idnits 2.17.1 draft-bormann-coap-misc-20.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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 1451 has weird spacing: '... code mid...' -- The document date (July 31, 2012) is 4279 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 1009, but not defined -- Looks like a reference, but probably isn't: '0' on line 1810 == Unused Reference: 'CoRE201' is defined on line 1075, but no explicit reference was found in the text == Outdated reference: A later version (-18) exists of draft-ietf-core-coap-11 == Outdated reference: A later version (-16) exists of draft-ietf-core-observe-05 == Outdated reference: A later version (-26) exists of draft-ietf-httpbis-p1-messaging-20 == Outdated reference: A later version (-26) exists of draft-ietf-httpbis-p4-conditional-20 == Outdated reference: A later version (-26) exists of draft-ietf-httpbis-p6-cache-20 ** Obsolete normative reference: RFC 2616 (Obsoleted by RFC 7230, RFC 7231, RFC 7232, RFC 7233, RFC 7234, RFC 7235) -- Obsolete informational reference (is this intentional?): RFC 5226 (Obsoleted by RFC 8126) Summary: 1 error (**), 0 flaws (~~), 9 warnings (==), 4 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: February 1, 2013 July 31, 2012 7 Miscellaneous additions to CoAP 8 draft-bormann-coap-misc-20 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 February 1, 2013. 35 Copyright Notice 37 Copyright (c) 2012 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 . . . . . . . . . . . . . . . . . . . . . . . . . 4 53 2. Getting rid of artificial limitations . . . . . . . . . . . . 5 54 2.1. Option Length encoding beyond 270 bytes . . . . . . . . . 5 55 3. Registered Option . . . . . . . . . . . . . . . . . . . . . . 8 56 3.1. A Separate Suboption Number Space . . . . . . . . . . . . 8 57 3.2. Opening Up the Option Number Space . . . . . . . . . . . . 9 58 3.2.1. Long Jump construct . . . . . . . . . . . . . . . . . 10 59 3.2.2. Discussion . . . . . . . . . . . . . . . . . . . . . . 11 60 3.2.3. Example . . . . . . . . . . . . . . . . . . . . . . . 12 61 3.2.4. IANA considerations . . . . . . . . . . . . . . . . . 12 62 4. Patience, Leisure, and Pledge . . . . . . . . . . . . . . . . 13 63 4.1. Patience . . . . . . . . . . . . . . . . . . . . . . . . . 13 64 4.2. Leisure . . . . . . . . . . . . . . . . . . . . . . . . . 13 65 4.3. Pledge . . . . . . . . . . . . . . . . . . . . . . . . . . 14 66 4.4. Option Formats . . . . . . . . . . . . . . . . . . . . . . 14 67 5. Observing Resources in CoAP . . . . . . . . . . . . . . . . . 16 68 6. CONNECT . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 69 6.1. Requesting a Tunnel with CONNECT . . . . . . . . . . . . . 19 70 6.2. Using a CONNECT Tunnel . . . . . . . . . . . . . . . . . . 19 71 6.3. Closing down a CONNECT Tunnel . . . . . . . . . . . . . . 20 72 7. Enabling Protocol Evolution . . . . . . . . . . . . . . . . . 21 73 7.1. Potential new option number allocation . . . . . . . . . . 22 74 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 24 75 9. Security Considerations . . . . . . . . . . . . . . . . . . . 25 76 10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 26 77 11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 27 78 11.1. Normative References . . . . . . . . . . . . . . . . . . . 27 79 11.2. Informative References . . . . . . . . . . . . . . . . . . 27 80 Appendix A. The Nursery (Things that still need to ripen a 81 bit) . . . . . . . . . . . . . . . . . . . . . . . . 29 82 A.1. Envelope Options . . . . . . . . . . . . . . . . . . . . . 29 83 A.2. Payload-Length Option . . . . . . . . . . . . . . . . . . 30 84 A.3. URI Authorities with Binary Adresses . . . . . . . . . . . 30 85 A.4. Length-aware number encoding (o256) . . . . . . . . . . . 31 86 A.5. SMS encoding . . . . . . . . . . . . . . . . . . . . . . . 33 87 A.5.1. ASCII-optimized SMS encoding . . . . . . . . . . . . . 34 88 Appendix B. The Cemetery (Things we won't do) . . . . . . . . . . 37 89 B.1. Example envelope option: solving #230 . . . . . . . . . . 37 90 B.2. Example envelope option: proxy-elective options . . . . . 38 91 B.3. Stateful URI compression . . . . . . . . . . . . . . . . . 38 92 B.4. Beyond 270 bytes in a single option . . . . . . . . . . . 39 93 B.5. Beyond 15 options . . . . . . . . . . . . . . . . . . . . 40 94 B.5.1. Implementation considerations . . . . . . . . . . . . 42 95 B.5.2. What should we do now? . . . . . . . . . . . . . . . . 43 96 B.5.3. Alternatives . . . . . . . . . . . . . . . . . . . . . 43 97 B.5.4. Alternative: Going to a delimiter model . . . . . . . 43 99 B.6. Implementing the option delimiter for 15 or more 100 options . . . . . . . . . . . . . . . . . . . . . . . . . 43 101 Appendix C. Experimental Options . . . . . . . . . . . . . . . . 45 102 C.1. Options indicating absolute time . . . . . . . . . . . . . 45 103 C.2. Representing Durations . . . . . . . . . . . . . . . . . . 46 104 C.3. Rationale . . . . . . . . . . . . . . . . . . . . . . . . 47 105 C.4. Pseudo-Floating Point . . . . . . . . . . . . . . . . . . 48 106 C.5. A Duration Type for CoAP . . . . . . . . . . . . . . . . . 49 107 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 56 109 1. Introduction 111 The CoRE WG is tasked with standardizing an Application Protocol for 112 Constrained Networks/Nodes, CoAP [I-D.ietf-core-coap]. This protocol 113 is intended to provide RESTful [REST] services not unlike HTTP 114 [RFC2616], while reducing the complexity of implementation as well as 115 the size of packets exchanged in order to make these services useful 116 in a highly constrained network of themselves highly constrained 117 nodes. 119 This objective requires restraint in a number of sometimes 120 conflicting ways: 122 o reducing implementation complexity in order to minimize code size, 124 o reducing message sizes in order to minimize the number of 125 fragments needed for each message (in turn to maximize the 126 probability of delivery of the message), the amount of 127 transmission power needed and the loading of the limited-bandwidth 128 channel, 130 o reducing requirements on the environment such as stable storage, 131 good sources of randomness or user interaction capabilities. 133 This draft attempts to address a number of problems not yet 134 adequately solved in [I-D.ietf-core-coap]. The solutions proposed to 135 these problems are somewhat interrelated and are therefore presented 136 in one draft. 138 The appendix contains the "CoAP cemetery" (possibly later to move 139 into its own draft), documenting roads that the WG decided not to 140 take, in order to spare readers from reinventing them in vain. 142 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 143 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 144 document are to be interpreted as described in [RFC2119]. 146 The term "byte" is used in its now customary sense as a synonym for 147 "octet". 149 2. Getting rid of artificial limitations 151 _Artificial limitations_ are limitations of a protocol or system that 152 are not rooted in limitations of actual capabilities, but in 153 arbitrary design decisions. Proper system design tries to avoid 154 artificial limitations, as these tend to cause complexity in systems 155 that need to work with these limitations. 157 E.g., the original UNIX filesystem had an artificial limitation of 158 the length of a path name component to 14 bytes. This led to a 159 cascade of workarounds in programs that manipulate file names: E.g., 160 systematically replacing a ".el" extension in a filename with a 161 ".elc" for the compiled file might exceed the limit, so all ".el" 162 files were suddenly limited to 13-byte filenames. 164 Note that, today, there still is a limitation in most file system 165 implementations, typically at 255. This just happens to be high 166 enough to rarely be of real-world concern; we will refer to this case 167 as a "painless" artificial limitation. 169 CoAP-08 had two highly recognizable artificial limitations in its 170 protocol encoding 172 o The number of options in a single message is limited to 15 max. 174 o The length of an option is limited to 270 max. 176 It has been argued that the latter limitation causes few problems, 177 just as the 255-byte path name component limitation in filenames 178 today causes few problems. Appendix B.4 provided a design to extend 179 this; as a precaution to future extensions of this kind, the current 180 encoding for length 270 (eight ones in the extension byte) could be 181 marked as reserved today. Since, Matthias Kovatsch has proposed a 182 simpler scheme that seems to gain favor in the WG, see Section 2.1. 184 The former limitation has been solved in CoAP-09. A historical 185 discussion of other approaches for going beyond 15 options is in 186 Appendix B.5. Appendix B.6 discusses implementation. 188 2.1. Option Length encoding beyond 270 bytes 190 For option lengths beyond 270 bytes, we reserve the value 255 of an 191 extension byte to mean "add 255, read another extension byte" 192 Figure 1. While this causes the length of the option header to grow 193 linearly with the size of the option value, only 0.4 % of that size 194 is used. With a focus on short options, this encoding is justified. 196 for 15..269: 197 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 198 | Option Delta | 1 1 1 1 | Length - 15 | 199 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 200 | Option Value ... 201 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 203 for 270..524: 204 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 205 | Option Delta | 1 1 1 1 | 1 1 1 1 1 1 1 1 | 206 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 207 | Length - 270 | Option Value ... 208 +---+---+---+---+---+---+---+---+ 209 | 210 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 212 for 525..779: 213 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 214 | Option Delta | 1 1 1 1 | 1 1 1 1 1 1 1 1 | 215 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 216 | 1 1 1 1 1 1 1 1 | Length - 525 | 217 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 218 | Option Value ... 219 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 221 for 780..1034: 222 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 223 | Option Delta | 1 1 1 1 | 1 1 1 1 1 1 1 1 | 224 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 225 | 1 1 1 1 1 1 1 1 | 1 1 1 1 1 1 1 1 | 226 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 227 | Length - 780 | Option Value ... 228 +---+---+---+---+---+---+---+---+ 229 | 230 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 232 Figure 1: Options beyond 270 bytes 234 Options that are longer than 1034 bytes MUST NOT be sent; an option 235 that has 255 (all one bits) in the field called "Length - 780" MUST 236 be rejected upon reception as an invalid option. 238 In the process, the maximum length of all options that are currently 239 set at 270 should now be set to a carefully chosen value. With the 240 purely encoding-based limit gone, Uri-Proxy should now be restored to 241 be a non-repeatable option. 