idnits 2.17.1 draft-ietf-httpbis-messaging-08.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- -- The draft header indicates that this document obsoletes RFC7230, but the abstract doesn't seem to directly say this. It does mention RFC7230 though, so this could be OK. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document seems to contain a disclaimer for pre-RFC5378 work, but was first submitted on or after 10 November 2008. The disclaimer is usually necessary only for documents that revise or obsolete older RFCs, and that take significant amounts of text from those RFCs. If you can contact all authors of the source material and they are willing to grant the BCP78 rights to the IETF Trust, you can and should remove the disclaimer. Otherwise, the disclaimer is needed and you can ignore this comment. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (May 26, 2020) is 1425 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Unused Reference: 'RFC3986' is defined on line 2103, but no explicit reference was found in the text == Unused Reference: 'RFC7231' is defined on line 2190, but no explicit reference was found in the text == Outdated reference: A later version (-19) exists of draft-ietf-httpbis-cache-08 -- Possible downref: Normative reference to a draft: ref. 'Caching' ** Downref: Normative reference to an Informational RFC: RFC 1950 ** Downref: Normative reference to an Informational RFC: RFC 1951 ** Downref: Normative reference to an Informational RFC: RFC 1952 == Outdated reference: A later version (-19) exists of draft-ietf-httpbis-semantics-08 -- Possible downref: Normative reference to a draft: ref. 'Semantics' -- Possible downref: Non-RFC (?) normative reference: ref. 'USASCII' -- Possible downref: Non-RFC (?) normative reference: ref. 'Welch' -- Obsolete informational reference (is this intentional?): RFC 7230 (ref. 'Err4667') (Obsoleted by RFC 9110, RFC 9112) -- Obsolete informational reference (is this intentional?): RFC 2068 (Obsoleted by RFC 2616) -- Duplicate reference: RFC7230, mentioned in 'RFC7230', was also mentioned in 'Err4667'. -- Obsolete informational reference (is this intentional?): RFC 7230 (Obsoleted by RFC 9110, RFC 9112) -- Obsolete informational reference (is this intentional?): RFC 7231 (Obsoleted by RFC 9110) Summary: 3 errors (**), 0 flaws (~~), 6 warnings (==), 11 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 HTTP Working Group R. Fielding, Ed. 3 Internet-Draft Adobe 4 Obsoletes: 7230 (if approved) M. Nottingham, Ed. 5 Intended status: Standards Track Fastly 6 Expires: November 27, 2020 J. Reschke, Ed. 7 greenbytes 8 May 26, 2020 10 HTTP/1.1 Messaging 11 draft-ietf-httpbis-messaging-08 13 Abstract 15 The Hypertext Transfer Protocol (HTTP) is a stateless application- 16 level protocol for distributed, collaborative, hypertext information 17 systems. This document specifies the HTTP/1.1 message syntax, 18 message parsing, connection management, and related security 19 concerns. 21 This document obsoletes portions of RFC 7230. 23 Editorial Note 25 This note is to be removed before publishing as an RFC. 27 Discussion of this draft takes place on the HTTP working group 28 mailing list (ietf-http-wg@w3.org), which is archived at 29 . 31 Working Group information can be found at ; 32 source code and issues list for this draft can be found at 33 . 35 The changes in this draft are summarized in Appendix D.9. 37 Status of This Memo 39 This Internet-Draft is submitted in full conformance with the 40 provisions of BCP 78 and BCP 79. 42 Internet-Drafts are working documents of the Internet Engineering 43 Task Force (IETF). Note that other groups may also distribute 44 working documents as Internet-Drafts. The list of current Internet- 45 Drafts is at https://datatracker.ietf.org/drafts/current/. 47 Internet-Drafts are draft documents valid for a maximum of six months 48 and may be updated, replaced, or obsoleted by other documents at any 49 time. It is inappropriate to use Internet-Drafts as reference 50 material or to cite them other than as "work in progress." 52 This Internet-Draft will expire on November 27, 2020. 54 Copyright Notice 56 Copyright (c) 2020 IETF Trust and the persons identified as the 57 document authors. All rights reserved. 59 This document is subject to BCP 78 and the IETF Trust's Legal 60 Provisions Relating to IETF Documents 61 (https://trustee.ietf.org/license-info) in effect on the date of 62 publication of this document. Please review these documents 63 carefully, as they describe your rights and restrictions with respect 64 to this document. Code Components extracted from this document must 65 include Simplified BSD License text as described in Section 4.e of 66 the Trust Legal Provisions and are provided without warranty as 67 described in the Simplified BSD License. 69 This document may contain material from IETF Documents or IETF 70 Contributions published or made publicly available before November 71 10, 2008. The person(s) controlling the copyright in some of this 72 material may not have granted the IETF Trust the right to allow 73 modifications of such material outside the IETF Standards Process. 74 Without obtaining an adequate license from the person(s) controlling 75 the copyright in such materials, this document may not be modified 76 outside the IETF Standards Process, and derivative works of it may 77 not be created outside the IETF Standards Process, except to format 78 it for publication as an RFC or to translate it into languages other 79 than English. 81 Table of Contents 83 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 4 84 1.1. Requirements Notation . . . . . . . . . . . . . . . . . . 5 85 1.2. Syntax Notation . . . . . . . . . . . . . . . . . . . . . 5 86 2. Message . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 87 2.1. Message Format . . . . . . . . . . . . . . . . . . . . . 6 88 2.2. Message Parsing . . . . . . . . . . . . . . . . . . . . . 7 89 2.3. HTTP Version . . . . . . . . . . . . . . . . . . . . . . 8 90 3. Request Line . . . . . . . . . . . . . . . . . . . . . . . . 9 91 3.1. Method . . . . . . . . . . . . . . . . . . . . . . . . . 9 92 3.2. Request Target . . . . . . . . . . . . . . . . . . . . . 10 93 3.2.1. origin-form . . . . . . . . . . . . . . . . . . . . . 10 94 3.2.2. absolute-form . . . . . . . . . . . . . . . . . . . . 11 95 3.2.3. authority-form . . . . . . . . . . . . . . . . . . . 12 96 3.2.4. asterisk-form . . . . . . . . . . . . . . . . . . . . 12 98 3.3. Reconstructing the Target URI . . . . . . . . . . . . . . 13 99 4. Status Line . . . . . . . . . . . . . . . . . . . . . . . . . 14 100 5. Field Syntax . . . . . . . . . . . . . . . . . . . . . . . . 15 101 5.1. Field Line Parsing . . . . . . . . . . . . . . . . . . . 16 102 5.2. Obsolete Line Folding . . . . . . . . . . . . . . . . . . 16 103 6. Message Body . . . . . . . . . . . . . . . . . . . . . . . . 17 104 6.1. Transfer-Encoding . . . . . . . . . . . . . . . . . . . . 17 105 6.2. Content-Length . . . . . . . . . . . . . . . . . . . . . 19 106 6.3. Message Body Length . . . . . . . . . . . . . . . . . . . 19 107 7. Transfer Codings . . . . . . . . . . . . . . . . . . . . . . 21 108 7.1. Chunked Transfer Coding . . . . . . . . . . . . . . . . . 22 109 7.1.1. Chunk Extensions . . . . . . . . . . . . . . . . . . 23 110 7.1.2. Chunked Trailer Section . . . . . . . . . . . . . . . 24 111 7.1.3. Decoding Chunked . . . . . . . . . . . . . . . . . . 24 112 7.2. Transfer Codings for Compression . . . . . . . . . . . . 25 113 7.3. Transfer Coding Registry . . . . . . . . . . . . . . . . 25 114 7.4. TE . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 115 8. Handling Incomplete Messages . . . . . . . . . . . . . . . . 27 116 9. Connection Management . . . . . . . . . . . . . . . . . . . . 28 117 9.1. Connection . . . . . . . . . . . . . . . . . . . . . . . 28 118 9.2. Establishment . . . . . . . . . . . . . . . . . . . . . . 30 119 9.3. Associating a Response to a Request . . . . . . . . . . . 30 120 9.4. Persistence . . . . . . . . . . . . . . . . . . . . . . . 31 121 9.4.1. Retrying Requests . . . . . . . . . . . . . . . . . . 32 122 9.4.2. Pipelining . . . . . . . . . . . . . . . . . . . . . 32 123 9.5. Concurrency . . . . . . . . . . . . . . . . . . . . . . . 33 124 9.6. Failures and Timeouts . . . . . . . . . . . . . . . . . . 33 125 9.7. Tear-down . . . . . . . . . . . . . . . . . . . . . . . . 34 126 9.8. TLS Connection Closure . . . . . . . . . . . . . . . . . 35 127 9.9. Upgrade . . . . . . . . . . . . . . . . . . . . . . . . . 36 128 9.9.1. Upgrade Protocol Names . . . . . . . . . . . . . . . 38 129 9.9.2. Upgrade Token Registry . . . . . . . . . . . . . . . 38 130 10. Enclosing Messages as Data . . . . . . . . . . . . . . . . . 39 131 10.1. Media Type message/http . . . . . . . . . . . . . . . . 39 132 10.2. Media Type application/http . . . . . . . . . . . . . . 40 133 11. Security Considerations . . . . . . . . . . . . . . . . . . . 42 134 11.1. Response Splitting . . . . . . . . . . . . . . . . . . . 42 135 11.2. Request Smuggling . . . . . . . . . . . . . . . . . . . 43 136 11.3. Message Integrity . . . . . . . . . . . . . . . . . . . 43 137 11.4. Message Confidentiality . . . . . . . . . . . . . . . . 43 138 12. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 44 139 12.1. Field Name Registration . . . . . . . . . . . . . . . . 44 140 12.2. Media Type Registration . . . . . . . . . . . . . . . . 44 141 12.3. Transfer Coding Registration . . . . . . . . . . . . . . 44 142 12.4. Upgrade Token Registration . . . . . . . . . . . . . . . 44 143 12.5. ALPN Protocol ID Registration . . . . . . . . . . . . . 44 144 13. References . . . . . . . . . . . . . . . . . . . . . . . . . 45 145 13.1. Normative References . . . . . . . . . . . . . . . . . . 45 146 13.2. Informative References . . . . . . . . . . . . . . . . . 46 147 Appendix A. Collected ABNF . . . . . . . . . . . . . . . . . . . 48 148 Appendix B. Differences between HTTP and MIME . . . . . . . . . 49 149 B.1. MIME-Version . . . . . . . . . . . . . . . . . . . . . . 50 150 B.2. Conversion to Canonical Form . . . . . . . . . . . . . . 