 00:01:11 0x00000047 0x47 2144 00:01:12 0x00000048 0x48 2145 00:01:13 0x00000049 0x49 2146 00:01:14 0x0000004a 0x4a 2147 00:01:15 0x0000004b 0x4b 2148 00:01:16 0x0000004c 0x4c 2149 00:01:17 0x0000004d 0x4d 2150 00:01:18 0x0000004e 0x4e 2151 00:01:19 0x0000004f 0x4f 2152 00:01:20 0x00000050 0x50 2153 00:01:21 0x00000051 0x51 2154 00:01:22 0x00000052 0x52 2155 00:01:23 0x00000053 0x53 2156 00:01:24 0x00000054 0x54 2157 00:01:25 0x00000055 0x55 2158 00:01:26 0x00000056 0x56 2159 00:01:27 0x00000057 0x57 2160 00:01:28 0x00000058 0x58 2161 00:01:29 0x00000059 0x59 2162 00:01:30 0x0000005a 0x5a 2163 00:01:31 0x0000005b 0x5b 2164 00:01:32 0x0000005c 0x5c 2165 00:01:33 0x0000005d 0x5d 2166 00:01:34 0x0000005e 0x5e 2167 00:01:35 0x0000005f 0x5f 2168 00:01:36 0x00000060 0x60 2169 00:01:37 0x00000061 0x61 2170 00:01:38 0x00000062 0x62 2171 00:01:39 0x00000063 0x63 2172 00:01:40 0x00000064 0x64 2173 00:01:41 0x00000065 0x65 2174 00:01:42 0x00000066 0x66 2175 00:01:43 0x00000067 0x67 2176 00:01:44 0x00000068 0x68 2177 00:01:45 0x00000069 0x69 2178 00:01:46 0x0000006a 0x6a 2179 00:01:47 0x0000006b 0x6b 2180 00:01:48 0x0000006c 0x6c 2181 00:01:49 0x0000006d 0x6d 2182 00:01:50 0x0000006e 0x6e 2183 00:01:51 0x0000006f 0x6f 2184 00:01:52 0x00000070 0x70 2185 00:01:53 0x00000071 0x71 2186 00:01:54 0x00000072 0x72 2187 00:01:55 0x00000073 0x73 2188 00:01:56 0x00000074 0x74 2189 00:01:57 0x00000075 0x75 2190 00:01:58 0x00000076 0x76 2191 00:01:59 0x00000077 0x77 2192 00:02:00 0x00000078 0x78 2193 00:02:01 0x00000079 0x79 2194 00:02:02 0x0000007a 0x7a 2195 00:02:03 0x0000007b 0x7b 2196 00:02:04 0x0000007c 0x7c 2197 00:02:05 0x0000007d 0x7d 2198 00:02:06 0x0000007e 0x7e 2199 00:02:07 0x0000007f 0x7f 2200 00:02:08 0x00000080 0x80 2201 00:02:24 0x00000090 0x90 2202 00:02:40 0x000000a0 0xa0 2203 00:02:56 0x000000b0 0xb0 2204 00:03:12 0x000000c0 0xc0 2205 00:03:28 0x000000d0 0xd0 2206 00:03:44 0x000000e0 0xe0 2207 00:04:00 0x000000f0 0xf0 2208 00:04:16 0x00000100 0x81 2209 00:04:48 0x00000120 0x91 2210 00:05:20 0x00000140 0xa1 2211 00:05:52 0x00000160 0xb1 2212 00:06:24 0x00000180 0xc1 2213 00:06:56 0x000001a0 0xd1 2214 00:07:28 0x000001c0 0xe1 2215 00:08:00 0x000001e0 0xf1 2216 00:08:32 0x00000200 0x82 2217 00:09:36 0x00000240 0x92 2218 00:10:40 0x00000280 0xa2 2219 00:11:44 0x000002c0 0xb2 2220 00:12:48 0x00000300 0xc2 2221 00:13:52 0x00000340 0xd2 2222 00:14:56 0x00000380 0xe2 2223 00:16:00 0x000003c0 0xf2 2224 00:17:04 0x00000400 0x83 2225 00:19:12 0x00000480 0x93 2226 00:21:20 0x00000500 0xa3 2227 00:23:28 0x00000580 0xb3 2228 00:25:36 0x00000600 0xc3 2229 00:27:44 0x00000680 0xd3 2230 00:29:52 0x00000700 0xe3 2231 00:32:00 0x00000780 0xf3 2232 00:34:08 0x00000800 0x84 2233 00:38:24 0x00000900 0x94 2234 00:42:40 0x00000a00 0xa4 2235 00:46:56 0x00000b00 0xb4 2236 00:51:12 0x00000c00 0xc4 2237 00:55:28 0x00000d00 0xd4 2238 00:59:44 0x00000e00 0xe4 2239 01:04:00 0x00000f00 0xf4 2240 01:08:16 0x00001000 0x85 2241 01:16:48 0x00001200 0x95 2242 01:25:20 0x00001400 0xa5 2243 01:33:52 0x00001600 0xb5 2244 01:42:24 0x00001800 0xc5 2245 01:50:56 0x00001a00 0xd5 2246 01:59:28 0x00001c00 0xe5 2247 02:08:00 0x00001e00 0xf5 2248 02:16:32 0x00002000 0x86 2249 02:33:36 0x00002400 0x96 2250 02:50:40 0x00002800 0xa6 2251 03:07:44 0x00002c00 0xb6 2252 03:24:48 0x00003000 0xc6 2253 03:41:52 0x00003400 0xd6 2254 03:58:56 0x00003800 0xe6 2255 04:16:00 0x00003c00 0xf6 2256 04:33:04 0x00004000 0x87 