243 A first proposal for a new set of per-option length restrictions 244 follows: 246 +--------+---------------------+-----+------+--------+--------+ 247 | number | name | min | max | type | repeat | 248 +--------+---------------------+-----+------+--------+--------+ 249 | 1 | content_type | 0 | 2 | uint | - | 250 | | | | | | | 251 | 2 | max_age | 0 | 4 | uint | - | 252 | | | | | | | 253 | 3 | proxy_uri | 1 | 1023 | string | - | 254 | | | | | | | 255 | 4 | etag | 1 | 8 | opaque | yes | 256 | | | | | | | 257 | 5 | uri_host | 1 | 255 | string | - | 258 | | | | | | | 259 | 6 | location_path | 0 | 255 | string | yes | 260 | | | | | | | 261 | 7 | uri_port | 0 | 2 | uint | - | 262 | | | | | | | 263 | 8 | location_query | 0 | 255 | string | yes | 264 | | | | | | | 265 | 9 | uri_path | 0 | 255 | string | yes | 266 | | | | | | | 267 | 10 | observe | 0 | 2 | uint | - | 268 | | | | | | | 269 | 11 | token | 1 | 8 | opaque | - | 270 | | | | | | | 271 | 12 | accept | 0 | 2 | uint | yes | 272 | | | | | | | 273 | 13 | if_match | 0 | 8 | opaque | yes | 274 | | | | | | | 275 | 14 | registered_elective | 1 | 1023 | opaque | yes | 276 | | | | | | | 277 | 15 | uri_query | 1 | 255 | string | yes | 278 | | | | | | | 279 | 17 | block2 | 0 | 3 | uint | - | 280 | | | | | | | 281 | 18 | size | 0 | 4 | uint | - | 282 | | | | | | | 283 | 19 | block1 | 0 | 3 | uint | - | 284 | | | | | | | 285 | 21 | if_none_match | 0 | 0 | empty | - | 286 | | | | | | | 287 | 25 | registered_critical | 1 | 1023 | opaque | yes | 288 +--------+---------------------+-----+------+--------+--------+ 290 (Option 14 with a length of 0 is a fencepost only.) 292 3. Registered Option 294 CoAP's option encoding is highly efficient, but works best with small 295 option numbers that do not require much fenceposting. The CoAP 296 Option Number Registry therefore has a relatively heavyweight 297 registration requirement: "IETF Review" as described in [RFC5226]. 299 However, there is also considerable benefit in a much looser registry 300 policy, enabling a first-come-first-served policy for a relatively 301 large option number space. 303 Here, we discuss two solutions that enable such a registry. One is 304 to define a separate mechanism for registered options, discussed in 305 Section 3.1. Alternatively, we could make it easier to use a larger 306 main option number space, discussed in Section 3.2. 308 3.1. A Separate Suboption Number Space 310 This alternative defines a separate space of suboption numbers, with 311 an expert review [RFC5226] (or even first-come-first-served) 312 registration policy. If expert review is selected for this registry, 313 it would be with a relatively loose policy delegated to the expert. 314 This draft proposes leaving the registered suboption numbers 0-127 to 315 expert review with a policy that mainly focuses on the availability 316 of a specification, and 128-16383 for first-come-first-served where 317 essentially only a name is defined. 319 The "registered" options are used in conjunction with this suboption 320 number registry. They use two normal CoAP option numbers, one for 321 options with elective semantics (Registered-Elective) and one for 322 options with critical semantics (Registered-Critical). The suboption 323 numbers are not separate, i.e. one registered suboption number might 324 have some elective semantics and some other critical semantics (e.g., 325 for the request and the response leg of an exchange). The option 326 value starts with an SDNV [RFC6256] of the registered suboption 327 number. (Note that there is no need for an implementation to 328 understand SDNVs, it can treat the prefixes as opaque. One could 329 consider the SDNVs as a suboption prefix allocation guideline for 330 IANA as opposed to a number encoding.) 332 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 333 |1 0 0 0 0 0 0 1|0 1 1 1 0 0 1 1| value... | 334 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 335 \___SDNV of registered number___/ 337 Figure 2: Example option value for registered option 339 Note that a Registered Option cannot be empty, because there would be 340 no space for the SDNV. Also, the empty option 14 is reserved for 341 fenceposting ([I-D.ietf-core-coap], section 3.2). (Obviously, once a 342 Registered-Elective Option is in use, there is never a need for a 343 fence-post for option number 14.) 345 The Registered-Elective and Registered-Critical Options are 346 repeatable. 348 +-----+----------+---------------------+---------+--------+---------+ 349 | No. | C/E | Name | Format | Length | Default | 350 +-----+----------+---------------------+---------+--------+---------+ 351 | 14 | Elective | Registered-Elective | (see | 1-1023 | (none) | 352 | | | | above) | B | | 353 | | | | | | | 354 | 25 | Critical | Registered-Critical | (see | 1-1023 | (none) | 355 | | | | above) | B | | 356 +-----+----------+---------------------+---------+--------+---------+ 358 This solves CoRE issue #214 [CoRE214]. (How many options we need 359 will depend on the resolution of #241 [CoRE241].) 361 3.2. Opening Up the Option Number Space 363 The disadvantage of the registered-... options is that there is a 364 significant syntactic difference between options making use of this 365 space and the usual standard options. This creates a problem not 366 unlike that decried in [RFC6648]. 368 The alternative discussed in this section reduces the distance by 369 opening up the main Option number space instead. 371 There is still a significant incentive to use low-numbered Options. 372 However, the proposal reduces the penalty for using a high-numbered 373 Option to two or three bytes. More importantly, using a cluster of 374 related high-numbered options only carries a total penalty of two or 375 three bytes. 377 The main reason high-numbered options are expensive to use and thus 378 the total space is relatively limited is that the option delta 379 mechanism only allows increasing the current option number by up to 380 14 per one-byte fencepost. To use, e.g., Option number 1234 together 381 with the usual set of low-numbered Options, one needs to insert 88 382 fence-post bytes. This is prohibitive. 384 Enabling first-come-first-served probably requires easily addressing 385 a 16-bit option number space, with some potential increase later in 386 the lifetime of the protocol (say, 10 to 15 years from now). 388 To enable the use of large option numbers, one needs a way to advance 389 the Option number in bigger steps than possible by the Option Delta. 390 So we propose a new construct, the Long Jump construct, to move the 391 Option number forward. 393 3.2.1. Long Jump construct 395 The following construct can occur in front of any Option: 397 0 1 2 3 4 5 6 7 398 +---+---+---+---+---+---+---+---+ 399 | 1 1 1 1 | 0 0 0 1 | 0xf1 (Delta = 15) 400 +---+---+---+---+---+---+---+---+ 402 0 1 2 3 4 5 6 7 403 +---+---+---+---+---+---+---+---+ 404 | 1 1 1 1 | 0 0 1 0 | 0xf2 405 +---+---+---+---+---+---+---+---+ 406 | Long Jump Value | (Delta/8)-2 407 +---+---+---+---+---+---+---+---+ 409 0 1 2 3 4 5 6 7 410 +---+---+---+---+---+---+---+---+ 411 | 1 1 1 1 | 0 0 1 1 | 0xf3 412 +---+---+---+---+---+---+---+---+ 413 | Long Jump Value, MSB | 414 +---+---+---+---+---+---+---+---+ (Delta/8)-258 415 | Long Jump Value, LSB | 416 +---+---+---+---+---+---+---+---+ 418 Figure 3: Long Jump Format 420 This construct is not by itself an Option. It can occur in front of 421 any Option to increase the current Option number that then goes into 422 its Option number calculation. The increase is done in multiples of 423 eight. More specifically, the actual addition to the current Option 424 number is computed as follows: 426 Delta = ((Long Jump Value) + N) * 8 428 where N is 2 for the one-byte version and N is 258 for the two-byte 429 version. 431 A Long Jump MUST be followed by an actual Option, i.e., it MUST NOT 432 be followed by another Long Jump or an end-of-options indicator. A 433 message violating this MUST be rejected as malformed. 435 Long Jumps do NOT count as Options in the Option Count field of the 436 header (i.e., they cannot by themselves end the Option sequence). 438 3.2.2. Discussion 440 Adding a mechanism at this late stage creates concerns of backwards 441 compatibility. A message sender never needs to implement long-jumps 442 unless it wants to make use of a high-numbered option. So this 443 mechanism can be added once a high-numbered option is added. A 444 message receiver, though, would more or less unconditionally have to 445 implement the mechanism, leading to unconditional additional 446 complexity. There are good reasons to minimize this, as follows: 448 o The increase in multiples of eight allows looking at an option and 449 finding out whether it is critical or not even if the Long Jump 450 value has just been skipped (as opposed to having been processed 451 fully). (It also allows accessing up to approximately 2048 452 options with a two-byte Long Jump.) This allows a basic 453 implementation that does not implement any high-numbered options 454 to simply ignore long jumps and any elective options behind them, 455 while still properly reacting to critical options. 457 o There is probably a good reason to disallow long-jumps that lead 458 to an option number of 42 and less, enabling simple receivers to 459 do the above simplification. 461 o It might seem obvious to remove the fenceposting mechanism 462 altogether in favor of long jumps. This is not advisable: 463 Fenceposting already has zero implementation effort at the 464 receiver, and the overhead at the sender is very limited (it is 465 just a third kind of jump, at one byte per jump). Beyond 42, 466 senders can ignore the existence of fenceposts if they want 467 (possibly obviating the need for more complex base-14 arithmetic). 469 There is no need for a finer granularity than 8, as the Option 470 construct following can also specify a Delta of 0..14. (A 471 granularity of 16 will require additional fenceposting where an 472 option delta of 15 would happen to be required otherwise, which we 473 have reserved. It can be argued that 16 is still the better choice, 474 as fenceposting is already in the code path.) 476 The Long Jump construct takes 0xf1 and 0xf2 from the space available 477 for initial bytes of Options. (Note that we previously took 0xf0 to 478 indicate end-of-options for OC=15.) 480 Varying N with the length as defined above makes it unambiguous 481 whether a one- or two-byte Long Jump is to be used. Setting N=2 for 482 the one-byte version makes it clear that a Delta of 8 is to be 483 handled the usual way (i.e., by Option Delta itself and/or 484 fenceposting). If the delta is not small and not 7 modulo 8, there 485 is still a choice between using the smaller multiple of 8 and a 486 larger Delta in the actual Option or v.v., this biases the choice 487 towards a larger Long Jump and a smaller following Delta, which is 488 also easier to implement as it reduces the number of choice points. 490 3.2.3. Example 492 The following sequence of bytes would encode a Uri-Path Option of 493 "foo" followed by Options 1357 (value "bar") and 1360 (value "baz"): 495 93 65 6f 6f Option 9 (0 + 9, "foo") 496 f1 a6 Long Jump by 1344 497 43 62 61 72 Option 1357 (9 + 1344 + 4, "bar") 498 33 62 61 7a Option 1360 (1357 + 3, "baz") 500 Figure 4: Example using a Long Jump construct 502 where f1 a6 is the long jump forward by (0xa6+2)*8=1344 option 503 numbers. The total option count (OC) for the CoAP header is 3. Note 504 that even if f1 a6 is skipped, the 1357 (which then appears as an 505 Option number 13) is clearly visible as Critical. 507 3.2.4. IANA considerations 509 With the scheme proposed above, we could have three tiers of Option 510 Numbers: 512 +---------------+-------------------------+ 513 | Option Number | Policy [RFC5226] | 514 +---------------+-------------------------+ 515 | 0..255 | Standards Action | 516 | | | 517 | 256..2047 | Designated Expert | 518 | | | 519 | 2048..65535 | First Come First Served | 520 +---------------+-------------------------+ 522 For the inventor of a new option, this would provide a small 523 incentive to go through the designated expert for some minimal cross- 524 checking in order to be able to use the two-byte long-jump. 526 4. Patience, Leisure, and Pledge 528 A number of options might be useful for controlling the timing of 529 interactions. 531 (This section also addresses core-coap ticket #177.) 533 4.1. Patience 535 A client may have a limited time period in which it can actually make 536 use of the response for a request. Using the Patience option, it can 537 provide an (elective) indication how much time it is willing to wait 538 for the response from the server, giving the server license to ignore 539 or reject the request if it cannot fulfill it in this period. 