50 151 B.3. Conversion of Date Formats . . . . . . . . . . . . . . . 50 152 B.4. Conversion of Content-Encoding . . . . . . . . . . . . . 51 153 B.5. Conversion of Content-Transfer-Encoding . . . . . . . . . 51 154 B.6. MHTML and Line Length Limitations . . . . . . . . . . . . 51 155 Appendix C. HTTP Version History . . . . . . . . . . . . . . . . 51 156 C.1. Changes from HTTP/1.0 . . . . . . . . . . . . . . . . . . 52 157 C.1.1. Multihomed Web Servers . . . . . . . . . . . . . . . 52 158 C.1.2. Keep-Alive Connections . . . . . . . . . . . . . . . 53 159 C.1.3. Introduction of Transfer-Encoding . . . . . . . . . . 53 160 C.2. Changes from RFC 7230 . . . . . . . . . . . . . . . . . . 53 161 Appendix D. Change Log . . . . . . . . . . . . . . . . . . . . . 54 162 D.1. Between RFC7230 and draft 00 . . . . . . . . . . . . . . 54 163 D.2. Since draft-ietf-httpbis-messaging-00 . . . . . . . . . . 54 164 D.3. Since draft-ietf-httpbis-messaging-01 . . . . . . . . . . 55 165 D.4. Since draft-ietf-httpbis-messaging-02 . . . . . . . . . . 56 166 D.5. Since draft-ietf-httpbis-messaging-03 . . . . . . . . . . 56 167 D.6. Since draft-ietf-httpbis-messaging-04 . . . . . . . . . . 56 168 D.7. Since draft-ietf-httpbis-messaging-05 . . . . . . . . . . 56 169 D.8. Since draft-ietf-httpbis-messaging-06 . . . . . . . . . . 57 170 D.9. Since draft-ietf-httpbis-messaging-07 . . . . . . . . . . 57 171 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 172 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . 60 173 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 60 175 1. Introduction 177 The Hypertext Transfer Protocol (HTTP) is a stateless application- 178 level request/response protocol that uses extensible semantics and 179 self-descriptive messages for flexible interaction with network-based 180 hypertext information systems. HTTP is defined by a series of 181 documents that collectively form the HTTP/1.1 specification: 183 o "HTTP Semantics" [Semantics] 185 o "HTTP Caching" [Caching] 187 o "HTTP/1.1 Messaging" (this document) 189 This document defines HTTP/1.1 message syntax and framing 190 requirements and their associated connection management. Our goal is 191 to define all of the mechanisms necessary for HTTP/1.1 message 192 handling that are independent of message semantics, thereby defining 193 the complete set of requirements for message parsers and message- 194 forwarding intermediaries. 196 This document obsoletes the portions of RFC 7230 related to HTTP/1.1 197 messaging and connection management, with the changes being 198 summarized in Appendix C.2. The other parts of RFC 7230 are 199 obsoleted by "HTTP Semantics" [Semantics]. 201 1.1. Requirements Notation 203 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 204 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 205 "OPTIONAL" in this document are to be interpreted as described in BCP 206 14 [RFC2119] [RFC8174] when, and only when, they appear in all 207 capitals, as shown here. 209 Conformance criteria and considerations regarding error handling are 210 defined in Section 3 of [Semantics]. 212 1.2. Syntax Notation 214 This specification uses the Augmented Backus-Naur Form (ABNF) 215 notation of [RFC5234], extended with the notation for case- 216 sensitivity in strings defined in [RFC7405]. 218 It also uses a list extension, defined in Section 4.5 of [Semantics], 219 that allows for compact definition of comma-separated lists using a 220 '#' operator (similar to how the '*' operator indicates repetition). 221 Appendix A shows the collected grammar with all list operators 222 expanded to standard ABNF notation. 224 As a convention, ABNF rule names prefixed with "obs-" denote 225 "obsolete" grammar rules that appear for historical reasons. 227 The following core rules are included by reference, as defined in 228 [RFC5234], Appendix B.1: ALPHA (letters), CR (carriage return), CRLF 229 (CR LF), CTL (controls), DIGIT (decimal 0-9), DQUOTE (double quote), 230 HEXDIG (hexadecimal 0-9/A-F/a-f), HTAB (horizontal tab), LF (line 231 feed), OCTET (any 8-bit sequence of data), SP (space), and VCHAR (any 232 visible [USASCII] character). 234 The rules below are defined in [Semantics]: 236 BWS = 237 OWS = 238 RWS = 239 absolute-URI = 240 absolute-path = 241 authority = 242 comment = 243 field-name = 244 field-value = 245 obs-text = 246 port = 247 query = 248 quoted-string = 249 token = 250 uri-host = 252 2. Message 254 2.1. Message Format 256 An HTTP/1.1 message consists of a start-line followed by a CRLF and a 257 sequence of octets in a format similar to the Internet Message Format 258 [RFC5322]: zero or more header field lines (collectively referred to 259 as the "headers" or the "header section"), an empty line indicating 260 the end of the header section, and an optional message body. 262 HTTP-message = start-line CRLF 263 *( field-line CRLF ) 264 CRLF 265 [ message-body ] 267 A message can be either a request from client to server or a response 268 from server to client. Syntactically, the two types of message 269 differ only in the start-line, which is either a request-line (for 270 requests) or a status-line (for responses), and in the algorithm for 271 determining the length of the message body (Section 6). 273 start-line = request-line / status-line 275 In theory, a client could receive requests and a server could receive 276 responses, distinguishing them by their different start-line formats. 277 In practice, servers are implemented to only expect a request (a 278 response is interpreted as an unknown or invalid request method) and 279 clients are implemented to only expect a response. 281 Although HTTP makes use of some protocol elements similar to the 282 Multipurpose Internet Mail Extensions (MIME) [RFC2045], see 283 Appendix B for the differences between HTTP and MIME messages. 285 2.2. Message Parsing 287 The normal procedure for parsing an HTTP message is to read the 288 start-line into a structure, read each header field line into a hash 289 table by field name until the empty line, and then use the parsed 290 data to determine if a message body is expected. If a message body 291 has been indicated, then it is read as a stream until an amount of 292 octets equal to the message body length is read or the connection is 293 closed. 295 A recipient MUST parse an HTTP message as a sequence of octets in an 296 encoding that is a superset of US-ASCII [USASCII]. Parsing an HTTP 297 message as a stream of Unicode characters, without regard for the 298 specific encoding, creates security vulnerabilities due to the 299 varying ways that string processing libraries handle invalid 300 multibyte character sequences that contain the octet LF (%x0A). 301 String-based parsers can only be safely used within protocol elements 302 after the element has been extracted from the message, such as within 303 a header field line value after message parsing has delineated the 304 individual field lines. 306 Although the line terminator for the start-line and header fields is 307 the sequence CRLF, a recipient MAY recognize a single LF as a line 308 terminator and ignore any preceding CR. 310 Older HTTP/1.0 user agent implementations might send an extra CRLF 311 after a POST request as a workaround for some early server 312 applications that failed to read message body content that was not 313 terminated by a line-ending. An HTTP/1.1 user agent MUST NOT preface 314 or follow a request with an extra CRLF. If terminating the request 315 message body with a line-ending is desired, then the user agent MUST 316 count the terminating CRLF octets as part of the message body length. 318 In the interest of robustness, a server that is expecting to receive 319 and parse a request-line SHOULD ignore at least one empty line (CRLF) 320 received prior to the request-line. 322 A sender MUST NOT send whitespace between the start-line and the 323 first header field. A recipient that receives whitespace between the 324 start-line and the first header field MUST either reject the message 325 as invalid or consume each whitespace-preceded line without further 326 processing of it (i.e., ignore the entire line, along with any 327 subsequent lines preceded by whitespace, until a properly formed 328 header field is received or the header section is terminated). 330 The presence of such whitespace in a request might be an attempt to 331 trick a server into ignoring that field line or processing the line 332 after it as a new request, either of which might result in a security 333 vulnerability if other implementations within the request chain 334 interpret the same message differently. Likewise, the presence of 335 such whitespace in a response might be ignored by some clients or 336 cause others to cease parsing. 338 When a server listening only for HTTP request messages, or processing 339 what appears from the start-line to be an HTTP request message, 340 receives a sequence of octets that does not match the HTTP-message 341 grammar aside from the robustness exceptions listed above, the server 342 SHOULD respond with a 400 (Bad Request) response. 344 2.3. HTTP Version 346 HTTP uses a "." numbering scheme to indicate versions 347 of the protocol. This specification defines version "1.1". 348 Section 3.5 of [Semantics] specifies the semantics of HTTP version 349 numbers. 351 The version of an HTTP/1.x message is indicated by an HTTP-version 352 field in the start-line. HTTP-version is case-sensitive. 354 HTTP-version = HTTP-name "/" DIGIT "." DIGIT 355 HTTP-name = %s"HTTP" 357 When an HTTP/1.1 message is sent to an HTTP/1.0 recipient [RFC1945] 358 or a recipient whose version is unknown, the HTTP/1.1 message is 359 constructed such that it can be interpreted as a valid HTTP/1.0 360 message if all of the newer features are ignored. This specification 361 places recipient-version requirements on some new features so that a 362 conformant sender will only use compatible features until it has 363 determined, through configuration or the receipt of a message, that 364 the recipient supports HTTP/1.1. 