2257 05:07:12 0x00004800 0x97 2258 05:41:20 0x00005000 0xa7 2259 06:15:28 0x00005800 0xb7 2260 06:49:36 0x00006000 0xc7 2261 07:23:44 0x00006800 0xd7 2262 07:57:52 0x00007000 0xe7 2263 08:32:00 0x00007800 0xf7 2264 09:06:08 0x00008000 0x88 2265 10:14:24 0x00009000 0x98 2266 11:22:40 0x0000a000 0xa8 2267 12:30:56 0x0000b000 0xb8 2268 13:39:12 0x0000c000 0xc8 2269 14:47:28 0x0000d000 0xd8 2270 15:55:44 0x0000e000 0xe8 2271 17:04:00 0x0000f000 0xf8 2272 18:12:16 0x00010000 0x89 2273 20:28:48 0x00012000 0x99 2274 22:45:20 0x00014000 0xa9 2275 1d 01:01:52 0x00016000 0xb9 2276 1d 03:18:24 0x00018000 0xc9 2277 1d 05:34:56 0x0001a000 0xd9 2278 1d 07:51:28 0x0001c000 0xe9 2279 1d 10:08:00 0x0001e000 0xf9 2280 1d 12:24:32 0x00020000 0x8a 2281 1d 16:57:36 0x00024000 0x9a 2282 1d 21:30:40 0x00028000 0xaa 2283 2d 02:03:44 0x0002c000 0xba 2284 2d 06:36:48 0x00030000 0xca 2285 2d 11:09:52 0x00034000 0xda 2286 2d 15:42:56 0x00038000 0xea 2287 2d 20:16:00 0x0003c000 0xfa 2288 3d 00:49:04 0x00040000 0x8b 2289 3d 09:55:12 0x00048000 0x9b 2290 3d 19:01:20 0x00050000 0xab 2291 4d 04:07:28 0x00058000 0xbb 2292 4d 13:13:36 0x00060000 0xcb 2293 4d 22:19:44 0x00068000 0xdb 2294 5d 07:25:52 0x00070000 0xeb 2295 5d 16:32:00 0x00078000 0xfb 2296 6d 01:38:08 0x00080000 0x8c 2297 6d 19:50:24 0x00090000 0x9c 2298 7d 14:02:40 0x000a0000 0xac 2299 8d 08:14:56 0x000b0000 0xbc 2300 9d 02:27:12 0x000c0000 0xcc 2301 9d 20:39:28 0x000d0000 0xdc 2302 10d 14:51:44 0x000e0000 0xec 2303 11d 09:04:00 0x000f0000 0xfc 2304 12d 03:16:16 0x00100000 0x8d 2305 13d 15:40:48 0x00120000 0x9d 2306 15d 04:05:20 0x00140000 0xad 2307 16d 16:29:52 0x00160000 0xbd 2308 18d 04:54:24 0x00180000 0xcd 2309 19d 17:18:56 0x001a0000 0xdd 2310 21d 05:43:28 0x001c0000 0xed 2311 22d 18:08:00 0x001e0000 0xfd 2312 24d 06:32:32 0x00200000 0x8e 2313 27d 07:21:36 0x00240000 0x9e 2314 30d 08:10:40 0x00280000 0xae 2315 33d 08:59:44 0x002c0000 0xbe 2316 36d 09:48:48 0x00300000 0xce 2317 39d 10:37:52 0x00340000 0xde 2318 42d 11:26:56 0x00380000 0xee 2319 45d 12:16:00 0x003c0000 0xfe 2320 48d 13:05:04 0x00400000 0x8f 2321 54d 14:43:12 0x00480000 0x9f 2322 60d 16:21:20 0x00500000 0xaf 2323 66d 17:59:28 0x00580000 0xbf 2324 72d 19:37:36 0x00600000 0xcf 2325 78d 21:15:44 0x00680000 0xdf 2326 84d 22:53:52 0x00700000 0xef 2327 91d 00:32:00 0x00780000 0xff (reserved) 2329 Figure 24 2331 D.6. CONTOUR (CoAP Non-trivial Option Useful Representation) 2333 So far, CoAP options have been simple. For non-trivial options of 2334 the future, e.g. for interaction with cellular systems, it is useful 2335 to have an option value encoding that can represent tree-structured 2336 data. Several such data structures are good candidates. If none of 2337 them work out, the PlanB encoding below may be useful. 2339 D.6.1. Specification of the PlanB Encoding 2341 A PlanB encoded data item is structured and encoded as described in 2342 this section. 2344 D.6.1.1. Structure of a Data Item 2346 +-+-+-+-+-+-+-+-+ 2347 | mt | ai | 2348 +-+-+-+-+-+-+-+-+ 2350 Figure 25: Initial Byte of a Date Item 2352 The initial byte Figure 25 conveys two pieces of information: 2354 o The major type "mt" (3 bits) 2356 o The argument information "ai" (5 bits) 2358 * The argument value "av" itself, if between 0 and 27 (embedded 2359 argument) 2361 * The encoded length of the argument data (separate argument) 2363 The initial byte is followed by the separate argument data "ad", if 2364 any, as controlled by the argument information Table 3; initial byte 2365 and argument data together define the argument value "av" ("uint" 2366 stands for an unsigned integer represented in network byte order). 