541 If the server knows early that it cannot fulfill the request in the 542 time requested, it MAY indicate this with a 5.04 "Timeout" response. 543 For non-safe methods (such as PUT, POST, DELETE), the server SHOULD 544 indicate whether it has fulfilled the request by either responding 545 with 5.04 "Timeout" (and not further processing the request) or by 546 processing the request normally. 548 Note that the value of the Patience option should be chosen such that 549 the client will be able to make use of the result even in the 550 presence of the expected network delays for the request and the 551 response. Similarly, when a proxy receives a request with a Patience 552 option and cannot fulfill that request from its cache, it may want to 553 adjust the value of the option before forwarding it to an upstream 554 server. 556 (TBD: The various cases that arise when combining Patience with 557 Observe.) 559 The Patience option is elective. Hence, a client MUST be prepared to 560 receive a normal response even after the chosen Patience period (plus 561 an allowance for network delays) has elapsed. 563 4.2. Leisure 565 Servers generally will compute an internal value that we will call 566 Leisure, which indicates the period of time that will be used for 567 responding to a request. A Patience option, if present, can be used 568 as an upper bound for the Leisure. Leisure may be non-zero for 569 congestion control reasons, in particular for responses to multicast 570 requests. For these, the server should have a group size estimate G, 571 a target rate R (which both should be chosen conservatively) and an 572 estimated response size S; a rough lower bound for Leisure can then 573 be computed as follows: 575 lb_Leisure = S * G / R 577 Figure 5: Computing a lower bound for the Leisure 579 E.g., for a multicast request with link-local scope on an 2.4 GHz 580 IEEE 802.15.4 (6LoWPAN) network, G could be (relatively 581 conservatively) set to 100, S to 100 bytes, and the target rate to 8 582 kbit/s = 1 kB/s. The resulting lower bound for the Leisure is 10 583 seconds. 585 To avoid response implosion, responses to multicast requests SHOULD 586 be dithered within a Leisure period chosen by the server to fall 587 between these two bounds. 589 Currently, we don't foresee a need to signal a value for Leisure from 590 client to server (beyond the signalling provided by Patience) or from 591 server to client, but an appropriate Option might be added later. 593 4.3. Pledge 595 In a basic observation relationship [I-D.ietf-core-observe], the 596 server makes a pledge to keep the client in the observation 597 relationship for a resource at least until the max-age for the 598 resource is reached. 600 To save the client some effort in re-establishing observation 601 relationships each time max-age is reached, the server MAY want to 602 extend its pledge beyond the end of max-age by signalling in a 603 response/notification an additional time period using the Pledge 604 Option, in parallel to the Observe Option. 606 The Pledge Option MUST NOT be used unless the server can make a 607 reasonable promise not to lose the observation relationship in this 608 time frame. 610 Currently, we don't foresee a need to signal a value for Pledge from 611 client to server, but an appropriate behavior might be added later 612 for this option when sent in a request. 614 4.4. Option Formats 616 +-----+----------+----------+-----------------+--------+---------+ 617 | No. | C/E | Name | Format | Length | Default | 618 +-----+----------+----------+-----------------+--------+---------+ 619 | 22 | Elective | Patience | Duration in mis | 1 B | (none) | 620 | | | | | | | 621 | 24 | Elective | Pledge | Duration in s | 1 B | 0 | 622 +-----+----------+----------+-----------------+--------+---------+ 624 All timing options use the Duration data type (see Appendix C.2), 625 however Patience (and Leisure, if that ever becomes an option) uses a 626 timebase of mibiseconds (mis = 1/1024 s) instead of seconds. (This 627 reduces the range of the Duration from ~ 91 days to 128 minutes.) 629 Implementation note: As there are no strong accuracy requirements on 630 the clocks employed, making use of any existing time base of 631 milliseconds is a valid implementation approach (2.4 % off). 633 None of the options may be repeated. 635 5. Observing Resources in CoAP 637 (Co-Author for this section: Matthias Kovatsch) 639 There are two open issues related to -observe 640 [I-D.ietf-core-observe]: 642 o mixing freshness and observation lifetime, and 644 o non-cacheable resources. 646 To solve the first issue, we think that -observe should be clarified 647 as follows: 649 A server sends at least some notifications as confirmable messages. 650 Each confirmable notification is an opportunity for the server to 651 check if the client is still there. If the client acknowledges the 652 notification, it is assumed to be well and alive and still interested 653 in the resource. If it rejects the message with a reset message or 654 if it doesn't respond, it is assumed not longer to be interested and 655 is removed from the list of observers. So an observation 656 relationship can potentially go on forever, if the client 657 acknowledges each confirmable notification. If the server doesn't 658 send a notification for a while and wants to check if the client is 659 still there, it may send a confirmable notification with the current 660 resource state to check that. 662 So there is no mixing of freshness and lifetime going on. 664 The other issue is a bit less trivial to solve. The problem is that 665 normal CoAP and -observe actually have very different freshness 666 models: 668 Normally, when a client wants to know the current state of a 669 resource, it retrieves a representation, uses it and stores it in its 670 cache. Later, when it wants to know the current state again, it can 671 either use the stored representation provided that it's still fresh, 672 or retrieve a new representation, use it and store it in its cache. 674 If a server knows when the state of the resource will change the next 675 time, it can set the Max-Age of the representation to an accurate 676 time span. So the change of the resource state will coincide with 677 the expiration of the freshness of the representation stored in the 678 client's cache (ignoring network latency). 680 But if the resource changes its state unpredictably at any time, the 681 server can set the Max-Age only to an estimate. If the state then 682 actually changes before the freshness expires, the client wrongly 683 believes it has fresh information. Conversely, if the freshness 684 expires and the client wants to know the current state, the client 685 wrongly believes it has to make a new request although the 686 representation is actually still fresh (this is defused by ETag 687 validation). 689 -observe doesn't have these kinds of problems: the server does not 690 have to predict when the resource will change its state the next 691 time. It just sends a notification when it does. The new 692 representation invalidates the old representation stored in the 693 client's cache. So the client always has a fresh representation that 694 it can use when it wants to know the current resource state without 695 ever having to make a request. An explicit Max-Age is not needed for 696 determining freshness. 698 But -observe has a different set of problems: 700 The first problem is that the resource may change its state more 701 often than there is bandwidth available or the client can handle. 702 Thus, -observe cannot make any guarantee that a client will see every 703 state change. The solution is that -observe guarantees that the 704 client will eventually see the latest state change, and follows a 705 best effort approach to enable the client to see as many state 706 changes as possible. 708 The second problem is that, when a notification doesn't arrive for a 709 while, the client does not know if the resource did not change its 710 state or if the server lost its state and forgot that the client is 711 interested in the resource. We propose the following solution: With 712 each notification that the server sends, it makes a promise to send 713 another notification, and that it will send this next notification at 714 latest after a certain time span. This time span is included with 715 each notification. So when no notification arrives for a while and 716 the time span has not expired yet, the client assumes that the 717 resource did not change its state. If the time span has expired, no 718 notification has arrived and the client wants to know the current 719 state of the resource, it has to make a new request. 721 The third problem is that, when an intermediary is observing a 722 resource and wants to create a response from a representation stored 723 in its cache, it needs to specify a Max-Age. But the intermediary 724 cannot predict when it will receive the next notification, because 725 the next notification can arrive at any time. Unlike the origin 726 server, it also doesn't have the application-specific knowledge that 727 the origin server has. We propose the following solution: With each 728 notification a server sends, it includes a value that an intermediary 729 should use to calculate the Max-Age. 731 To summarize: 733 o A notification doesn't have a Max-Age; it's fresh until the next 734 notification arrives. A notification is the promise for another 735 notification that will arrive at latest after Next-Notification- 736 At-Latest. This value is included with every notification. The 737 promise includes that the server attempts to transmit a 738 notification to the client for the promised time span, even if the 739 client does not seem to respond, e.g., due to a temporary network 740 outage. 742 o A notification also contains another value, called Max-Age-Hint. 743 This value is used by a cache to calculate a Max-Age for the 744 representation if needed. In a cache, the Max-Age-Hint of a 745 representation is counted down like Max-Age. When it reaches 746 zero, however, the representation can be still used to satisfy 747 requests, but is non-cacheable (i.e., Max-Age is 0). The Max-Age- 748 Hint must be less than or equal to Next-Notification-At-Latest. 750 We see two possible ways to encode Next-Notification-At-Latest and 751 Max-Age-Hint in a message: 753 o The first way is to require the values of Next-Notification-At- 754 Latest and Max-Age-Hint to be the same, although they are 755 conceptually unrelated. Then, a single option in the message can 756 be used to hold both values. 758 o The second way is to include two options, one for Next- 759 Notification-At-Latest and one for Max-Age-Hint. Since Next- 760 Notification-At-Latest is less than or equal to Max-Age-Hint, the 761 first option should indicates Max-Age-Hint, and the second option 762 Next-Notification-At-Latest minus Max-Age-Hint with a default 763 value of 0. 765 6. CONNECT 767 [RFC2817] defines the HTTP CONNECT method to establish a TCP tunnel 768 through a proxy so that end-to-end TLS connections can be made 769 through the proxy. Recently, a requirement for similar functionality 770 has been discussed for CoAP. This section defines a straw-man 771 CONNECT method and related methods and response codes for CoAP. 773 (IANA considerations for this section TBD.) 775 6.1. Requesting a Tunnel with CONNECT 777 CONNECT is allocated as a new method code in the "CoAP Method Codes" 778 registry. When a client makes a CONNECT request to an intermediary, 779 the intermediary evaluates the Uri-Host, Uri-Port, and/or the 780 authority part of the Proxy-Uri Options in a way that is defined by 781 the security policy of the intermediary. If the security policy 782 allows the allocation of a tunnel based on these parameters, the 783 method returns an empty payload and a response code of 2.30 Tunnel 784 Established. Other possible response codes include 4.03 Forbidden. 