366 Intermediaries that process HTTP messages (i.e., all intermediaries 367 other than those acting as tunnels) MUST send their own HTTP-version 368 in forwarded messages. In other words, they are not allowed to 369 blindly forward the start-line without ensuring that the protocol 370 version in that message matches a version to which that intermediary 371 is conformant for both the receiving and sending of messages. 372 Forwarding an HTTP message without rewriting the HTTP-version might 373 result in communication errors when downstream recipients use the 374 message sender's version to determine what features are safe to use 375 for later communication with that sender. 377 A server MAY send an HTTP/1.0 response to an HTTP/1.1 request if it 378 is known or suspected that the client incorrectly implements the HTTP 379 specification and is incapable of correctly processing later version 380 responses, such as when a client fails to parse the version number 381 correctly or when an intermediary is known to blindly forward the 382 HTTP-version even when it doesn't conform to the given minor version 383 of the protocol. Such protocol downgrades SHOULD NOT be performed 384 unless triggered by specific client attributes, such as when one or 385 more of the request header fields (e.g., User-Agent) uniquely match 386 the values sent by a client known to be in error. 388 3. Request Line 390 A request-line begins with a method token, followed by a single space 391 (SP), the request-target, another single space (SP), and ends with 392 the protocol version. 394 request-line = method SP request-target SP HTTP-version 396 Although the request-line grammar rule requires that each of the 397 component elements be separated by a single SP octet, recipients MAY 398 instead parse on whitespace-delimited word boundaries and, aside from 399 the CRLF terminator, treat any form of whitespace as the SP separator 400 while ignoring preceding or trailing whitespace; such whitespace 401 includes one or more of the following octets: SP, HTAB, VT (%x0B), FF 402 (%x0C), or bare CR. However, lenient parsing can result in request 403 smuggling security vulnerabilities if there are multiple recipients 404 of the message and each has its own unique interpretation of 405 robustness (see Section 11.2). 407 HTTP does not place a predefined limit on the length of a request- 408 line, as described in Section 3 of [Semantics]. A server that 409 receives a method longer than any that it implements SHOULD respond 410 with a 501 (Not Implemented) status code. A server that receives a 411 request-target longer than any URI it wishes to parse MUST respond 412 with a 414 (URI Too Long) status code (see Section 9.5.15 of 413 [Semantics]). 415 Various ad hoc limitations on request-line length are found in 416 practice. It is RECOMMENDED that all HTTP senders and recipients 417 support, at a minimum, request-line lengths of 8000 octets. 419 3.1. Method 421 The method token indicates the request method to be performed on the 422 target resource. The request method is case-sensitive. 424 method = token 426 The request methods defined by this specification can be found in 427 Section 7 of [Semantics], along with information regarding the HTTP 428 method registry and considerations for defining new methods. 430 3.2. Request Target 432 The request-target identifies the target resource upon which to apply 433 the request. The client derives a request-target from its desired 434 target URI. There are four distinct formats for the request-target, 435 depending on both the method being requested and whether the request 436 is to a proxy. 438 request-target = origin-form 439 / absolute-form 440 / authority-form 441 / asterisk-form 443 No whitespace is allowed in the request-target. Unfortunately, some 444 user agents fail to properly encode or exclude whitespace found in 445 hypertext references, resulting in those disallowed characters being 446 sent as the request-target in a malformed request-line. 448 Recipients of an invalid request-line SHOULD respond with either a 449 400 (Bad Request) error or a 301 (Moved Permanently) redirect with 450 the request-target properly encoded. A recipient SHOULD NOT attempt 451 to autocorrect and then process the request without a redirect, since 452 the invalid request-line might be deliberately crafted to bypass 453 security filters along the request chain. 455 3.2.1. origin-form 457 The most common form of request-target is the origin-form. 459 origin-form = absolute-path [ "?" query ] 461 When making a request directly to an origin server, other than a 462 CONNECT or server-wide OPTIONS request (as detailed below), a client 463 MUST send only the absolute path and query components of the target 464 URI as the request-target. If the target URI's path component is 465 empty, the client MUST send "/" as the path within the origin-form of 466 request-target. A Host header field is also sent, as defined in 467 Section 5.6 of [Semantics]. 469 For example, a client wishing to retrieve a representation of the 470 resource identified as 472 http://www.example.org/where?q=now 474 directly from the origin server would open (or reuse) a TCP 475 connection to port 80 of the host "www.example.org" and send the 476 lines: 478 GET /where?q=now HTTP/1.1 479 Host: www.example.org 481 followed by the remainder of the request message. 483 3.2.2. absolute-form 485 When making a request to a proxy, other than a CONNECT or server-wide 486 OPTIONS request (as detailed below), a client MUST send the target 487 URI in absolute-form as the request-target. 489 absolute-form = absolute-URI 491 The proxy is requested to either service that request from a valid 492 cache, if possible, or make the same request on the client's behalf 493 to either the next inbound proxy server or directly to the origin 494 server indicated by the request-target. Requirements on such 495 "forwarding" of messages are defined in Section 5.7 of [Semantics]. 497 An example absolute-form of request-line would be: 499 GET http://www.example.org/pub/WWW/TheProject.html HTTP/1.1 501 A client MUST send a Host header field in an HTTP/1.1 request even if 502 the request-target is in the absolute-form, since this allows the 503 Host information to be forwarded through ancient HTTP/1.0 proxies 504 that might not have implemented Host. 506 When a proxy receives a request with an absolute-form of request- 507 target, the proxy MUST ignore the received Host header field (if any) 508 and instead replace it with the host information of the request- 509 target. A proxy that forwards such a request MUST generate a new 510 Host field value based on the received request-target rather than 511 forward the received Host field value. 513 When an origin server receives a request with an absolute-form of 514 request-target, the origin server MUST ignore the received Host 515 header field (if any) and instead use the host information of the 516 request-target. Note that if the request-target does not have an 517 authority component, an empty Host header field will be sent in this 518 case. 520 To allow for transition to the absolute-form for all requests in some 521 future version of HTTP, a server MUST accept the absolute-form in 522 requests, even though HTTP/1.1 clients will only send them in 523 requests to proxies. 525 3.2.3. authority-form 527 The authority-form of request-target is only used for CONNECT 528 requests (Section 7.3.6 of [Semantics]). 530 authority-form = authority 532 When making a CONNECT request to establish a tunnel through one or 533 more proxies, a client MUST send only the target URI's authority 534 component (excluding any userinfo and its "@" delimiter) as the 535 request-target. For example, 537 CONNECT www.example.com:80 HTTP/1.1 539 3.2.4. asterisk-form 541 The asterisk-form of request-target is only used for a server-wide 542 OPTIONS request (Section 7.3.7 of [Semantics]). 544 asterisk-form = "*" 546 When a client wishes to request OPTIONS for the server as a whole, as 547 opposed to a specific named resource of that server, the client MUST 548 send only "*" (%x2A) as the request-target. For example, 550 OPTIONS * HTTP/1.1 552 If a proxy receives an OPTIONS request with an absolute-form of 553 request-target in which the URI has an empty path and no query 554 component, then the last proxy on the request chain MUST send a 555 request-target of "*" when it forwards the request to the indicated 556 origin server. 558 For example, the request 560 OPTIONS http://www.example.org:8001 HTTP/1.1 562 would be forwarded by the final proxy as 564 OPTIONS * HTTP/1.1 565 Host: www.example.org:8001 567 after connecting to port 8001 of host "www.example.org". 569 3.3. Reconstructing the Target URI 571 Since the request-target often contains only part of the user agent's 572 target URI, a server constructs its value to properly service the 573 request (Section 5.1 of [Semantics]). 575 If the request-target is in absolute-form, the target URI is the same 576 as the request-target. Otherwise, the target URI is constructed as 577 follows: 579 If the server's configuration (or outbound gateway) provides a 580 fixed URI scheme, that scheme is used for the target URI. 581 Otherwise, if the request is received over a TLS-secured TCP 582 connection, the target URI's scheme is "https"; if not, the scheme 583 is "http". 585 If the server's configuration (or outbound gateway) provides a 586 fixed URI authority component, that authority is used for the 587 target URI. If not, then if the request-target is in authority- 588 form, the target URI's authority component is the same as the 589 request-target. If not, then if a Host header field is supplied 590 with a non-empty field-value, the authority component is the same 591 as the Host field-value. Otherwise, the authority component is 592 assigned the default name configured for the server and, if the 593 connection's incoming TCP port number differs from the default 594 port for the target URI's scheme, then a colon (":") and the 595 incoming port number (in decimal form) are appended to the 596 authority component. 598 If the request-target is in authority-form or asterisk-form, the 599 target URI's combined path and query component is empty. 