2368 +----------+--------------------+----------------------------+ 2369 | ai value | ad length in bytes | interpretation of "av" | 2370 +----------+--------------------+----------------------------+ 2371 | 0-27 | 0 | value of "ai" | 2372 | | | | 2373 | 28 | 1 | uint value of "ad" + 28 | 2374 | | | | 2375 | 29 | 2 | uint value of "ad" | 2376 | | | | 2377 | 30 | 4 | uint value of "ad" | 2378 | | | | 2379 | 31 | 8 | uint integer value of "ad" | 2380 +----------+--------------------+----------------------------+ 2382 Table 3: Length and Interpretation of Argument Data 2384 As a quality of implementation concern, implementations are expected 2385 to use the shortest representation for an argument value. However, 2386 decoders MUST be able to decode longer representations. (See also 2387 Appendix D.6.3.2 below.) 2389 Finally, if the major type is one that carries Content Data, the 2390 content data follows as indicated by the argument. 2392 +-----+--------------------------+-----------------+ 2393 | mt | semantics | content data | 2394 +-----+--------------------------+-----------------+ 2395 | 000 | unsigned integer av | -\u002D | 2396 | | | | 2397 | 001 | negative integer -1 - av | -\u002D | 2398 | | | | 2399 | 010 | (special av) | -\u002D | 2400 | | | | 2401 | 011 | (tagged by av) | 1 data item | 2402 | | | | 2403 | 100 | byte string | av bytes | 2404 | | | | 2405 | 101 | UTF-8 string | av bytes | 2406 | | | | 2407 | 110 | array | av data items | 2408 | | | | 2409 | 111 | map | 2 av data items | 2410 +-----+--------------------------+-----------------+ 2412 Table 4: Semantics of a Data Item 2414 D.6.1.2. Integers 2416 Unsigned (non-negative) and negative integers are indicated by the 2417 major types mt=000 and mt=001, respectively. 2419 D.6.1.3. Special Items 2420 The semantics of a Special Item depends on the value of "ai", as 2421 defined in Table 5. 2423 +--------+------------------------------------------------+ 2424 | ai | semantics | 2425 +--------+------------------------------------------------+ 2426 | 0..15 | reserved for future optimization | 2427 | | | 2428 | 16..23 | (reserved) | 2429 | | | 2430 | 24 | False | 2431 | | | 2432 | 25 | True | 2433 | | | 2434 | 26 | Nil | 2435 | | | 2436 | 27 | Undefined | 2437 | | | 2438 | 28 | (reserved) | 2439 | | | 2440 | 29 | ad = IEEE 754 Half-Precision Float (16 bits) | 2441 | | | 2442 | 30 | ad = IEEE 754 Single-Precision Float (32 bits) | 2443 | | | 2444 | 31 | ad = IEEE 754 Double-Precision Float (64 bits) | 2445 +--------+------------------------------------------------+ 2447 Table 5: Special Items 2449 D.6.1.3.1. Special Values 2451 False, True, Nil, and Undefined are four special values that can be 2452 encoded in PlanB. False, True, and Nil play the same roles as False, 2453 True, and Null in JSON. A value of Undefined can be used as a 2454 substitute for a data item with an encoding problem, in order to 2455 allow the rest of the enclosing data items to be encoded without 2456 harm. 2458 D.6.1.3.2. Floating Point Values 2460 Floating point values are encoded in argument data of the appropriate 2461 size. PlanB supports 16-bit, 32-bit, and 64-bit IEEE 754 floating 2462 point values. 