786 It may be the case that the intermediary itself can only reach the 787 requested origin server through another intermediary. In this case, 788 the first intermediary SHOULD make a CONNECT request of that next 789 intermediary, requesting a tunnel to the authority. A proxy MUST NOT 790 respond with any 2.xx status code unless it has either a direct or 791 tunnel connection established to the authority. 793 An origin server which receives a CONNECT request for itself MAY 794 respond with a 2.xx status code to indicate that a tunnel is 795 established to itself. 797 Code 2.30 "Tunnel Established" is allocated as a new response code in 798 the "CoAP Response Codes" registry. 800 6.2. Using a CONNECT Tunnel 802 Any successful (2.xx) response to a CONNECT request indicates that 803 the intermediary has established a tunnel to the requested host and 804 port. The tunnel is bound to the requesting end-point and the Token 805 supplied in the request (as always, the default Token is admissible). 806 The tunnel can be used by the client by making a DATAGRAM request. 808 DATAGRAM is allocated as a new method code in the "CoAP Method Codes" 809 registry. When a client makes a DATAGRAM request to an intermediary, 810 the intermediary looks up the tunnel bound to the client end-point 811 and Token supplied in the DATAGRAM request (no other Options are 812 permitted). If a tunnel is found and the intermediary's security 813 policy permits, the intermediary forwards the payload of the DATAGRAM 814 request as the UDP payload towards the host and port established for 815 the tunnel. No response is defined for this request (note that the 816 request can be given as a CON or NON request; for CON, there will be 817 an ACK on the message layer if the tunnel exists). 819 The security policy on the intermediary may restrict the allowable 820 payloads based on its security policy, possibly considering host and 821 port. An inadmissible payload SHOULD cause a 4.03 Forbidden response 822 with a diagnostic message as payload. 824 The UDP payload of any datagram received from the tunnel and admitted 825 by the security policy is forwarded to the client as the payload of a 826 2.31 "Datagram Received" response. The response does not carry any 827 Option except for Token, which identifies the tunnel towards the 828 client. 830 Code 2.31 "Datagram Received" is allocated as a new response code in 831 the "CoAP Response Codes" registry. 833 An origin server that has established a tunnel to itself processes 834 the CoAP payloads of related DATAGRAM requests as it would process an 835 incoming UDP payload, and forwards what would be outgoing UDP 836 payloads in 2.31 "Datagram Received" responses. 838 6.3. Closing down a CONNECT Tunnel 840 A 2.31 "Datagram Received" response may be replied to with a RST, 841 which closes down the tunnel. Similarly, the Token used in the 842 tunnel may be reused by the client for a different purpose, which 843 also closes down the tunnel. 845 7. Enabling Protocol Evolution 847 To enable a protocol to evolve, it is critical that new capabilities 848 can be introduced without requiring changes in components that don't 849 really care about the capability. One such probem is exhibited by 850 CoAP options: If a proxy does not understand an elective option in a 851 request, it will not be able to forward it to the origin server, 852 rendering the new option ineffectual. Worse, if a proxy does not 853 understand a critical option in a request, it will not be able to 854 operate on the request, rendering the new option damaging. 856 As a conclusion to the Ticket #230 discussion in the June 4th interim 857 call, we decided to solve the identification of options that a proxy 858 can safely forward even if not understood (previously called Proxy- 859 Elective). 861 The proposal is to encode this information in the option number, just 862 like the way the information that an option is critical is encoded 863 now. This leads to two bits with semantics: the lowest bit continues 864 to be the critical bit, and the next higher bit is now the "unsafe" 865 bit (i.e., this option is not safe to forward unless understood by 866 the proxy). 868 Another consideration (for options that are not unsafe to forward) is 869 whether the option should serve as a cache key in a request. HTTP 870 has a vary header that indicates in the response which header fields 871 were considered by the origin server to be cache keys. In order to 872 avoid this complexity, we should be able to indicate this information 873 right in the option number. However, reserving another bit is 874 wasteful, in particular as there are few safe-to-forward options that 875 are not cache-keys. 877 Therefore, we propose the following bit allocation in an option 878 number: 880 xxx nnn UC 882 Figure 6 884 (where xxx is a variable length prefix, as option numbers are not 885 bounded upwards). UC is the unsafe and critical bits. For U=0 only, 886 if nnn is equal to 111 binary, the option does not serve as a cache 887 key (for U=1, the proxy has to know the option to act on it, so there 888 is no point in indicating whether it is a cache key). There is no 889 semantic meaning of xxx. 891 Note that clients and servers are generally not interested in this 892 information. A proxy may use an equivalent of the following C code 893 to derive the characteristics of an option number "onum": 895 Critical = (onum & 1); 896 UnSafe = (onum & 2); 897 NoCache = ((onum & 0x1e) == 0x1c); 899 Figure 7 901 Discussion: This requires a renumbering of all options. 903 This renumbering may also be considered as an opportunity to make 904 the numbering straight again shortly before nailing down the 905 protocol 907 In particular, Content-Type is now probably better considered to 908 be elective. 910 7.1. Potential new option number allocation 912 We want to give one example for a revised allocation of option 913 numbers. Option numbers are given as decimal numbers, one each for 914 xxx, nnn, and UC, with the UC values as follows 916 +-----------+------------+------------------------------------+ 917 | UC binary | UC decimal | meaning | 918 +-----------+------------+------------------------------------+ 919 | 00 | 0 | (safe, elective, 111=no-cache-key) | 920 | | | | 921 | 01 | 1 | (safe, critical, 111=no-cache-key) | 922 | | | | 923 | 10 | 2 | (unsafe, elective) | 924 | | | | 925 | 11 | 3 | (unsafe, critical) | 926 +-----------+------------+------------------------------------+ 928 The table is: 930 +-----+---------+-------+-------------------+-----------------------+ 931 | New | xx nnn | Old | Name | Comment | 932 | | UC | | | | 933 +-----+---------+-------+-------------------+-----------------------+ 934 | 4 | 0 1 0 | 1 | Content-Type | category change | 935 | | | | | (elective) | 936 | | | | | | 937 | 8 | 0 2 0 | 4 | ETag | | 938 | | | | | | 939 | 12 | 0 3 0 | 12 | Accept | | 940 | | | | | | 941 | 16 | 0 4 0 | 6 | Location-Path | | 942 | | | | | | 943 | 20 | 0 5 0 | 8 | Location-Query | | 944 | | | | | | 945 | 24 | 0 6 0 | - | (unused) | | 946 | | | | | | 947 | 28 | 0 7 0 | 18 | Size | needs nnn=111 | 948 | | | | | | 949 | 32 | 1 0 0 | 20/22 | Patience | | 950 | | | | | | 951 | 64 | 2 x 0 | - | Location-reserved | (nnn = 0..3, 4 | 952 | | | | | reserved numbers) | 953 | | | | | | 954 | 1 | 0 0 1 | 13 | If-Match | | 955 | | | | | | 956 | 5 | 0 1 1 | 21 | If-None-Match | | 957 | | | | | | 958 | 2 | 0 0 2 | 2 | Max-Age | | 959 | | | | | | 960 | 6 | 0 1 2 | 10 | Observe | | 961 | | | | | | 962 | 10 | 0 2 2 | xx | Observe-2 | | 963 | | | | | | 964 | 14 | 0 3 2 | xx | (unused) | was fencepost | 965 | | | | | | 966 | 3 | 0 0 3 | 3 | Proxy-Uri | | 967 | | | | | | 968 | 7 | 0 1 3 | 5 | Uri-Host | | 969 | | | | | | 970 | 11 | 0 2 3 | 7 | Uri-Port | | 971 | | | | | | 972 | 15 | 0 3 3 | 9 | Uri-Path | | 973 | | | | | | 974 | 19 | 0 4 3 | 15 | Uri-Query | | 975 | | | | | | 976 | 23 | 0 5 3 | 11 | Token | | 977 | | | | | | 978 | 27 | 0 6 3 | 17 | Block2 | | 979 | | | | | | 980 | 31 | 0 7 3 | 19 | Block1 | yes, we can use | 981 | | | | | nnn=111 with U=1 | 982 +-----+---------+-------+-------------------+-----------------------+ 984 8. IANA Considerations 986 This draft adds option numbers to Table 2 of [I-D.ietf-core-coap]: 988 +--------+---------------------+-----------+ 989 | Number | Name | Reference | 990 +--------+---------------------+-----------+ 991 | 14 | Registered-Elective | [RFCXXXX] | 992 | | | | 993 | 22 | Patience | [RFCXXXX] | 994 | | | | 995 | 24 | Pledge | [RFCXXXX] | 996 | | | | 997 | 25 | Registered-Critical | [RFCXXXX] | 998 +--------+---------------------+-----------+ 1000 Table 1: New CoAP Option Numbers 1002 This draft adds a suboption registry, initially empty. 1004 +------------+-----------------------------+-----------+ 1005 | Number | Name | Reference | 1006 +------------+-----------------------------+-----------+ 1007 | 0..127 | (allocate on export review) | [RFCXXXX] | 1008 | | | | 1009 | 128..16383 | (allocate fcfs) | [RFCXXXX] | 1010 +------------+-----------------------------+-----------+ 1012 Table 2: CoAP Suboption Numbers 1014 9. Security Considerations 1016 TBD. 1018 10. Acknowledgements 1020 This work was partially funded by the Klaus Tschira Foundation and by 1021 Intel Corporation. 1023 Of course, much of the content of this draft is the result of 1024 discussions with the [I-D.ietf-core-coap] authors. 1026 Patience and Leisure were influenced by a mailing list discussion 1027 with Esko Dijk, Kepeng Li, and Salvatore Loreto - thanks! 1029 11. References 1031 11.1. Normative References 1033 [I-D.ietf-core-coap] 1034 Shelby, Z., Hartke, K., Bormann, C., and B. Frank, 1035 "Constrained Application Protocol (CoAP)", 1036 draft-ietf-core-coap-11 (work in progress), July 2012. 1038 [I-D.ietf-core-observe] 1039 Hartke, K., "Observing Resources in CoAP", 1040 draft-ietf-core-observe-05 (work in progress), March 2012. 1042 [I-D.ietf-httpbis-p1-messaging] 1043 Fielding, R., Lafon, Y., and J. Reschke, "HTTP/1.1, part 1044 1: Message Routing and Syntax"", 1045 draft-ietf-httpbis-p1-messaging-20 (work in progress), 1046 July 2012. 1048 [I-D.ietf-httpbis-p4-conditional] 1049 Fielding, R., Lafon, Y., and J. Reschke, "HTTP/1.1, part 1050 4: Conditional Requests", 1051 draft-ietf-httpbis-p4-conditional-20 (work in progress), 1052 July 2012. 1054 [I-D.ietf-httpbis-p6-cache] 1055 Fielding, R., Lafon, Y., Nottingham, M., and J. Reschke, 1056 "HTTP/1.1, part 6: Caching", 1057 draft-ietf-httpbis-p6-cache-20 (work in progress), 1058 July 2012. 1060 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1061 Requirement Levels", BCP 14, RFC 2119, March 1997. 1063 [RFC2616] Fielding, R., Gettys, J., Mogul, J., Frystyk, H., 1064 Masinter, L., Leach, P., and T. Berners-Lee, "Hypertext 1065 Transfer Protocol -- HTTP/1.1", RFC 2616, June 1999. 1067 [RFC4648] Josefsson, S., "The Base16, Base32, and Base64 Data 1068 Encodings", RFC 4648, October 2006. 1070 [RFC6256] Eddy, W. and E. Davies, "Using Self-Delimiting Numeric 1071 Values in Protocols", RFC 6256, May 2011. 1073 11.2. Informative References 1075 [CoRE201] "Multiple Location options need to be processed as a 1076 unit", CoRE ticket #201, 2012, 1077 . 1079 [CoRE214] "Adopt vendor-defined option into core-coap", CoRE 1080 ticket #214, 2012, 1081 . 1083 [CoRE230] "Multiple Location options need to be processed as a 1084 unit", CoRE ticket #230, 2012, 1085 . 1087 [CoRE241] "Proxy Safe & Cache Key indication for options", CoRE 1088 ticket #241, 2012, 1089 . 1091 [REST] Fielding, R., "Architectural Styles and the Design of 1092 Network-based Software Architectures", 2000. 1094 [RFC1924] Elz, R., "A Compact Representation of IPv6 Addresses", 1095 RFC 1924, April 1996. 1097 [RFC2817] Khare, R. and S. Lawrence, "Upgrading to TLS Within 1098 HTTP/1.1", RFC 2817, May 2000. 1100 [RFC5198] Klensin, J. and M. Padlipsky, "Unicode Format for Network 1101 Interchange", RFC 5198, March 2008. 1103 [RFC5226] Narten, T. and H. Alvestrand, "Guidelines for Writing an 1104 IANA Considerations Section in RFCs", BCP 26, RFC 5226, 1105 May 2008. 