600 Otherwise, the combined path and query component is the same as 601 the request-target. 603 The components of the target URI, once determined as above, can be 604 combined into absolute-URI form by concatenating the scheme, 605 "://", authority, and combined path and query component. 607 Example 1: the following message received over an insecure TCP 608 connection 610 GET /pub/WWW/TheProject.html HTTP/1.1 611 Host: www.example.org:8080 613 has a target URI of 615 http://www.example.org:8080/pub/WWW/TheProject.html 617 Example 2: the following message received over a TLS-secured TCP 618 connection 620 OPTIONS * HTTP/1.1 621 Host: www.example.org 623 has a target URI of 625 https://www.example.org 627 Recipients of an HTTP/1.0 request that lacks a Host header field 628 might need to use heuristics (e.g., examination of the URI path for 629 something unique to a particular host) in order to guess the target 630 URI's authority component. 632 4. Status Line 634 The first line of a response message is the status-line, consisting 635 of the protocol version, a space (SP), the status code, another 636 space, and ending with an OPTIONAL textual phrase describing the 637 status code. 639 status-line = HTTP-version SP status-code SP [reason-phrase] 641 Although the status-line grammar rule requires that each of the 642 component elements be separated by a single SP octet, recipients MAY 643 instead parse on whitespace-delimited word boundaries and, aside from 644 the line terminator, treat any form of whitespace as the SP separator 645 while ignoring preceding or trailing whitespace; such whitespace 646 includes one or more of the following octets: SP, HTAB, VT (%x0B), FF 647 (%x0C), or bare CR. However, lenient parsing can result in response 648 splitting security vulnerabilities if there are multiple recipients 649 of the message and each has its own unique interpretation of 650 robustness (see Section 11.1). 652 The status-code element is a 3-digit integer code describing the 653 result of the server's attempt to understand and satisfy the client's 654 corresponding request. The rest of the response message is to be 655 interpreted in light of the semantics defined for that status code. 656 See Section 9 of [Semantics] for information about the semantics of 657 status codes, including the classes of status code (indicated by the 658 first digit), the status codes defined by this specification, 659 considerations for the definition of new status codes, and the IANA 660 registry. 662 status-code = 3DIGIT 664 The reason-phrase element exists for the sole purpose of providing a 665 textual description associated with the numeric status code, mostly 666 out of deference to earlier Internet application protocols that were 667 more frequently used with interactive text clients. 669 reason-phrase = 1*( HTAB / SP / VCHAR / obs-text ) 671 A client SHOULD ignore the reason-phrase content because it is not a 672 reliable channel for information (it might be translated for a given 673 locale, overwritten by intermediaries, or discarded when the message 674 is forwarded via other versions of HTTP). A server MUST send the 675 space that separates status-code from the reason-phrase even when the 676 reason-phrase is absent (i.e., the status-line would end with the 677 three octets SP CR LF). 679 5. Field Syntax 681 Each field line consists of a case-insensitive field name followed by 682 a colon (":"), optional leading whitespace, the field line value, and 683 optional trailing whitespace. 685 field-line = field-name ":" OWS field-value OWS 687 Most HTTP field names and the rules for parsing within field values 688 are defined in Section 4 of [Semantics]. This section covers the 689 generic syntax for header field inclusion within, and extraction 690 from, HTTP/1.1 messages. In addition, the following header fields 691 are defined by this document because they are specific to HTTP/1.1 692 message processing: 694 +-------------------+----------+---------------+ 695 | Field Name | Status | Reference | 696 +-------------------+----------+---------------+ 697 | Connection | standard | Section 9.1 | 698 | MIME-Version | standard | Appendix B.1 | 699 | TE | standard | Section 7.4 | 700 | Transfer-Encoding | standard | Section 6.1 | 701 | Upgrade | standard | Section 9.9 | 702 +-------------------+----------+---------------+ 704 Table 1 706 Furthermore, the field name "Close" is reserved, since using that 707 name as an HTTP header field might conflict with the "close" 708 connection option of the Connection header field (Section 9.1). 710 +-------------------+----------+----------+------------+ 711 | Header Field Name | Protocol | Status | Reference | 712 +-------------------+----------+----------+------------+ 713 | Close | http | reserved | Section 5 | 714 +-------------------+----------+----------+------------+ 716 5.1. Field Line Parsing 718 Messages are parsed using a generic algorithm, independent of the 719 individual field names. The contents within a given field line value 720 are not parsed until a later stage of message interpretation (usually 721 after the message's entire header section has been processed). 723 No whitespace is allowed between the field name and colon. In the 724 past, differences in the handling of such whitespace have led to 725 security vulnerabilities in request routing and response handling. A 726 server MUST reject any received request message that contains 727 whitespace between a header field name and colon with a response 728 status code of 400 (Bad Request). A proxy MUST remove any such 729 whitespace from a response message before forwarding the message 730 downstream. 732 A field line value might be preceded and/or followed by optional 733 whitespace (OWS); a single SP preceding the field line value is 734 preferred for consistent readability by humans. The field line value 735 does not include any leading or trailing whitespace: OWS occurring 736 before the first non-whitespace octet of the field line value or 737 after the last non-whitespace octet of the field line value ought to 738 be excluded by parsers when extracting the field line value from a 739 header field line. 741 5.2. Obsolete Line Folding 743 Historically, HTTP header field line values could be extended over 744 multiple lines by preceding each extra line with at least one space 745 or horizontal tab (obs-fold). This specification deprecates such 746 line folding except within the message/http media type 747 (Section 10.1). 749 obs-fold = OWS CRLF RWS 750 ; obsolete line folding 752 A sender MUST NOT generate a message that includes line folding 753 (i.e., that has any field line value that contains a match to the 754 obs-fold rule) unless the message is intended for packaging within 755 the message/http media type. 757 A server that receives an obs-fold in a request message that is not 758 within a message/http container MUST either reject the message by 759 sending a 400 (Bad Request), preferably with a representation 760 explaining that obsolete line folding is unacceptable, or replace 761 each received obs-fold with one or more SP octets prior to 762 interpreting the field value or forwarding the message downstream. 764 A proxy or gateway that receives an obs-fold in a response message 765 that is not within a message/http container MUST either discard the 766 message and replace it with a 502 (Bad Gateway) response, preferably 767 with a representation explaining that unacceptable line folding was 768 received, or replace each received obs-fold with one or more SP 769 octets prior to interpreting the field value or forwarding the 770 message downstream. 772 A user agent that receives an obs-fold in a response message that is 773 not within a message/http container MUST replace each received obs- 774 fold with one or more SP octets prior to interpreting the field 775 value. 777 6. Message Body 779 The message body (if any) of an HTTP message is used to carry the 780 payload body (Section 6.3.3 of [Semantics]) of that request or 781 response. The message body is identical to the payload body unless a 782 transfer coding has been applied, as described in Section 6.1. 784 message-body = *OCTET 786 The rules for determining when a message body is present in an 787 HTTP/1.1 message differ for requests and responses. 789 The presence of a message body in a request is signaled by a Content- 790 Length or Transfer-Encoding header field. Request message framing is 791 independent of method semantics, even if the method does not define 792 any use for a message body. 794 The presence of a message body in a response depends on both the 795 request method to which it is responding and the response status code 796 (Section 4), and corresponds to when a payload body is allowed; see 797 Section 6.3.3 of [Semantics]. 799 6.1. Transfer-Encoding 801 The Transfer-Encoding header field lists the transfer coding names 802 corresponding to the sequence of transfer codings that have been (or 803 will be) applied to the payload body in order to form the message 804 body. Transfer codings are defined in Section 7. 806 Transfer-Encoding = 1#transfer-coding 808 Transfer-Encoding is analogous to the Content-Transfer-Encoding field 809 of MIME, which was designed to enable safe transport of binary data 810 over a 7-bit transport service ([RFC2045], Section 6). However, safe 811 transport has a different focus for an 8bit-clean transfer protocol. 812 In HTTP's case, Transfer-Encoding is primarily intended to accurately 813 delimit a dynamically generated payload and to distinguish payload 814 encodings that are only applied for transport efficiency or security 815 from those that are characteristics of the selected resource. 817 A recipient MUST be able to parse the chunked transfer coding 818 (Section 7.1) because it plays a crucial role in framing messages 819 when the payload body size is not known in advance. A sender MUST 820 NOT apply chunked more than once to a message body (i.e., chunking an 821 already chunked message is not allowed). If any transfer coding 822 other than chunked is applied to a request payload body, the sender 823 MUST apply chunked as the final transfer coding to ensure that the 824 message is properly framed. If any transfer coding other than 825 chunked is applied to a response payload body, the sender MUST either 826 apply chunked as the final transfer coding or terminate the message 827 by closing the connection. 