2464 As a quality of implementation concern, implementations are expected 2465 to use the shortest representation for a floating point value. As 2466 Half-Precision is not yet as universally supported in programming 2467 languages and libraries as the other two sizes, there is less of an 2468 expectation that it will be used. Implementations MAY also replace 2469 floating point value that happen to have zero fractional parts by the 2470 equivalent integer, in particular if the encoding is shorter. (See 2471 also Appendix D.6.3.2.) 2473 D.6.1.3.3. Reserved special item numbers 2475 Some argument information numbers have not been assigned semantics in 2476 this specification. These are registered in an IANA registry and 2477 allocated by Standards Action ([RFC5226]). (TBD: Do we need 2478 experimental values here?) 2480 D.6.1.4. Tagged Items 2482 In PlanB, data items can be "tagged" to give them different or 2483 additional semantics. The tag is an integer number as indicated by 2484 the argument value; the (sole) data item is carried as content data. 2485 If a tag requires structured data, this structure is encoded into the 2486 data item as defined by the tag. 2488 While tagging is a required part of PlanB, the support of any 2489 specific tag is an optional feature (see Appendix D.6.2.2). 2491 D.6.1.5. Byte Strings 2493 The strings transported have application-specific semantics. In a 2494 schemaless environment, the semantics is typically specified by 2495 adjacent items, such as keys in a map. Byte strings are also useful 2496 as data items within tagged items. 2498 D.6.1.6. UTF-8 Strings 2500 The strings transported MUST be UTF-8 strings [RFC3629]. (The 2501 general assumption is that these UTF-8 strings are in Network Unicode 2502 form [RFC5198].) 2504 D.6.1.7. Arrays 2506 Arrays are sequences of data items, which need not all be of the same 2507 type. 2509 D.6.1.8. Maps 2511 The content data for a map (often also called table, dictionary, or 2512 hash, or object in JSON) is comprised of pairs of data items, the 2513 even-numbered ones serving as keys and the following odd-numbered 2514 ones serving as values for this key. 2516 A JSON-like profile may want to restrict keys to UTF-8 strings. 2518 D.6.2. Optional Features 2520 This section describes features that may not be used in every 2521 implementation of PlanB. While it is easy to build a complete PlanB 2522 implementation on a general purpoe computer, more specialized 2523 implementations, such as on constrained nodes, may want to leave off 2524 one or more of the features described here. 2526 D.6.2.1. Backrefs 2528 A backref indicates that a copy of a previously encoded data item is 2529 inserted at its place. A backref with an "ai" value of 0 inserts a 2530 copy of the most recently completed data item, a backref with an "ai" 2531 value of 1 inserts a copy of the data item completed immediately 2532 preceding that, etc. Note that data items can be nested, so the 2533 immediately preceding data item may be part of the content data of 2534 the data item enclosing that, etc. "Completion" in the sense of this 2535 section is inside out, i.e. nested data items complete before their 2536 enclosing data items complete. 