1107 [RFC6648] Saint-Andre, P., Crocker, D., and M. Nottingham, 1108 "Deprecating the "X-" Prefix and Similar Constructs in 1109 Application Protocols", BCP 178, RFC 6648, June 2012. 1111 Appendix A. The Nursery (Things that still need to ripen a bit) 1113 A.1. Envelope Options 1115 As of [I-D.ietf-core-coap], options can take one of four types, two 1116 of which are mostly identical: 1118 o uint: A non-negative integer which is represented in network byte 1119 order using a variable number of bytes (see [I-D.ietf-core-coap] 1120 Appendix A); 1122 o string: a sequence of bytes that is nominally a Net-Unicode string 1123 [RFC5198]; 1125 o opaque: a sequence of bytes. 1127 o empty (not explicitly identified as a fourth type in 1128 [I-D.ietf-core-coap]). 1130 It turns out some options would benefit from some internal structure. 1131 Also, it may be a good idea to be able to bundle multiple options 1132 into one, in order to ensure consistency for a set of elective 1133 options that need to be processed all or nothing (i.e., the option 1134 becomes critical as soon as another option out of the set is 1135 processed, too). 1137 In this section, we introduce a fifth CoAP option type: Envelope 1138 options. 1140 An envelope option is a sequence of bytes that looks and is 1141 interpreted exactly like a CoAP sequence of options. Instead of an 1142 option count or an end-of-option marker, the sequence of options is 1143 terminated by the end of the envelope option. 1145 The nested options (options inside the envelope option) may come from 1146 the same number space as the top-level CoAP options, or the envelope 1147 option may define its own number space - this choice needs to be 1148 defined for each envelope option. 1150 If the top-level number space is used, the envelope option typically 1151 will restrict the set of options that actually can be used in the 1152 envelope. In particular, it is unlikely that an envelope option will 1153 allow itself inside the envelope (this would be a recursive option). 1155 Envelope options are a general, but simple mechanism. Some of its 1156 potential uses are illustrated by two examples in the cemetery: 1157 Appendix B.1 and Appendix B.2. (Each of these examples has its own 1158 merits and demerits, which led us to decide not to pursue either of 1159 them right now, but this should be discussed separately from the 1160 concept of Envelope options employed in the examples.) 1162 A.2. Payload-Length Option 1164 Not all transport mappings may provide an unambiguous length of the 1165 CoAP message. For UDP, it may also be desirable to pack more than 1166 one CoAP message into one UDP payload (aggregation); in that case, 1167 for all but the last message there needs to be a way to delimit the 1168 payload of that message. 1170 This can be solved using a new option, the Payload-Length option. If 1171 this option is present, the value of this option is an unsigned 1172 integer giving the length of the payload of the message (note that 1173 this integer can be zero for a zero-length payload, which can in turn 1174 be represented by a zero-length option value). (In the UDP 1175 aggregation case, what would have been in the payload of this message 1176 after "payload-length" bytes is then actually one or more additional 1177 messages.) 1179 A.3. URI Authorities with Binary Adresses 1181 One problem with the way URI authorities are represented in the URI 1182 syntax is that the authority part can be very bulky if it encodes an 1183 IPv6 address in ASCII. 1185 Proposal: Provide an option "Uri-Authority-Binary" that can be an 1186 even number of bytes between 2 and 18 except 12 or 14. 1188 o If the number of bytes is 2, the destination IP address of the 1189 packet transporting the CoAP message is implied. 1191 o If the number of bytes is 4 or 6, the first four bytes of the 1192 option value are an IPv4 address in binary. 1194 o If the number of bytes is 8 or 10, the first eight bytes are the 1195 lower 64 bits of an IPv6 address; the upper eight bytes are 1196 implied from the destination address of the packet transporting 1197 the CoAP message. 1199 o If the number of bytes is 16 or 18, the first 16 bytes are an IPv6 1200 address. 1202 o If two more bytes remain, this is a port number (as always in 1203 network byte order). 1205 The resulting authority is (conceptually translated into ASCII and) 1206 used in place of an Uri-Authority option, or inserted into a Proxy- 1207 Uri. Examples: 1209 +-------------+------------------+---------+------------------------+ 1210 | Proxy-Uri | Uri-Authority-Bi | Uri-Pat | URI | 1211 | | nary | h | | 1212 +-------------+------------------+---------+------------------------+ 1213 | (none) | (none) | (none) | "/" | 1214 | | | | | 1215 | (none) | (none) | 'temp' | "/temp" | 1216 | | | | | 1217 | (none) | 2 bytes: 61616 | 'temp' | "coap://[DA]:61616/tem | 1218 | | | | p" | 1219 | | | | | 1220 | (none) | 16 bytes: | temp | "coap://[2000::1]/temp | 1221 | | 2000::1 | | " | 1222 | | | | | 1223 | 'http://' | 10 bytes: | (none) | "http://[DA::123:45]:6 | 1224 | | ::123:45 + 616 | | 16" | 1225 | | | | | 1226 | 'http:///te | 18 bytes: | (none) | "http://[2000::1]:616/ | 1227 | mp' | 2000::1 + 616 | | temp" | 1228 +-------------+------------------+---------+------------------------+ 1230 A.4. Length-aware number encoding (o256) 1232 The number encoding defined in Appendix A of [I-D.ietf-core-coap] has 1233 one significant flaw: Every number has an infinite number of 1234 representations, which can be derived by adding leading zero bytes. 1235 This runs against the principle of minimizing unnecessary choice. 1236 The resulting uncertainty in encoding ultimately leads to unnecessary 1237 interoperability failures. (It also wastes a small fraction of the 1238 encoding space, i.e., it wastes bytes.) 1240 We could solve the first, but not the second, by outlawing leading 1241 zeroes, but then we have to cope with error cases caused by illegal 1242 values, another source of interoperability problems. 1244 The number encoding "o256" defined in this section avoids this flaw. 1245 The suggestion is not to replace CoAP's "uint" encoding wholesale 1246 (CoAP is already too widely implemented for such a change), but to 1247 consider this format for new options. 1249 The basic requirements for such an encoding are: 1251 o numbers are encoded as a sequence of zero or more bytes 1253 o each number has exactly one encoding 1254 o for a < b, encoding-size(a) <= encoding-size(b) -- i.e., with 1255 larger numbers, the encoding only gets larger, never smaller 1256 again. 1258 o within each encoding size (0 bytes, 1 byte, etc.), lexicographical 1259 ordering of the bytes is the same as numeric ordering 1261 Obviously, there is only one encoding that satisfies all these 1262 requirements. As illustrated by Figure 8, this is unambiguously 1263 derived by 1265 1. enumerating all possible byte sequences, ordered by length and 1266 within the same length in lexicographic ordering, and, 1268 2. assigning sequential cardinals. 1270 0x'' -> 0 1271 0x'00' -> 1 1272 0x'01' -> 2 1273 0x'02' -> 3 1274 ... 1275 0x'fe' -> 255 1276 0x'ff' -> 256 1277 0x'0000' -> 257 1278 0x'0001' -> 258 1279 ... 1280 0x'fefd' -> 65534 1281 0x'fefe' -> 65535 1282 0x'feff' -> 65536 1283 ... 1284 0x'ffff' -> 65792 1285 0x'000000' -> 65793 1286 0x'000001' -> 65794 1288 Figure 8: Enumerating byte sequences by length and then lexicographic 1289 order 1291 This results in an exceedingly simple algorithm: each byte is 1292 interpreted in the base-256 place-value system, but stands for a 1293 number between 1 and 256 instead of 0 to 255. We therefore call this 1294 encoding "o256" (one-to-256). 0 is always encoded in zero bytes; 1 to 1295 256 is one byte, 257 (0x101) to 65792 (0x10100) is two bytes, 65793 1296 (0x10101) to 16843008 (0x1010100) is three bytes, etc. 1298 To further illustrate the algorithmic simplicity, pseudocode for 1299 encoding and decoding is given in Figure 9 and Figure 10, 1300 respectively (in the encoder, "prepend" stands for adding a byte at 1301 the _leading_ edge, the requirement for which is a result of the 1302 network byte order). Note that this differs only in a single 1303 subtraction/addition (resp.) of one from the canonical algorithm for 1304 Appendix A uints. 1306 while num > 0 1307 num -= 1 1308 prepend(num & 0xFF) 1309 num >>= 8 1310 end 1312 Figure 9: o256 encoder (pseudocode) 1314 num = 0 1315 each_byte do |b| 1316 num <<= 8 1317 num += b + 1 1318 end 1320 Figure 10: o256 decoder (pseudocode) 1322 On a more philosophical note, it can be observed that o256 solves the 1323 inverse problem of Self-Delimiting Numeric Values (SDNV) [RFC6256]: 1324 SDNV encodes variable-length numbers together with their length 1325 (allowing decoding without knowing their length in advance, deriving 1326 delimiting information from the number encoding). o256 encodes 1327 variable-length numbers when there is a way to separately convey the 1328 length (as in CoAP options), encoding (and later deriving) a small 1329 part of the numeric value into/from that size information. 1331 A.5. SMS encoding 1333 For use in SMS applications, CoAP messages can be transferred using 1334 SMS binary mode. However, there is operational experience showing 1335 that some environments cannot successfully send a binary mode SMS. 1337 For transferring SMS in character mode (7-bit characters), base64- 1338 encoding [RFC4648] is an obvious choice. 3 bytes of message (24 bits) 1339 turn into 4 characters, which cosume 28 bits. The overall overhead 1340 is approximately 17 %; the maximum message size is 120 bytes (160 SMS 1341 characters). 1343 If a more compact encoding is desired, base85 encoding can be 1344 employed (however, probably not the version defined in [RFC1924] -- 1345 instead, the version used in tools such as btoa and PDF should be 1346 chosen). However, this requires division operations. Also, the 1347 base85 character set includes several characters that cannot be 1348 transferred in a single 7-bit unit in SMS and/or are known to cause 1349 operational problems. A modified base85 character set can be defined 1350 to solve the latter problem. 4 bytes of message (32 bits) turn into 5 1351 characters, which consume 35 bits. The overall overhead is 1352 approximately 9.3 %; the resulting maximum message size is 128 bytes 1353 (160 SMS characters). 1355 Base64 and base85 do not make use of the fact that much CoAP data 1356 will be ASCII-based. Therefore, we define the following experimental 1357 SMS encoding. 1359 A.5.1. ASCII-optimized SMS encoding 1361 Not all 128 theoretically possible SMS characters are operationally 1362 free of problems. We therefore define: 1364 Shunned code characters: @ sign, as it maps to 0x00 1366 LF and CR signs (0x0A, 0x0D) 1368 uppercase C cedilla (0x09), as it is often mistranslated in 1369 gateways 1371 ESC (0x1B), as it is used in certain character combinations only 1373 Some ASCII characters cannot be transferred in the base SMS character 1374 set, as their code positions are taken by non-ASCII characters. 1375 These are simply encoded with their ASCII code positions, e.g., an 1376 underscore becomes a section mark (even though underscore has a 1377 different code position in the SMS character set). 