829 For example, 831 Transfer-Encoding: gzip, chunked 833 indicates that the payload body has been compressed using the gzip 834 coding and then chunked using the chunked coding while forming the 835 message body. 837 Unlike Content-Encoding (Section 6.1.2 of [Semantics]), Transfer- 838 Encoding is a property of the message, not of the representation, and 839 any recipient along the request/response chain MAY decode the 840 received transfer coding(s) or apply additional transfer coding(s) to 841 the message body, assuming that corresponding changes are made to the 842 Transfer-Encoding field value. Additional information about the 843 encoding parameters can be provided by other header fields not 844 defined by this specification. 846 Transfer-Encoding MAY be sent in a response to a HEAD request or in a 847 304 (Not Modified) response (Section 9.4.5 of [Semantics]) to a GET 848 request, neither of which includes a message body, to indicate that 849 the origin server would have applied a transfer coding to the message 850 body if the request had been an unconditional GET. This indication 851 is not required, however, because any recipient on the response chain 852 (including the origin server) can remove transfer codings when they 853 are not needed. 855 A server MUST NOT send a Transfer-Encoding header field in any 856 response with a status code of 1xx (Informational) or 204 (No 857 Content). A server MUST NOT send a Transfer-Encoding header field in 858 any 2xx (Successful) response to a CONNECT request (Section 7.3.6 of 859 [Semantics]). 861 Transfer-Encoding was added in HTTP/1.1. It is generally assumed 862 that implementations advertising only HTTP/1.0 support will not 863 understand how to process a transfer-encoded payload. A client MUST 864 NOT send a request containing Transfer-Encoding unless it knows the 865 server will handle HTTP/1.1 requests (or later minor revisions); such 866 knowledge might be in the form of specific user configuration or by 867 remembering the version of a prior received response. A server MUST 868 NOT send a response containing Transfer-Encoding unless the 869 corresponding request indicates HTTP/1.1 (or later minor revisions). 871 A server that receives a request message with a transfer coding it 872 does not understand SHOULD respond with 501 (Not Implemented). 874 6.2. Content-Length 876 When a message does not have a Transfer-Encoding header field, a 877 Content-Length header field can provide the anticipated size, as a 878 decimal number of octets, for a potential payload body. For messages 879 that do include a payload body, the Content-Length field value 880 provides the framing information necessary for determining where the 881 body (and message) ends. For messages that do not include a payload 882 body, the Content-Length indicates the size of the selected 883 representation (Section 6.2.4 of [Semantics]). 885 Note: HTTP's use of Content-Length for message framing differs 886 significantly from the same field's use in MIME, where it is an 887 optional field used only within the "message/external-body" media- 888 type. 890 6.3. Message Body Length 892 The length of a message body is determined by one of the following 893 (in order of precedence): 895 1. Any response to a HEAD request and any response with a 1xx 896 (Informational), 204 (No Content), or 304 (Not Modified) status 897 code is always terminated by the first empty line after the 898 header fields, regardless of the header fields present in the 899 message, and thus cannot contain a message body. 901 2. Any 2xx (Successful) response to a CONNECT request implies that 902 the connection will become a tunnel immediately after the empty 903 line that concludes the header fields. A client MUST ignore any 904 Content-Length or Transfer-Encoding header fields received in 905 such a message. 907 3. If a Transfer-Encoding header field is present and the chunked 908 transfer coding (Section 7.1) is the final encoding, the message 909 body length is determined by reading and decoding the chunked 910 data until the transfer coding indicates the data is complete. 912 If a Transfer-Encoding header field is present in a response and 913 the chunked transfer coding is not the final encoding, the 914 message body length is determined by reading the connection until 915 it is closed by the server. If a Transfer-Encoding header field 916 is present in a request and the chunked transfer coding is not 917 the final encoding, the message body length cannot be determined 918 reliably; the server MUST respond with the 400 (Bad Request) 919 status code and then close the connection. 921 If a message is received with both a Transfer-Encoding and a 922 Content-Length header field, the Transfer-Encoding overrides the 923 Content-Length. Such a message might indicate an attempt to 924 perform request smuggling (Section 11.2) or response splitting 925 (Section 11.1) and ought to be handled as an error. A sender 926 MUST remove the received Content-Length field prior to forwarding 927 such a message downstream. 929 4. If a message is received without Transfer-Encoding and with an 930 invalid Content-Length header field, then the message framing is 931 invalid and the recipient MUST treat it as an unrecoverable 932 error, unless the field value can be successfully parsed as a 933 comma-separated list (Section 4.5 of [Semantics]), all values in 934 the list are valid, and all values in the list are the same. If 935 this is a request message, the server MUST respond with a 400 936 (Bad Request) status code and then close the connection. If this 937 is a response message received by a proxy, the proxy MUST close 938 the connection to the server, discard the received response, and 939 send a 502 (Bad Gateway) response to the client. If this is a 940 response message received by a user agent, the user agent MUST 941 close the connection to the server and discard the received 942 response. 944 5. If a valid Content-Length header field is present without 945 Transfer-Encoding, its decimal value defines the expected message 946 body length in octets. If the sender closes the connection or 947 the recipient times out before the indicated number of octets are 948 received, the recipient MUST consider the message to be 949 incomplete and close the connection. 951 6. If this is a request message and none of the above are true, then 952 the message body length is zero (no message body is present). 954 7. Otherwise, this is a response message without a declared message 955 body length, so the message body length is determined by the 956 number of octets received prior to the server closing the 957 connection. 959 Since there is no way to distinguish a successfully completed, close- 960 delimited message from a partially received message interrupted by 961 network failure, a server SHOULD generate encoding or length- 962 delimited messages whenever possible. The close-delimiting feature 963 exists primarily for backwards compatibility with HTTP/1.0. 965 A server MAY reject a request that contains a message body but not a 966 Content-Length by responding with 411 (Length Required). 968 Unless a transfer coding other than chunked has been applied, a 969 client that sends a request containing a message body SHOULD use a 970 valid Content-Length header field if the message body length is known 971 in advance, rather than the chunked transfer coding, since some 972 existing services respond to chunked with a 411 (Length Required) 973 status code even though they understand the chunked transfer coding. 974 This is typically because such services are implemented via a gateway 975 that requires a content-length in advance of being called and the 976 server is unable or unwilling to buffer the entire request before 977 processing. 979 A user agent that sends a request containing a message body MUST send 980 a valid Content-Length header field if it does not know the server 981 will handle HTTP/1.1 (or later) requests; such knowledge can be in 982 the form of specific user configuration or by remembering the version 983 of a prior received response. 985 If the final response to the last request on a connection has been 986 completely received and there remains additional data to read, a user 987 agent MAY discard the remaining data or attempt to determine if that 988 data belongs as part of the prior response body, which might be the 989 case if the prior message's Content-Length value is incorrect. A 990 client MUST NOT process, cache, or forward such extra data as a 991 separate response, since such behavior would be vulnerable to cache 992 poisoning. 994 7. Transfer Codings 996 Transfer coding names are used to indicate an encoding transformation 997 that has been, can be, or might need to be applied to a payload body 998 in order to ensure "safe transport" through the network. This 999 differs from a content coding in that the transfer coding is a 1000 property of the message rather than a property of the representation 1001 that is being transferred. 1003 transfer-coding = token *( OWS ";" OWS transfer-parameter ) 1005 Parameters are in the form of a name=value pair. 1007 transfer-parameter = token BWS "=" BWS ( token / quoted-string ) 1009 All transfer-coding names are case-insensitive and ought to be 1010 registered within the HTTP Transfer Coding registry, as defined in 1011 Section 7.3. They are used in the TE (Section 7.4) and Transfer- 1012 Encoding (Section 6.1) header fields. 