2538 A simple implementation of a decoder keeps a pointer to each of the 2539 16 most recently completed items. If a backref is encountered, it 2540 inserts a copy of the respective item. 2542 A backref nominally creates a copy. An implementation can still use 2543 sharing if that does not change the implementation-internal semantics 2544 (e.g., when the decoded value is immutable). Since backrefs can only 2545 refer to completed items, a backref cannot refer to an enclosing data 2546 items, which avoids creating loops. 2548 Backrefs might be encoded in the first 16 positions of the "special" 2549 space, using one byte each then. 2551 (TODO: Insert example here.) 2553 D.6.2.2. Tags 2555 Profiles may be defined that include specific tags defined in the 2556 following list and/or defined by standard action or in the registry. 2558 +------------+---------------------+--------------------------------+ 2559 | tag | data item | semantics | 2560 +------------+---------------------+--------------------------------+ 2561 | 0 | byte string | positive bignum: data as uint | 2562 | | | | 2563 | 1 | byte string | negative bignum: as tag 0, but | 2564 | | | subtracted from -1 | 2565 | | | | 2566 | 2 | unsigned integer | shared backref | 2567 | | | | 2568 | 3 | (see below) | date/time | 2569 | | | | 2570 | 4 | byte string | base64url encoding | 2571 | | | | 2572 | 5 | byte string | base64 encoding | 2573 | | | | 2574 | 6 | byte string | base32 encoding | 2575 | | | | 2576 | 7 | byte string | base32hex encoding | 2577 | | | | 2578 | 8 | byte string | base16 encoding | 2579 | | | | 2580 | 9..23 | (reserved) | allocated by IETF review | 2581 | | | | 2582 | 24..39 | (reserved) | for experimental use only | 2583 | | | | 2584 | 40..283 | (reserved) | allocated by IETF review | 2585 | | | | 2586 | 284.. | (unallocated) | allocated by registry | 2587 +------------+---------------------+--------------------------------+ 2589 D.6.2.2.1. Bignums 2591 Bignums are integers that do not fit into the basic integer 2592 representations provided by mt=000 and mt=001. 2594 As a quality of implementation concern, encoders are expected to use 2595 the minimum number of bytes required to represent the bignum (i.e., 2596 no leading zeroes). However, a decoder MUST be able to decode 2597 bignums with leading zeroes. 2599 D.6.2.2.2. shared backrefs 2601 To enable the representation of data structures with subtree sharing 2602 or with loops, shared backrefs may be employed. The unsigned integer 2603 carried as content operates like a normal backref, except that the 2604 referenced data item is actually shared between the original site and 2605 this site. The backref number counts back from the tag and is not 2606 limited to a value below 16. 2608 D.6.2.2.3. Date/Time 2609 A Date/Time tag may be placed before one of the following kinds of 2610 data items: 2612 +--------------+----------------------------------------------+ 2613 | data item | Semantics | 2614 +--------------+----------------------------------------------+ 2615 | UTF-8 string | Internet date/time [RFC3339] | 2616 | | | 2617 | number | UTC in seconds relative to 1970-01-01T00:00Z | 2618 | | | 2619 | byte string | NTP timestamp: 4, 8, or 16 bytes [RFC5905] | 2620 | | | 2621 | array | see below | 2622 +--------------+----------------------------------------------+ 2624 Note that the number can be fractional and/or negative. The array 2625 contains two numbers, one for seconds and one less than 10**9 for 2626 nanoseconds since 1970-01-01T00:00Z; a third number can be added to 2627 the array to indicate a timezone in minutes east of UTC (the time 2628 itself stays in UTC). If the first number is a floating point 2629 number, the second number MUST be the integer number zero. 2631 A profile may restrict the variations allowed. 