1379 Equivalently translated input bytes: $, @, [, \, ], ^, _, `, {, |, 1380 }, ~, DEL 1382 In other words, bytes 0x20 to 0x7F are encoded into the same code 1383 positions in the 7-bit character set. 1385 Out of the remaining code characters, the following SMS characters 1386 are available for encoding: 1388 Non-equivalently translated (NET) code characters: 0x01 to 0x08, (8 1389 characters) 1391 0x0B, 0x0C, (2 characters) 1392 0x0E to 0x1A, (13 characters) 1394 0x1C to 0x1F, (4 characters) 1396 Of the 27 NET code characters, 18 are taken as prefix characters (see 1397 below), and 8 are defined as directly translated characters: 1399 Directly translated bytes: Equivalently translated input bytes are 1400 represented as themselves 1402 0x00 to 0x07 are represented as 0x01 to 0x08 1404 This leaves 0x08 to 0x1F and 0x80 to 0xFF. Of these, the bytes 0x80 1405 to 0x87 and 0xA0 to 0xFF are represented as the bytes 0x00 to 0x07 1406 (represented by characters 0x01 to 0x08) and 0x20 to 0x7F, with a 1407 prefix of 1 (see below). The characters 0x08 to 0x1F are represented 1408 as the characters 0x28 to 0x3F with a prefix of 2 (see below). The 1409 characters 0x88 to 0x9F are represented as the characters 0x48 to 1410 0x5F with a prefix of 2 (see below). (Characters 0x01 to 0x08, 0x20 1411 to 0x27, 0x40 to 0x47, and 0x60 to 0x7f with a prefix of 2 are 1412 reserved for future extensions, which could be used for some 1413 backreferencing or run-length compression.) 1415 Bytes that do not need a prefix (directly translated bytes) are sent 1416 as is. Any byte that does need a prefix (i.e., 1 or 2) is preceded 1417 by a prefix character, which provides a prefix for this and the 1418 following two bytes as follows: 1420 +------+-----+---+------+-----+ 1421 | 0x0B | 100 | | 0x15 | 200 | 1422 +------+-----+---+------+-----+ 1423 | 0x0C | 101 | | 0x16 | 201 | 1424 | | | | | | 1425 | 0x0E | 102 | | 0x17 | 202 | 1426 | | | | | | 1427 | 0x0F | 110 | | 0x18 | 210 | 1428 | | | | | | 1429 | 0x10 | 111 | | 0x19 | 211 | 1430 | | | | | | 1431 | 0x11 | 112 | | 0x1A | 212 | 1432 | | | | | | 1433 | 0x12 | 120 | | 0x1C | 220 | 1434 | | | | | | 1435 | 0x13 | 121 | | 0x1D | 221 | 1436 | | | | | | 1437 | 0x14 | 122 | | 0x1E | 222 | 1438 +------+-----+---+------+-----+ 1440 (This leaves one non-shunned character, 0x1F, for future extension.) 1442 The coding overhead of this encoding for random bytes is similar to 1443 Base85, without the need for a division/multiplication. For bytes 1444 that are mostly ASCII characters, the overhead can easily become 1445 negative. (Conversely, for bytes that are more likely to be non- 1446 ASCII than in a random sequence of bytes, the overhead becomes 1447 greater.) 1449 So, for instance, for the CoAP message in Figure 11: 1451 ver tt code mid 1452 1 ack 2.05 17033 1453 content_type 40 1454 token sometok 1455 3c 2f 3e 3b 74 69 74 6c 65 3d 22 47 65 6e 65 72 |;title="Gener| 1456 61 6c 20 49 6e 66 6f 22 3b 63 74 3d 30 2c 3c 2f |al Info";ct=0,;if="clock"| 1458 3b 72 74 3d 22 54 69 63 6b 73 22 3b 74 69 74 6c |;rt="Ticks";titl| 1459 65 3d 22 49 6e 74 65 72 6e 61 6c 20 43 6c 6f 63 |e="Internal Cloc| 1460 6b 22 3b 63 74 3d 30 2c 3c 2f 61 73 79 6e 63 3e |k";ct=0,| 1461 3b 63 74 3d 30 |;ct=0 | 1463 Figure 11: CoAP response message as captured and decoded 1465 The 116 byte unencoded message is shown as ASCII characters in 1466 Figure 12 (\xDD stands for the byte with the hex digits DD): 1468 bEB\x89\x11(\xA7sometok;title="General Info";ct=0, 1469 ;if="clock";rt="Ticks";title="Internal Clock";ct=0,;ct=0 1471 Figure 12: CoAP response message shown as unencoded characters 1473 The equivalent SMS encoding is shown as equivalent-coded SMS 1474 characters in Figure 13 (7 bits per character, \x12 is a 220 prefix 1475 and \x0B is a 100 prefix, the rest is shown in equivalent encoding), 1476 adding two characters of prefix overhead, for a total length of 118 1477 7-bit characters or 104 (103.25 plus padding) bytes: 1479 bEB\x12I1(\x0B'sometok;title="General Info";ct=0, 1480 ;if="clock";rt="Ticks";title="Internal Clock";ct=0,;ct=0 1482 Figure 13: CoAP response message shown as SMS-encoded characters 1484 Appendix B. The Cemetery (Things we won't do) 1486 This annex documents roads that the WG decided not to take, in order 1487 to spare readers from reinventing them in vain. 1489 B.1. Example envelope option: solving #230 1491 Ticket #230 [CoRE230] points out a design flaw of 1492 [I-D.ietf-core-coap]: When we split the elective Location option of 1493 draft -01 into multiple elective options, we made it possible that an 1494 implementation might process some of these and ignore others, leading 1495 to an incorrect interpretation of the Location expressed by the 1496 server. 1498 There are several more or less savory solutions to #230. 1500 Each of the elective options that together make up the Location could 1501 be defined in such a way that it makes a requirement on the 1502 processing of the related option (essentially revoking their elective 1503 status once the option under consideration is actually processed). 1504 This falls flat as soon as another option is defined that would also 1505 become part of the Location: existing implementations would not know 1506 that the new option is also part of the cluster that is re- 1507 interpreted as critical. The potential future addition of Location- 1508 Host and Location-Port makes this a valid consideration. 1510 A better solution would be to define an elective Envelope Option 1511 called Location. Within a Location Option, the following top-level 1512 options might be allowed (now or in the future): 1514 o Uri-Host 1516 o Uri-Port 1518 o Uri-Path 1520 o Uri-Query 1522 This would unify the code for interpreting the top-level request 1523 options that indicate the request URI with the code that interprets 1524 the Location URI. 1526 The four options listed are all critical, while the envelope is 1527 elective. This gives exactly the desired semantics: If the envelope 1528 is processed at all (which is elective), the nested options are 1529 critical and all need to be processed. 1531 B.2. Example envelope option: proxy-elective options 1533 Another potential application of envelope options is motivated by the 1534 observation that new critical options might not be implemented by all 1535 proxies on the CoAP path to an origin server. So that this does not 1536 become an obstacle to introducing new critical options that are of 1537 interest only to client and origin server, the client might want to 1538 mark some critical options proxy-elective, i.e. elective for a proxy 1539 but still critical for the origin server. 1541 One way to do this would be an Envelope option, the Proxy-Elective 1542 Option. A client might bundle a number of critical options into a 1543 critical Proxy-Elective Option. A proxy that processes the message 1544 is obliged to process the envelope (or reject the message), where 1545 processing means passing on the nested options towards the origin 1546 server (preferably again within a Proxy-Elective option). It can 1547 pass on the nested options, even ones unknown to the proxy, knowing 1548 that the client is happy with proxies not processing all of them. 1550 (The assumption here is that the Proxy-Elective option becomes part 1551 of the base standard, so all but the most basic proxies would know 1552 how to handle it.) 1554 B.3. Stateful URI compression 1556 Is the approximately 25 % average saving achievable with Huffman- 1557 based URI compression schemes worth the complexity? Probably not, 1558 because much higher average savings can be achieved by introducing 1559 state. 1561 Henning Schulzrinne has proposed for a server to be able to supply a 1562 shortened URI once a resource has been requested using the full- 1563 length URI. Let's call such a shortened referent a _Temporary 1564 Resource Identifier_, _TeRI_ for short. This could be expressed by a 1565 response option as shown in Figure 14. 1567 0 1568 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 1569 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1570 | duration | TeRI... 1571 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1573 Figure 14: Option for offering a TeRI in a response 1575 The TeRI offer option indicates that the server promises to offer 1576 this resources under the TeRI given for at least the time given as 1577 the duration. Another TeRI offer can be made later to extend the 1578 duration. 1580 Once a TeRI for a URI is known (and still within its lifetime), the 1581 client can supply a TeRI instead of a URI in its requests. The same 1582 option format as an offer could be used to allow the client to 1583 indicate how long it believes the TeRI will still be valid (so that 1584 the server can decide when to update the lifetime duration). TeRIs 1585 in requests could be distinguished from URIs e.g. by using a 1586 different option number. 1588 Proposal: Add a TeRI option that can be used in CoAP requests and 1589 responses. 1591 Add a way to indicate a TeRI and its duration in a link-value. 1593 Do not add any form of stateless URI encoding. 1595 Benefits: Much higher reduction of message size than any stateless 1596 URI encoding could achieve. 1598 As the use of TeRIs is entirely optional, minimal complexity nodes 1599 can get by without implementing them. 1601 Drawbacks: Adds considerable state and complexity to the protocol. 1603 It turns out that real CoAP URIs are short enough that TeRIs are 1604 not needed. 1606 (Discuss the security implications of TeRIs.) 1608 B.4. Beyond 270 bytes in a single option 1610 The authors would argue that 270 as the maximum length of an option 1611 is already beyond the "painless" threshold. 1613 If that is not the consensus of the WG, the scheme can easily be 1614 extended as in Figure 15: 1616 for 15..269: 1617 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1618 | Option Delta | 1 1 1 1 | Length - 15 | 1619 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1620 | Option Value ... 1621 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1623 for 270..65805: 1624 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1625 | Option Delta | 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1626 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1627 | Length - 270 (in network byte order) | 1628 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1629 | Option Value ... 1630 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1632 Figure 15: Ridiculously Long Option Header 1634 The infinite number of obvious variations on this scheme are left as 1635 an exercise to the reader. 1637 Again, as a precaution to future extensions, the current encoding for 1638 length 270 (eight ones in the extension byte) could be marked as 1639 reserved today. 1641 B.5. Beyond 15 options 1643 (This section keeps discussion that is no longer needed as we have 1644 agreed to do what is documented in Appendix B.6). 1646 The limit of 15 options is motivated by the fixed four-bit field "OC" 1647 that is used for indicating the number of options in the fixed-length 1648 CoAP header (Figure 16). 1650 0 1 2 3 1651 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 1652 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1653 |Ver| T | OC | Code | Message ID | 1654 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1655 | Options (if any) ... 1656 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1657 | Payload (if any) ... 1658 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1660 Figure 16: Four-byte fixed header in a CoAP Message 1662 Note that there is another fixed four-bit field in CoAP: the option 1663 length (Figure 17 - note that this figure is not to the same scale as 1664 the previous figure): 1666 0 1 2 3 4 5 6 7 1667 +---+---+---+---+---+---+---+---+ 1668 | Option Delta | Length | for 0..14 1669 +---+---+---+---+---+---+---+---+ 1670 | Option Value ... 1671 +---+---+---+---+---+---+---+---+ 1673 Figure 17: Short Option Header 1675 Since 15 is inacceptable for a maximum option length, the all-ones 1676 value (15) was taken out of the set of allowable values for the short 1677 header, and a long header was introduced that allows the insertion of 1678 an extension byte (Figure 18): 1680 for 15..270: 1681 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1682 | Option Delta | 1 1 1 1 | Length - 15 | 1683 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1684 | Option Value ... 