1014 +------------+------------------------------------------+-----------+ 1015 | Name | Description | Reference | 1016 +------------+------------------------------------------+-----------+ 1017 | chunked | Transfer in a series of chunks | Section 7 | 1018 | | | .1 | 1019 | compress | UNIX "compress" data format [Welch] | Section 7 | 1020 | | | .2 | 1021 | deflate | "deflate" compressed data ([RFC1951]) | Section 7 | 1022 | | inside the "zlib" data format | .2 | 1023 | | ([RFC1950]) | | 1024 | gzip | GZIP file format [RFC1952] | Section 7 | 1025 | | | .2 | 1026 | trailers | (reserved) | Section 7 | 1027 | x-compress | Deprecated (alias for compress) | Section 7 | 1028 | | | .2 | 1029 | x-gzip | Deprecated (alias for gzip) | Section 7 | 1030 | | | .2 | 1031 +------------+------------------------------------------+-----------+ 1033 Table 2 1035 Note: the coding name "trailers" is reserved because its use would 1036 conflict with the keyword "trailers" in the TE header field 1037 (Section 7.4). 1039 7.1. Chunked Transfer Coding 1041 The chunked transfer coding wraps the payload body in order to 1042 transfer it as a series of chunks, each with its own size indicator, 1043 followed by an OPTIONAL trailer section containing trailer fields. 1044 Chunked enables content streams of unknown size to be transferred as 1045 a sequence of length-delimited buffers, which enables the sender to 1046 retain connection persistence and the recipient to know when it has 1047 received the entire message. 1049 chunked-body = *chunk 1050 last-chunk 1051 trailer-section 1052 CRLF 1054 chunk = chunk-size [ chunk-ext ] CRLF 1055 chunk-data CRLF 1056 chunk-size = 1*HEXDIG 1057 last-chunk = 1*("0") [ chunk-ext ] CRLF 1059 chunk-data = 1*OCTET ; a sequence of chunk-size octets 1061 The chunk-size field is a string of hex digits indicating the size of 1062 the chunk-data in octets. The chunked transfer coding is complete 1063 when a chunk with a chunk-size of zero is received, possibly followed 1064 by a trailer section, and finally terminated by an empty line. 1066 A recipient MUST be able to parse and decode the chunked transfer 1067 coding. 1069 Note that HTTP/1.1 does not define any means to limit the size of a 1070 chunked response such that an intermediary can be assured of 1071 buffering the entire response. 1073 The chunked encoding does not define any parameters. Their presence 1074 SHOULD be treated as an error. 1076 7.1.1. Chunk Extensions 1078 The chunked encoding allows each chunk to include zero or more chunk 1079 extensions, immediately following the chunk-size, for the sake of 1080 supplying per-chunk metadata (such as a signature or hash), mid- 1081 message control information, or randomization of message body size. 1083 chunk-ext = *( BWS ";" BWS chunk-ext-name 1084 [ BWS "=" BWS chunk-ext-val ] ) 1086 chunk-ext-name = token 1087 chunk-ext-val = token / quoted-string 1089 The chunked encoding is specific to each connection and is likely to 1090 be removed or recoded by each recipient (including intermediaries) 1091 before any higher-level application would have a chance to inspect 1092 the extensions. Hence, use of chunk extensions is generally limited 1093 to specialized HTTP services such as "long polling" (where client and 1094 server can have shared expectations regarding the use of chunk 1095 extensions) or for padding within an end-to-end secured connection. 1097 A recipient MUST ignore unrecognized chunk extensions. A server 1098 ought to limit the total length of chunk extensions received in a 1099 request to an amount reasonable for the services provided, in the 1100 same way that it applies length limitations and timeouts for other 1101 parts of a message, and generate an appropriate 4xx (Client Error) 1102 response if that amount is exceeded. 1104 7.1.2. Chunked Trailer Section 1106 A trailer section allows the sender to include additional fields at 1107 the end of a chunked message in order to supply metadata that might 1108 be dynamically generated while the message body is sent, such as a 1109 message integrity check, digital signature, or post-processing 1110 status. The proper use and limitations of trailer fields are defined 1111 in Section 4.6 of [Semantics]. 1113 trailer-section = *( field-line CRLF ) 1115 A recipient that decodes and removes the chunked encoding from a 1116 message (e.g., for storage or forwarding to a non-HTTP/1.1 peer) MUST 1117 discard any received trailer fields, store/forward them separately 1118 from the header fields, or selectively merge into the header section 1119 only those trailer fields corresponding to header field definitions 1120 that are understood by the recipient to explicitly permit and define 1121 how their corresponding trailer field value can be safely merged. 1123 7.1.3. Decoding Chunked 1125 A process for decoding the chunked transfer coding can be represented 1126 in pseudo-code as: 1128 length := 0 1129 read chunk-size, chunk-ext (if any), and CRLF 1130 while (chunk-size > 0) { 1131 read chunk-data and CRLF 1132 append chunk-data to decoded-body 1133 length := length + chunk-size 1134 read chunk-size, chunk-ext (if any), and CRLF 1135 } 1136 read trailer field 1137 while (trailer field is not empty) { 1138 if (trailer fields are stored/forwarded separately) { 1139 append trailer field to existing trailer fields 1140 } 1141 else if (trailer field is understood and defined as mergeable) { 1142 merge trailer field with existing header fields 1143 } 1144 else { 1145 discard trailer field 1146 } 1147 read trailer field 1148 } 1149 Content-Length := length 1150 Remove "chunked" from Transfer-Encoding 1151 Remove Trailer from existing header fields 1153 7.2. Transfer Codings for Compression 1155 The following transfer coding names for compression are defined by 1156 the same algorithm as their corresponding content coding: 1158 compress (and x-compress) 1159 See Section 6.1.2.1 of [Semantics]. 1161 deflate 1162 See Section 6.1.2.2 of [Semantics]. 1164 gzip (and x-gzip) 1165 See Section 6.1.2.3 of [Semantics]. 1167 The compression codings do not define any parameters. Their presence 1168 SHOULD be treated as an error. 1170 7.3. Transfer Coding Registry 1172 The "HTTP Transfer Coding Registry" defines the namespace for 1173 transfer coding names. It is maintained at 1174 . 1176 Registrations MUST include the following fields: 1178 o Name 1180 o Description 1182 o Pointer to specification text 1184 Names of transfer codings MUST NOT overlap with names of content 1185 codings (Section 6.1.2 of [Semantics]) unless the encoding 1186 transformation is identical, as is the case for the compression 1187 codings defined in Section 7.2. 1189 The TE header field (Section 7.4) uses a pseudo parameter named "q" 1190 as rank value when multiple transfer codings are acceptable. Future 1191 registrations of transfer codings SHOULD NOT define parameters called 1192 "q" (case-insensitively) in order to avoid ambiguities. 1194 Values to be added to this namespace require IETF Review (see 1195 Section 4.8 of [RFC8126]), and MUST conform to the purpose of 1196 transfer coding defined in this specification. 1198 Use of program names for the identification of encoding formats is 1199 not desirable and is discouraged for future encodings. 1201 7.4. TE 1203 The "TE" header field in a request indicates what transfer codings, 1204 besides chunked, the client is willing to accept in response, and 1205 whether or not the client is willing to accept trailer fields in a 1206 chunked transfer coding. 1208 The TE field-value consists of a list of transfer coding names, each 1209 allowing for optional parameters (as described in Section 7), and/or 1210 the keyword "trailers". A client MUST NOT send the chunked transfer 1211 coding name in TE; chunked is always acceptable for HTTP/1.1 1212 recipients. 1214 TE = #t-codings 1215 t-codings = "trailers" / ( transfer-coding [ t-ranking ] ) 1216 t-ranking = OWS ";" OWS "q=" rank 1217 rank = ( "0" [ "." 0*3DIGIT ] ) 1218 / ( "1" [ "." 0*3("0") ] ) 1220 Three examples of TE use are below. 1222 TE: deflate 1223 TE: 1224 TE: trailers, deflate;q=0.5 1226 When multiple transfer codings are acceptable, the client MAY rank 1227 the codings by preference using a case-insensitive "q" parameter 1228 (similar to the qvalues used in content negotiation fields, 1229 Section 6.4.4 of [Semantics]). The rank value is a real number in 1230 the range 0 through 1, where 0.001 is the least preferred and 1 is 1231 the most preferred; a value of 0 means "not acceptable". 1233 If the TE field value is empty or if no TE field is present, the only 1234 acceptable transfer coding is chunked. A message with no transfer 1235 coding is always acceptable. 1237 The keyword "trailers" indicates that the sender will not discard 1238 trailer fields, as described in Section 4.6 of [Semantics]. 1240 Since the TE header field only applies to the immediate connection, a 1241 sender of TE MUST also send a "TE" connection option within the 1242 Connection header field (Section 9.1) in order to prevent the TE 1243 field from being forwarded by intermediaries that do not support its 1244 semantics. 1246 8. Handling Incomplete Messages 1248 A server that receives an incomplete request message, usually due to 1249 a canceled request or a triggered timeout exception, MAY send an 1250 error response prior to closing the connection. 1252 A client that receives an incomplete response message, which can 1253 occur when a connection is closed prematurely or when decoding a 1254 supposedly chunked transfer coding fails, MUST record the message as 1255 incomplete. Cache requirements for incomplete responses are defined 1256 in Section 3 of [Caching]. 1258 If a response terminates in the middle of the header section (before 1259 the empty line is received) and the status code might rely on header 1260 fields to convey the full meaning of the response, then the client 1261 cannot assume that meaning has been conveyed; the client might need 1262 to repeat the request in order to determine what action to take next. 1264 A message body that uses the chunked transfer coding is incomplete if 1265 the zero-sized chunk that terminates the encoding has not been 1266 received. A message that uses a valid Content-Length is incomplete 1267 if the size of the message body received (in octets) is less than the 1268 value given by Content-Length. A response that has neither chunked 1269 transfer coding nor Content-Length is terminated by closure of the 1270 connection and, thus, is considered complete regardless of the number 1271 of message body octets received, provided that the header section was 1272 received intact. 