2633 D.6.2.2.4. Base Encoding 2635 To encode a UTF-8 string that happens to exactly encode a byte string 2636 in one of the encodings defined in [RFC4648], the equivalent tagged 2637 encoding of that byte string can be used. 2639 For example, to encode the UTF-8 string "Zm9vYmFy", which happens to 2640 be both the base64 and base64url encoding for the bytes "foobar", 2641 either of the following two encodings can be used: 2643 64 86 66 6f 6f 62 61 72 2644 65 86 66 6f 6f 62 61 72 2646 Figure 26 2648 (This is intended to enable JSON roundtripping with reasonably coding 2649 efficiency where large binary values are base-encoded. If PlanB 2650 encoding is used on the entire path, actual base encoding may never 2651 need to occur.) 2653 D.6.2.2.5. Registry of Tags 2654 Tags are the main extension mechanism provided by PlanB. New tags 2655 are registered in an IANA registry. For tags numbered below 284, a 2656 strict allocation policy applies: IETF review [RFC5226]. Within this 2657 space, the tags numbered 24 to 39 are reserved for experimental use. 2659 D.6.3. Discussion 2661 D.6.3.1. Single-Byte Argument Data 2663 The argument value of a single byte of argument data is determined by 2664 adding 28 to the integer value of the argument data byte. This 2665 trades some complexity (adding 28) for some minor enhancement in 2666 coding efficiency (saving a byte for "av" values of 256 to .) 2668 D.6.3.2. Integer vs. Floating Point encoding 2670 For the purposes of this specification, all number representations 2671 are equivalent. This means that an encoder can encode a floating 2672 point value of 0.0 as the integer 0. It, however, also means that an 2673 application that expects to find integer values only might find 2674 floating point values if the encoder decides these are desirable, 2675 e.g. where the floating point value is more compact than a 64-bit 2676 integer. 2678 An integer-only profile may want to exclude the use of floating point 2679 values. 2681 A compact profile may want to exclude integer encodings that are 2682 longer than necessary, e.g. to save the need to implement 64-bit 2683 integers. 2685 D.6.3.3. Backrefs 2687 The Backref feature trades nontrivial complexity for coding 2688 efficiency, which may be non-trivial as well in many applications. 2690 A no-backrefs profile may support an applications that need to 2691 minimize implementation or CPU complexity at the cost of potentially 2692 increased encoded size. 2694 A typical implementation of the backref feature in an encoder will 2695 keep a sliding window of 16 previous data items. 2697 D.6.4. Examples 2699 (TODO) 2701 D.6.5. Acknowledgements 2703 (This subsection needs a proper acknowledgements subsubsection.) 2705 The encoding of the argument information in PlanB was inspired by the 2706 encoding of length information designed by Klaus Hartke for CoAP. 2708 Authors' Addresses 2710 Carsten Bormann 2711 Universitaet Bremen TZI 2712 Postfach 330440 2713 Bremen D-28359 2714 Germany 2716 Phone: +49-421-218-63921 2717 Email: cabo@tzi.org 2719 Klaus Hartke 2720 Universitaet Bremen TZI 2721 Postfach 330440 2722 Bremen D-28359 2723 Germany 2725 Phone: +49-421-218-63905 2726 Email: hartke@tzi.org