1685 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1687 Figure 18: Long Option Header 1689 We might want to use the same technique for the CoAP header as well. 1690 There are two obvious places where the extension byte could be 1691 placed: 1693 1. right after the byte carrying the OC field, so the structure is 1694 the same as for the option header; 1696 2. right after the fixed-size CoAP header. 1698 Both solutions lose the fixed-size-ness of the CoAP header. 1700 Solution 1 has the disadvantage that the CoAP header is also changing 1701 in structure: The extension byte is wedged between the first and the 1702 second byte of the CoAP header. This is unfortunate, as the number 1703 of options only comes into play when the option processing begins, so 1704 it is more natural to use solution 2 (Figure 19): 1706 0 1 2 3 1707 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 1708 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1709 |Ver| T | 15 | Code | Message ID | 1710 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1711 | OC - 15 | Options ... 1712 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1713 | Payload (if any) ... 1714 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1716 Figure 19: Extended header for CoAP Messages with 15+ options 1718 This would allow for up to 270 options in a CoAP message, which is 1719 very likely way beyond the "painless" threshold. 1721 B.5.1. Implementation considerations 1723 For a message decoder, this extension creates relatively little pain, 1724 as the number of options only becomes interesting when the encoding 1725 turns to the options part of the message, which is then simply lead 1726 in by the extension byte if the four-bit field is 15. 1728 For a message encoder, this extension is not so rosy. If the encoder 1729 is constructing the message serially, it may not know in advance 1730 whether the number of options will exceed 14. None of the following 1731 implementation strategies is particularly savory, but all of them do 1732 work: 1734 1. Encode the options serially under the assumption that the number 1735 of options will be 14 or less. When the 15th option needs to be 1736 encoded, abort the option encoding, and restart it from scratch 1737 one byte further to the left. 1739 2. Similar to 1, except that the bytes already encoded are all moved 1740 one byte to right, the extension byte is inserted, and the option 1741 encoding process is continued. 1743 3. The encoder always leaves space for the extension byte (at least 1744 if it can't prove the number will be less thatn 14). If the 1745 extension byte is not needed, an Option 0 with length 0 is 1746 encoded instead (i.e., one byte is wasted - this option is 1747 elective and will be ignored by the receiver). 1749 As a minimum, to enable strategy 3, the option 0 should be reserved 1750 at least for the case of length=0. 1752 B.5.2. What should we do now? 1754 As a minimum proposal for the next version of CoAP, the value 15 for 1755 OC should be marked as reserved today. 1757 B.5.3. Alternatives 1759 One alternative that has been discussed previously is to have an 1760 "Options" Option, which allows the carriage of multiple options in 1761 the belly of a single one. This could also be used to carry more 1762 than 15 options. However: 1764 o The conditional introduction of an Options option has 1765 implementation considerations that are likely to be more severe 1766 than the ones listed above; 1768 o since 270 bytes may not be enough for the encoding of _all_ 1769 options, the "Options" option would need to be repeatable. This 1770 creates many different ways to encode the same message, leading to 1771 combinatorial explosion in test cases for ensuring 1772 interoperability. 1774 B.5.4. Alternative: Going to a delimiter model 1776 Another alternative is to spend the additional byte not as an 1777 extended count, but as an option terminator. 1779 B.6. Implementing the option delimiter for 15 or more options 1781 Implementation note: As can be seen from the proof of concept code 1782 in Figure 20, the actual implementation cost for a decoder is 1783 around 4 lines of code (or about 8-10 machine code instructions). 1785 while numopt > 0 1786 nextbyte = ... get next byte 1788 if numopt == 15 # new 1789 break if nextbyte == 0xF0 # new 1790 else # new 1791 numopt -= 1 1792 end # new 1794 ... decode the delta and length from nextbyte and handle them 1795 end 1797 Figure 20: Implementing the Option Terminator 1799 Similarly, creating the option terminator needs about four more lines 1800 (not marked "old" in the C code in Figure 21). 1802 b0 = 0x40 + (tt << 4); /* old */ 1803 buffer[0] = b0 + 15; /* guess first byte */ 1805 .... encode options .... /* old */ 1807 if (option_count >= 15 || first_fragment_already_shipped) 1808 buffer[pos++] = 0xF0; /* use delimiter */ 1809 else /* save a byte: */ 1810 buffer[0] = b0 + option_count; /* old: backpatch */ 1812 Figure 21: Creating the Option Terminator 1814 Appendix C. Experimental Options 1816 This annex documents proposals that need significant additional 1817 discussion before they can become part of (or go back to) the main 1818 CoAP specification. They are not dead, but might die if there turns 1819 out to be no good way to solve the problem. 1821 C.1. Options indicating absolute time 1823 HTTP has a number of headers that may indicate absolute time: 1825 o "Date", defined in Section 14.18 in [RFC2616] (Section 9.3 in 1826 [I-D.ietf-httpbis-p1-messaging]), giving the absolute time a 1827 response was generated; 1829 o "Last-Modified", defined in Section 14.29 in [RFC2616], (Section 1830 6.6 in [I-D.ietf-httpbis-p4-conditional], giving the absolute time 1831 of when the origin server believes the resource representation was 1832 last modified; 1834 o "If-Modified-Since", defined in Section 14.25 in [RFC2616], 1835 "If-Unmodified-Since", defined in Section 14.28 in [RFC2616], and 1836 "If-Range", defined in Section 14.27 in [RFC2616] can be used to 1837 supply absolute time to gate a conditional request; 1839 o "Expires", defined in Section 14.21 in [RFC2616] (Section 3.3 in 1840 [I-D.ietf-httpbis-p6-cache]), giving the absolute time after which 1841 a response is considered stale. 1843 o The more obscure headers "Retry-After", defined in Section 14.37 1844 in [RFC2616], and "Warning", defined in section 14.46 in 1845 [RFC2616], also may employ absolute time. 1847 [I-D.ietf-core-coap] defines a single "Date" option, which however 1848 "indicates the creation time and date of a given resource 1849 representation", i.e., is closer to a "Last-Modified" HTTP header. 1850 HTTP's caching rules [I-D.ietf-httpbis-p6-cache] make use of both 1851 "Date" and "Last-Modified", combined with "Expires". The specific 1852 semantics required for CoAP needs further consideration. 1854 In addition to the definition of the semantics, an encoding for 1855 absolute times needs to be specified. 1857 In UNIX-related systems, it is customary to indicate absolute time as 1858 an integer number of seconds, after midnight UTC, January 1, 1970. 1859 Unless negative numbers are employed, this time format cannot 1860 represent time values prior to January 1, 1970, which probably is not 1861 required for the uses ob absolute time in CoAP. 1863 If a 32-bit integer is used and allowance is made for a sign-bit in a 1864 local implementation, the latest UTC time value that can be 1865 represented by the resulting 31 bit integer value is 03:14:07 on 1866 January 19, 2038. If the 32-bit integer is used as an unsigned 1867 value, the last date is 2106-02-07, 06:28:15. 1869 The reach can be extended by: - moving the epoch forward, e.g. by 40 1870 years (= 1262304000 seconds) to 2010-01-01. This makes it impossible 1871 to represent Last-Modified times in that past (such as could be 1872 gatewayed in from HTTP). - extending the number of bits, e.g. by one 1873 more byte, either always or as one of two formats, keeping the 32-bit 1874 variant as well. 1876 Also, the resolution can be extended by expressing time in 1877 milliseconds etc., requiring even more bits (e.g., a 48-bit unsigned 1878 integer of milliseconds would last well after year 9999.) 1880 For experiments, an experimental "Date" option is defined with the 1881 semantics of HTTP's "Last-Modified". It can carry an unsigned 1882 integer of 32, 40, or 48 bits; 32- and 40-bit integers indicate the 1883 absolute time in seconds since 1970-01-01 00:00 UTC, while 48-bit 1884 integers indicate the absolute time in milliseconds since 1970-01-01 1885 00:00 UTC. 1887 However, that option is not really that useful until there is a 1888 "If-Modified-Since" option as well. 1890 (Also: Discuss nodes without clocks.) 1892 C.2. Representing Durations 1894 Various message types used in CoAP need the representation of 1895 *durations*, i.e. of the length of a timespan. In SI units, these 1896 are measured in seconds. CoAP durations represent integer numbers of 1897 seconds, but instead of representing these numbers as integers, a 1898 more compact single-byte pseudo-floating-point (pseudo-FP) 1899 representation is used (Figure 22). 1901 0 1 2 3 4 5 6 7 1902 +---+---+---+---+---+---+---+---+ 1903 | 0... value | 1904 +---+---+---+---+---+---+---+---+ 1906 +---+---+---+---+---+---+---+---+ 1907 | 1... mantissa | exponent | 1908 +---+---+---+---+---+---+---+---+ 1909 Figure 22: Duration in (8,4) pseudo-FP representation 1911 If the high bit is clear, the entire n-bit value (including the high 1912 bit) is the decoded value. If the high bit is set, the mantissa 1913 (including the high bit, with the exponent field cleared out but 1914 still present) is shifted left by the exponent to yield the decoded 1915 value. 1917 The (n,e)-pseudo-FP format can be decoded with a single line of code 1918 (plus a couple of constant definitions), as demonstrated in 1919 Figure 23. 1921 #define N 8 1922 #define E 4 1923 #define HIBIT (1 << (N - 1)) 1924 #define EMASK ((1 << E) - 1) 1925 #define MMASK ((1 << N) - 1 - EMASK) 1927 #define DECODE_8_4(r) (r < HIBIT ? r : (r & MMASK) << (r & EMASK)) 1929 Figure 23: Decoding an (8,4) pseudo-FP value 1931 Note that a pseudo-FP encoder needs to consider rounding; different 1932 applications of durations may favor rounding up or rounding down the 1933 value encoded in the message. 1935 The highest pseudo-FP value, represented by an all-ones byte (0xFF), 1936 is reserved to indicate an indefinite duration. The next lower value 1937 (0xEF) is thus the highest representable value and is decoded as 1938 7340032 seconds, a little more than 12 weeks. 1940 C.3. Rationale 1942 Where CPU power and memory is abundant, a duration can almost always 1943 be adequately represented by a non-negative floating-point number 1944 representing that number of seconds. Historically, many APIs have 1945 also used an integer representation, which limits both the resolution 1946 (e.g., if the integer represents the duration in seconds) and often 1947 the range (integer machine types have range limits that may become 1948 relevant). UNIX's "time_t" (which is used for both absolute time and 1949 durations) originally was a signed 32-bit value of seconds, but was 1950 later complemented by an additional integer to add microsecond 1951 ("struct timeval") and then later nanosecond ("struct timespec") 1952 resolution. 1954 Three decisions need to be made for each application of the concept 1955 of duration: 1957 o the *resolution*. What rounding error is acceptable? 1959 o the *range*. What is the maximum duration that needs to be 1960 represented? 1962 o the *number of bits* that can be expended. 1964 Obviously, these decisions are interrelated. Typically, a large 1965 range needs a large number of bits, unless resolution is traded. For 1966 most applications, the actual requirement for resolution are limited 1967 for longer durations, but can be more acute for shorter durations. 