1274 9. Connection Management 1276 HTTP messaging is independent of the underlying transport- or 1277 session-layer connection protocol(s). HTTP only presumes a reliable 1278 transport with in-order delivery of requests and the corresponding 1279 in-order delivery of responses. The mapping of HTTP request and 1280 response structures onto the data units of an underlying transport 1281 protocol is outside the scope of this specification. 1283 As described in Section 5.3 of [Semantics], the specific connection 1284 protocols to be used for an HTTP interaction are determined by client 1285 configuration and the target URI. For example, the "http" URI scheme 1286 (Section 2.5.1 of [Semantics]) indicates a default connection of TCP 1287 over IP, with a default TCP port of 80, but the client might be 1288 configured to use a proxy via some other connection, port, or 1289 protocol. 1291 HTTP implementations are expected to engage in connection management, 1292 which includes maintaining the state of current connections, 1293 establishing a new connection or reusing an existing connection, 1294 processing messages received on a connection, detecting connection 1295 failures, and closing each connection. Most clients maintain 1296 multiple connections in parallel, including more than one connection 1297 per server endpoint. Most servers are designed to maintain thousands 1298 of concurrent connections, while controlling request queues to enable 1299 fair use and detect denial-of-service attacks. 1301 9.1. Connection 1303 The "Connection" header field allows the sender to list desired 1304 control options for the current connection. 1306 When a field aside from Connection is used to supply control 1307 information for or about the current connection, the sender MUST list 1308 the corresponding field name within the Connection header field. 1310 Intermediaries MUST parse a received Connection header field before a 1311 message is forwarded and, for each connection-option in this field, 1312 remove any header or trailer field(s) from the message with the same 1313 name as the connection-option, and then remove the Connection header 1314 field itself (or replace it with the intermediary's own connection 1315 options for the forwarded message). 1317 Hence, the Connection header field provides a declarative way of 1318 distinguishing fields that are only intended for the immediate 1319 recipient ("hop-by-hop") from those fields that are intended for all 1320 recipients on the chain ("end-to-end"), enabling the message to be 1321 self-descriptive and allowing future connection-specific extensions 1322 to be deployed without fear that they will be blindly forwarded by 1323 older intermediaries. 1325 Furthermore, intermediaries SHOULD remove or replace field(s) whose 1326 semantics are known to require removal before forwarding, whether or 1327 not they appear as a Connection option, after applying those fields' 1328 semantics. This includes but is not limited to: 1330 o Proxy-Connection Appendix C.1.2 1332 o Keep-Alive Section 19.7.1 of [RFC2068] 1334 o TE Section 7.4 1336 o Trailer Section 4.6.3 of [Semantics] 1338 o Transfer-Encoding Section 6.1 1340 o Upgrade Section 9.9 1342 The Connection header field's value has the following grammar: 1344 Connection = 1#connection-option 1345 connection-option = token 1347 Connection options are case-insensitive. 1349 A sender MUST NOT send a connection option corresponding to a field 1350 that is intended for all recipients of the payload. For example, 1351 Cache-Control is never appropriate as a connection option 1352 (Section 5.2 of [Caching]). 1354 The connection options do not always correspond to a field present in 1355 the message, since a connection-specific field might not be needed if 1356 there are no parameters associated with a connection option. In 1357 contrast, a connection-specific field that is received without a 1358 corresponding connection option usually indicates that the field has 1359 been improperly forwarded by an intermediary and ought to be ignored 1360 by the recipient. 1362 When defining new connection options, specification authors ought to 1363 document it as reserved field name and register that definition in 1364 the Hypertext Transfer Protocol (HTTP) Field Name Registry 1365 (Section 4.3.2 of [Semantics]), to avoid collisions. 1367 The "close" connection option is defined for a sender to signal that 1368 this connection will be closed after completion of the response. For 1369 example, 1371 Connection: close 1373 in either the request or the response header fields indicates that 1374 the sender is going to close the connection after the current 1375 request/response is complete (Section 9.7). 1377 A client that does not support persistent connections MUST send the 1378 "close" connection option in every request message. 1380 A server that does not support persistent connections MUST send the 1381 "close" connection option in every response message that does not 1382 have a 1xx (Informational) status code. 1384 9.2. Establishment 1386 It is beyond the scope of this specification to describe how 1387 connections are established via various transport- or session-layer 1388 protocols. Each connection applies to only one transport link. 1390 9.3. Associating a Response to a Request 1392 HTTP/1.1 does not include a request identifier for associating a 1393 given request message with its corresponding one or more response 1394 messages. Hence, it relies on the order of response arrival to 1395 correspond exactly to the order in which requests are made on the 1396 same connection. More than one response message per request only 1397 occurs when one or more informational responses (1xx, see Section 9.2 1398 of [Semantics]) precede a final response to the same request. 1400 A client that has more than one outstanding request on a connection 1401 MUST maintain a list of outstanding requests in the order sent and 1402 MUST associate each received response message on that connection to 1403 the highest ordered request that has not yet received a final (non- 1404 1xx) response. 1406 If an HTTP/1.1 client receives data on a connection that doesn't have 1407 any outstanding requests, it MUST NOT consider them to be a response 1408 to a not-yet-issued request; it SHOULD close the connection, since 1409 message delimitation is now ambiguous, unless the data consists only 1410 of one or more CRLF (which can be discarded, as per Section 2.2). 1412 9.4. Persistence 1414 HTTP/1.1 defaults to the use of "persistent connections", allowing 1415 multiple requests and responses to be carried over a single 1416 connection. The "close" connection option is used to signal that a 1417 connection will not persist after the current request/response. HTTP 1418 implementations SHOULD support persistent connections. 1420 A recipient determines whether a connection is persistent or not 1421 based on the most recently received message's protocol version and 1422 Connection header field (if any): 1424 o If the "close" connection option is present, the connection will 1425 not persist after the current response; else, 1427 o If the received protocol is HTTP/1.1 (or later), the connection 1428 will persist after the current response; else, 1430 o If the received protocol is HTTP/1.0, the "keep-alive" connection 1431 option is present, either the recipient is not a proxy or the 1432 message is a response, and the recipient wishes to honor the 1433 HTTP/1.0 "keep-alive" mechanism, the connection will persist after 1434 the current response; otherwise, 1436 o The connection will close after the current response. 1438 A client MAY send additional requests on a persistent connection 1439 until it sends or receives a "close" connection option or receives an 1440 HTTP/1.0 response without a "keep-alive" connection option. 1442 In order to remain persistent, all messages on a connection need to 1443 have a self-defined message length (i.e., one not defined by closure 1444 of the connection), as described in Section 6. A server MUST read 1445 the entire request message body or close the connection after sending 1446 its response, since otherwise the remaining data on a persistent 1447 connection would be misinterpreted as the next request. Likewise, a 1448 client MUST read the entire response message body if it intends to 1449 reuse the same connection for a subsequent request. 1451 A proxy server MUST NOT maintain a persistent connection with an 1452 HTTP/1.0 client (see Section 19.7.1 of [RFC2068] for information and 1453 discussion of the problems with the Keep-Alive header field 1454 implemented by many HTTP/1.0 clients). 1456 See Appendix C.1.2 for more information on backwards compatibility 1457 with HTTP/1.0 clients. 1459 9.4.1. Retrying Requests 1461 Connections can be closed at any time, with or without intention. 1462 Implementations ought to anticipate the need to recover from 1463 asynchronous close events. The conditions under which a client can 1464 automatically retry a sequence of outstanding requests are defined in 1465 Section 7.2.2 of [Semantics]. 1467 9.4.2. Pipelining 1469 A client that supports persistent connections MAY "pipeline" its 1470 requests (i.e., send multiple requests without waiting for each 1471 response). A server MAY process a sequence of pipelined requests in 1472 parallel if they all have safe methods (Section 7.2.1 of 1473 [Semantics]), but it MUST send the corresponding responses in the 1474 same order that the requests were received. 1476 A client that pipelines requests SHOULD retry unanswered requests if 1477 the connection closes before it receives all of the corresponding 1478 responses. When retrying pipelined requests after a failed 1479 connection (a connection not explicitly closed by the server in its 1480 last complete response), a client MUST NOT pipeline immediately after 1481 connection establishment, since the first remaining request in the 1482 prior pipeline might have caused an error response that can be lost 1483 again if multiple requests are sent on a prematurely closed 1484 connection (see the TCP reset problem described in Section 9.7). 