1969 C.4. Pseudo-Floating Point 1971 Constrained systems typically avoid the use of floating-point (FP) 1972 values, as 1974 o simple CPUs often don't have support for floating-point datatypes 1976 o software floating-point libraries are expensive in code size and 1977 slow. 1979 In addition, floating-point datatypes used to be a significant 1980 element of market differentiation in CPU design; it has taken the 1981 industry a long time to agree on a standard floating point 1982 representation. 1984 These issues have led to protocols that try to constrain themselves 1985 to integer representation even where the ability of a floating point 1986 representation to trade range for resolution would be beneficial. 1988 The idea of introducing _pseudo-FP_ is to obtain the increased range 1989 provided by embedding an exponent, without necessarily getting stuck 1990 with hardware datatypes or inefficient software floating-point 1991 libraries. 1993 For the purposes of this draft, we define an (n,e)-pseudo-FP as a 1994 fixed-length value of n bits, e of which may be used for an exponent. 1995 Figure 22 illustrates an (8,4)-pseudo-FP value. 1997 If the high bit is clear, the entire n-bit value (including the high 1998 bit) is the decoded value. If the high bit is set, the mantissa 1999 (including the high bit, but with the exponent field cleared out) is 2000 shifted left by the exponent to yield the decoded value. 2002 The (n,e)-pseudo-FP format can be decoded with a single line of code 2003 (plus a couple of constant definition), as demonstrated in Figure 23. 2005 Only non-negative numbers can be represented by this format. It is 2006 designed to provide full integer resolution for values from 0 to 2007 2^(n-1)-1, i.e., 0 to 127 in the (8,4) case, and a mantissa of n-e 2008 bits from 2^(n-1) to (2^n-2^e)*2^(2^e-1), i.e., 128 to 7864320 in the 2009 (8,4) case. By choosing e carefully, resolution can be traded 2010 against range. 2012 Note that a pseudo-FP encoder needs to consider rounding; different 2013 applications of durations may favor rounding up or rounding down the 2014 value encoded in the message. This requires a little more than a 2015 single line of code (which is left as an exercise to the reader, as 2016 the most efficient expression depends on hardware details). 2018 C.5. A Duration Type for CoAP 2020 CoAP needs durations in a number of places. In [I-D.ietf-core-coap], 2021 durations occur in the option "Subscription-lifetime" as well as in 2022 the option "Max-age". (Note that the option "Date" is not a 2023 duration, but a point in time.) Other durations of this kind may be 2024 added later. 2026 Most durations relevant to CoAP are best expressed with a minimum 2027 resolution of one second. More detailed resolutions are unlikely to 2028 provide much benefit. 2030 The range of lifetimes and caching ages are probably best kept below 2031 the order of magnitude of months. An (8,4)-pseudo-FP has the maximum 2032 value of 7864320, which is about 91 days; this appears to be adequate 2033 for a subscription lifetime and probably even for a maximum cache 2034 age. Figure 24 shows the values that can be expressed. (If a larger 2035 range for the latter is indeed desired, an (8,5)-pseudo-FP could be 2036 used; this would last 15 milleniums, at the cost of having only 3 2037 bits of accuracy for values larger than 127 seconds.) 2039 Proposal: A single duration type is used throughout CoAP, based on 2040 an (8,4)-pseudo-FP giving a duration in seconds. 2042 Benefits: Implementations can use a single piece of code for 2043 managing all CoAP-related durations. 2045 In addition, length information never needs to be managed for 2046 durations that are embedded in other data structures: All 2047 durations are expressed by a single byte. 2049 It might be worthwhile to reserve one duration value, e.g. 0xFF, for 2050 an indefinite duration. 2052 Duration Seconds Encoded 2053 ----------- ---------- ------- 2054 00:00:00 0x00000000 0x00 2055 00:00:01 0x00000001 0x01 2056 00:00:02 0x00000002 0x02 2057 00:00:03 0x00000003 0x03 2058 00:00:04 0x00000004 0x04 2059 00:00:05 0x00000005 0x05 2060 00:00:06 0x00000006 0x06 2061 00:00:07 0x00000007 0x07 2062 00:00:08 0x00000008 0x08 2063 00:00:09 0x00000009 0x09 2064 00:00:10 0x0000000a 0x0a 2065 00:00:11 0x0000000b 0x0b 2066 00:00:12 0x0000000c 0x0c 2067 00:00:13 0x0000000d 0x0d 2068 00:00:14 0x0000000e 0x0e 2069 00:00:15 0x0000000f 0x0f 2070 00:00:16 0x00000010 0x10 2071 00:00:17 0x00000011 0x11 2072 00:00:18 0x00000012 0x12 2073 00:00:19 0x00000013 0x13 2074 00:00:20 0x00000014 0x14 2075 00:00:21 0x00000015 0x15 2076 00:00:22 0x00000016 0x16 2077 00:00:23 0x00000017 0x17 2078 00:00:24 0x00000018 0x18 2079 00:00:25 0x00000019 0x19 2080 00:00:26 0x0000001a 0x1a 2081 00:00:27 0x0000001b 0x1b 2082 00:00:28 0x0000001c 0x1c 2083 00:00:29 0x0000001d 0x1d 2084 00:00:30 0x0000001e 0x1e 2085 00:00:31 0x0000001f 0x1f 2086 00:00:32 0x00000020 0x20 2087 00:00:33 0x00000021 0x21 2088 00:00:34 0x00000022 0x22 2089 00:00:35 0x00000023 0x23 2090 00:00:36 0x00000024 0x24 2091 00:00:37 0x00000025 0x25 2092 00:00:38 0x00000026 0x26 2093 00:00:39 0x00000027 0x27 2094 00:00:40 0x00000028 0x28 2095 00:00:41 0x00000029 0x29 2096 00:00:42 0x0000002a 0x2a 2097 00:00:43 0x0000002b 0x2b 2098 00:00:44 0x0000002c 0x2c 2099 00:00:45 0x0000002d 0x2d 2100 00:00:46 0x0000002e 0x2e 2101 00:00:47 0x0000002f 0x2f 2102 00:00:48 0x00000030 0x30 2103 00:00:49 0x00000031 0x31 2104 00:00:50 0x00000032 0x32 2105 00:00:51 0x00000033 0x33 2106 00:00:52 0x00000034 0x34 2107 00:00:53 0x00000035 0x35 2108 00:00:54 0x00000036 0x36 2109 00:00:55 0x00000037 0x37 2110 00:00:56 0x00000038 0x38 2111 00:00:57 0x00000039 0x39 2112 00:00:58 0x0000003a 0x3a 2113 00:00:59 0x0000003b 0x3b 2114 00:01:00 0x0000003c 0x3c 2115 00:01:01 0x0000003d 0x3d 2116 00:01:02 0x0000003e 0x3e 2117 00:01:03 0x0000003f 0x3f 2118 00:01:04 0x00000040 0x40 2119 00:01:05 0x00000041 0x41 2120 00:01:06 0x00000042 0x42 2121 00:01:07 0x00000043 0x43 2122 00:01:08 0x00000044 0x44 2123 00:01:09 0x00000045 0x45 2124 00:01:10 0x00000046 0x46 2125 00:01:11 0x00000047 0x47 2126 00:01:12 0x00000048 0x48 2127 00:01:13 0x00000049 0x49 2128 00:01:14 0x0000004a 0x4a 2129 00:01:15 0x0000004b 0x4b 2130 00:01:16 0x0000004c 0x4c 2131 00:01:17 0x0000004d 0x4d 2132 00:01:18 0x0000004e 0x4e 2133 00:01:19 0x0000004f 0x4f 2134 00:01:20 0x00000050 0x50 2135 00:01:21 0x00000051 0x51 2136 00:01:22 0x00000052 0x52 2137 00:01:23 0x00000053 0x53 2138 00:01:24 0x00000054 0x54 2139 00:01:25 0x00000055 0x55 2140 00:01:26 0x00000056 0x56 2141 00:01:27 0x00000057 0x57 2142 00:01:28 0x00000058 0x58 2143 00:01:29 0x00000059 0x59 2144 00:01:30 0x0000005a 0x5a 2145 00:01:31 0x0000005b 0x5b 2146 00:01:32 0x0000005c 0x5c 2147 00:01:33 0x0000005d 0x5d 2148 00:01:34 0x0000005e 0x5e 2149 00:01:35 0x0000005f 0x5f 2150 00:01:36 0x00000060 0x60 2151 00:01:37 0x00000061 0x61 2152 00:01:38 0x00000062 0x62 2153 00:01:39 0x00000063 0x63 2154 00:01:40 0x00000064 0x64 2155 00:01:41 0x00000065 0x65 2156 00:01:42 0x00000066 0x66 2157 00:01:43 0x00000067 0x67 2158 00:01:44 0x00000068 0x68 2159 00:01:45 0x00000069 0x69 2160 00:01:46 0x0000006a 0x6a 2161 00:01:47 0x0000006b 0x6b 2162 00:01:48 0x0000006c 0x6c 2163 00:01:49 0x0000006d 0x6d 2164 00:01:50 0x0000006e 0x6e 2165 00:01:51 0x0000006f 0x6f 2166 00:01:52 0x00000070 0x70 2167 00:01:53 0x00000071 0x71 2168 00:01:54 0x00000072 0x72 2169 00:01:55 0x00000073 0x73 2170 00:01:56 0x00000074 0x74 2171 00:01:57 0x00000075 0x75 2172 00:01:58 0x00000076 0x76 2173 00:01:59 0x00000077 0x77 2174 00:02:00 0x00000078 0x78 2175 00:02:01 0x00000079 0x79 2176 00:02:02 0x0000007a 0x7a 2177 00:02:03 0x0000007b 0x7b 2178 00:02:04 0x0000007c 0x7c 2179 00:02:05 0x0000007d 0x7d 2180 00:02:06 0x0000007e 0x7e 2181 00:02:07 0x0000007f 0x7f 2182 00:02:08 0x00000080 0x80 2183 00:02:24 0x00000090 0x90 2184 00:02:40 0x000000a0 0xa0 2185 00:02:56 0x000000b0 0xb0 2186 00:03:12 0x000000c0 0xc0 2187 00:03:28 0x000000d0 0xd0 2188 00:03:44 0x000000e0 0xe0 2189 00:04:00 0x000000f0 0xf0 2190 00:04:16 0x00000100 0x81 2191 00:04:48 0x00000120 0x91 2192 00:05:20 0x00000140 0xa1 2193 00:05:52 0x00000160 0xb1 2194 00:06:24 0x00000180 0xc1 2195 00:06:56 0x000001a0 0xd1 2196 00:07:28 0x000001c0 0xe1 2197 00:08:00 0x000001e0 0xf1 2198 00:08:32 0x00000200 0x82 2199 00:09:36 0x00000240 0x92 2200 00:10:40 0x00000280 0xa2 2201 00:11:44 0x000002c0 0xb2 2202 00:12:48 0x00000300 0xc2 2203 00:13:52 0x00000340 0xd2 2204 00:14:56 0x00000380 0xe2 2205 00:16:00 0x000003c0 0xf2 2206 00:17:04 0x00000400 0x83 2207 00:19:12 0x00000480 0x93 2208 00:21:20 0x00000500 0xa3 2209 00:23:28 0x00000580 0xb3 2210 00:25:36 0x00000600 0xc3 2211 00:27:44 0x00000680 0xd3 2212 00:29:52 0x00000700 0xe3 2213 00:32:00 0x00000780 0xf3 2214 00:34:08 0x00000800 0x84 2215 00:38:24 0x00000900 0x94 2216 00:42:40 0x00000a00 0xa4 2217 00:46:56 0x00000b00 0xb4 2218 00:51:12 0x00000c00 0xc4 2219 00:55:28 0x00000d00 0xd4 2220 00:59:44 0x00000e00 0xe4 2221 01:04:00 0x00000f00 0xf4 2222 01:08:16 0x00001000 0x85 2223 01:16:48 0x00001200 0x95 2224 01:25:20 0x00001400 0xa5 2225 01:33:52 0x00001600 0xb5 2226 01:42:24 0x00001800 0xc5 2227 01:50:56 0x00001a00 0xd5 2228 01:59:28 0x00001c00 0xe5 2229 02:08:00 0x00001e00 0xf5 2230 02:16:32 0x00002000 0x86 2231 02:33:36 0x00002400 0x96 2232 02:50:40 0x00002800 0xa6 2233 03:07:44 0x00002c00 0xb6 2234 03:24:48 0x00003000 0xc6 2235 03:41:52 0x00003400 0xd6 2236 03:58:56 0x00003800 0xe6 2237 04:16:00 0x00003c00 0xf6 2238 04:33:04 0x00004000 0x87 2239 05:07:12 0x00004800 0x97 2240 05:41:20 0x00005000 0xa7 2241 06:15:28 0x00005800 0xb7 2242 06:49:36 0x00006000 0xc7 2243 07:23:44 0x00006800 0xd7 2244 07:57:52 0x00007000 0xe7 2245 08:32:00 0x00007800 0xf7 2246 09:06:08 0x00008000 0x88 2247 10:14:24 0x00009000 0x98 2248 11:22:40 0x0000a000 0xa8 2249 12:30:56 0x0000b000 0xb8 2250 13:39:12 0x0000c000 0xc8 2251 14:47:28 0x0000d000 0xd8 2252 15:55:44 0x0000e000 0xe8 2253 17:04:00 0x0000f000 0xf8 2254 18:12:16 0x00010000 0x89 2255 20:28:48 0x00012000 0x99 2256 22:45:20 0x00014000 0xa9 2257 1d 01:01:52 0x00016000 0xb9 2258 1d 03:18:24 0x00018000 0xc9 2259 1d 05:34:56 0x0001a000 0xd9 2260 1d 07:51:28 0x0001c000 0xe9 2261 1d 10:08:00 0x0001e000 0xf9 2262 1d 12:24:32 0x00020000 0x8a 2263 1d 16:57:36 0x00024000 0x9a 2264 1d 21:30:40 0x00028000 0xaa 2265 2d 02:03:44 0x0002c000 0xba 2266 2d 06:36:48 0x00030000 0xca 2267 2d 11:09:52 0x00034000 0xda 2268 2d 15:42:56 0x00038000 0xea 2269 2d 20:16:00 0x0003c000 0xfa 2270 3d 00:49:04 0x00040000 0x8b 2271 3d 09:55:12 0x00048000 0x9b 2272 3d 19:01:20 0x00050000 0xab 2273 4d 04:07:28 0x00058000 0xbb 2274 4d 13:13:36 0x00060000 0xcb 2275 4d 22:19:44 0x00068000 0xdb 2276 5d 07:25:52 0x00070000 0xeb 2277 5d 16:32:00 0x00078000 0xfb 2278 6d 01:38:08 0x00080000 0x8c 2279 6d 19:50:24 0x00090000 0x9c 2280 7d 14:02:40 0x000a0000 0xac 2281 8d 08:14:56 0x000b0000 0xbc 2282 9d 02:27:12 0x000c0000 0xcc 2283 9d 20:39:28 0x000d0000 0xdc 2284 10d 14:51:44 0x000e0000 0xec 2285 11d 09:04:00 0x000f0000 0xfc 2286 12d 03:16:16 0x00100000 0x8d 2287 13d 15:40:48 0x00120000 0x9d 2288 15d 04:05:20 0x00140000 0xad 2289 16d 16:29:52 0x00160000 0xbd 2290 18d 04:54:24 0x00180000 0xcd 2291 19d 17:18:56 0x001a0000 0xdd 2292 21d 05:43:28 0x001c0000 0xed 2293 22d 18:08:00 0x001e0000 0xfd 2294 24d 06:32:32 0x00200000 0x8e 2295 27d 07:21:36 0x00240000 0x9e 2296 30d 08:10:40 0x00280000 0xae 2297 33d 08:59:44 0x002c0000 0xbe 2298 36d 09:48:48 0x00300000 0xce 2299 39d 10:37:52 0x00340000 0xde 2300 42d 11:26:56 0x00380000 0xee 2301 45d 12:16:00 0x003c0000 0xfe 2302 48d 13:05:04 0x00400000 0x8f 2303 54d 14:43:12 0x00480000 0x9f 2304 60d 16:21:20 0x00500000 0xaf 2305 66d 17:59:28 0x00580000 0xbf 2306 72d 19:37:36 0x00600000 0xcf 2307 78d 21:15:44 0x00680000 0xdf 2308 84d 22:53:52 0x00700000 0xef 2309 91d 00:32:00 0x00780000 0xff (reserved) 2311 Figure 24 2313 Authors' Addresses 2315 Carsten Bormann 2316 Universitaet Bremen TZI 2317 Postfach 330440 2318 Bremen D-28359 2319 Germany 2321 Phone: +49-421-218-63921 2322 Email: cabo@tzi.org 2324 Klaus Hartke 2325 Universitaet Bremen TZI 2326 Postfach 330440 2327 Bremen D-28359 2328 Germany 2330 Phone: +49-421-218-63905 2331 Email: hartke@tzi.org