1486 Idempotent methods (Section 7.2.2 of [Semantics]) are significant to 1487 pipelining because they can be automatically retried after a 1488 connection failure. A user agent SHOULD NOT pipeline requests after 1489 a non-idempotent method, until the final response status code for 1490 that method has been received, unless the user agent has a means to 1491 detect and recover from partial failure conditions involving the 1492 pipelined sequence. 1494 An intermediary that receives pipelined requests MAY pipeline those 1495 requests when forwarding them inbound, since it can rely on the 1496 outbound user agent(s) to determine what requests can be safely 1497 pipelined. If the inbound connection fails before receiving a 1498 response, the pipelining intermediary MAY attempt to retry a sequence 1499 of requests that have yet to receive a response if the requests all 1500 have idempotent methods; otherwise, the pipelining intermediary 1501 SHOULD forward any received responses and then close the 1502 corresponding outbound connection(s) so that the outbound user 1503 agent(s) can recover accordingly. 1505 9.5. Concurrency 1507 A client ought to limit the number of simultaneous open connections 1508 that it maintains to a given server. 1510 Previous revisions of HTTP gave a specific number of connections as a 1511 ceiling, but this was found to be impractical for many applications. 1512 As a result, this specification does not mandate a particular maximum 1513 number of connections but, instead, encourages clients to be 1514 conservative when opening multiple connections. 1516 Multiple connections are typically used to avoid the "head-of-line 1517 blocking" problem, wherein a request that takes significant server- 1518 side processing and/or has a large payload blocks subsequent requests 1519 on the same connection. However, each connection consumes server 1520 resources. Furthermore, using multiple connections can cause 1521 undesirable side effects in congested networks. 1523 Note that a server might reject traffic that it deems abusive or 1524 characteristic of a denial-of-service attack, such as an excessive 1525 number of open connections from a single client. 1527 9.6. Failures and Timeouts 1529 Servers will usually have some timeout value beyond which they will 1530 no longer maintain an inactive connection. Proxy servers might make 1531 this a higher value since it is likely that the client will be making 1532 more connections through the same proxy server. The use of 1533 persistent connections places no requirements on the length (or 1534 existence) of this timeout for either the client or the server. 1536 A client or server that wishes to time out SHOULD issue a graceful 1537 close on the connection. Implementations SHOULD constantly monitor 1538 open connections for a received closure signal and respond to it as 1539 appropriate, since prompt closure of both sides of a connection 1540 enables allocated system resources to be reclaimed. 1542 A client, server, or proxy MAY close the transport connection at any 1543 time. For example, a client might have started to send a new request 1544 at the same time that the server has decided to close the "idle" 1545 connection. From the server's point of view, the connection is being 1546 closed while it was idle, but from the client's point of view, a 1547 request is in progress. 1549 A server SHOULD sustain persistent connections, when possible, and 1550 allow the underlying transport's flow-control mechanisms to resolve 1551 temporary overloads, rather than terminate connections with the 1552 expectation that clients will retry. The latter technique can 1553 exacerbate network congestion. 1555 A client sending a message body SHOULD monitor the network connection 1556 for an error response while it is transmitting the request. If the 1557 client sees a response that indicates the server does not wish to 1558 receive the message body and is closing the connection, the client 1559 SHOULD immediately cease transmitting the body and close its side of 1560 the connection. 1562 9.7. Tear-down 1564 The Connection header field (Section 9.1) provides a "close" 1565 connection option that a sender SHOULD send when it wishes to close 1566 the connection after the current request/response pair. 1568 A client that sends a "close" connection option MUST NOT send further 1569 requests on that connection (after the one containing "close") and 1570 MUST close the connection after reading the final response message 1571 corresponding to this request. 1573 A server that receives a "close" connection option MUST initiate a 1574 close of the connection (see below) after it sends the final response 1575 to the request that contained "close". The server SHOULD send a 1576 "close" connection option in its final response on that connection. 1577 The server MUST NOT process any further requests received on that 1578 connection. 1580 A server that sends a "close" connection option MUST initiate a close 1581 of the connection (see below) after it sends the response containing 1582 "close". The server MUST NOT process any further requests received 1583 on that connection. 1585 A client that receives a "close" connection option MUST cease sending 1586 requests on that connection and close the connection after reading 1587 the response message containing the "close"; if additional pipelined 1588 requests had been sent on the connection, the client SHOULD NOT 1589 assume that they will be processed by the server. 1591 If a server performs an immediate close of a TCP connection, there is 1592 a significant risk that the client will not be able to read the last 1593 HTTP response. If the server receives additional data from the 1594 client on a fully closed connection, such as another request that was 1595 sent by the client before receiving the server's response, the 1596 server's TCP stack will send a reset packet to the client; 1597 unfortunately, the reset packet might erase the client's 1598 unacknowledged input buffers before they can be read and interpreted 1599 by the client's HTTP parser. 1601 To avoid the TCP reset problem, servers typically close a connection 1602 in stages. First, the server performs a half-close by closing only 1603 the write side of the read/write connection. The server then 1604 continues to read from the connection until it receives a 1605 corresponding close by the client, or until the server is reasonably 1606 certain that its own TCP stack has received the client's 1607 acknowledgement of the packet(s) containing the server's last 1608 response. Finally, the server fully closes the connection. 1610 It is unknown whether the reset problem is exclusive to TCP or might 1611 also be found in other transport connection protocols. 1613 9.8. TLS Connection Closure 1615 TLS provides a facility for secure connection closure. When a valid 1616 closure alert is received, an implementation can be assured that no 1617 further data will be received on that connection. TLS 1618 implementations MUST initiate an exchange of closure alerts before 1619 closing a connection. A TLS implementation MAY, after sending a 1620 closure alert, close the connection without waiting for the peer to 1621 send its closure alert, generating an "incomplete close". Note that 1622 an implementation which does this MAY choose to reuse the session. 1623 This SHOULD only be done when the application knows (typically 1624 through detecting HTTP message boundaries) that it has received all 1625 the message data that it cares about. 1627 As specified in [RFC8446], any implementation which receives a 1628 connection close without first receiving a valid closure alert (a 1629 "premature close") MUST NOT reuse that session. Note that a 1630 premature close does not call into question the security of the data 1631 already received, but simply indicates that subsequent data might 1632 have been truncated. Because TLS is oblivious to HTTP request/ 1633 response boundaries, it is necessary to examine the HTTP data itself 1634 (specifically the Content-Length header) to determine whether the 1635 truncation occurred inside a message or between messages. 1637 When encountering a premature close, a client SHOULD treat as 1638 completed all requests for which it has received as much data as 1639 specified in the Content-Length header. 1641 A client detecting an incomplete close SHOULD recover gracefully. It 1642 MAY resume a TLS session closed in this fashion. 1644 Clients MUST send a closure alert before closing the connection. 1645 Clients which are unprepared to receive any more data MAY choose not 1646 to wait for the server's closure alert and simply close the 1647 connection, thus generating an incomplete close on the server side. 1649 Servers SHOULD be prepared to receive an incomplete close from the 1650 client, since the client can often determine when the end of server 1651 data is. Servers SHOULD be willing to resume TLS sessions closed in 1652 this fashion. 1654 Servers MUST attempt to initiate an exchange of closure alerts with 1655 the client before closing the connection. Servers MAY close the 1656 connection after sending the closure alert, thus generating an 1657 incomplete close on the client side. 1659 9.9. Upgrade 1661 The "Upgrade" header field is intended to provide a simple mechanism 1662 for transitioning from HTTP/1.1 to some other protocol on the same 1663 connection. 1665 A client MAY send a list of protocol names in the Upgrade header 1666 field of a request to invite the server to switch to one or more of 1667 the named protocols, in order of descending preference, before 1668 sending the final response. A server MAY ignore a received Upgrade 1669 header field if it wishes to continue using the current protocol on 1670 that connection. Upgrade cannot be used to insist on a protocol 1671 change. 1673 Upgrade = 1#protocol 1675 protocol = protocol-name ["/" protocol-version] 1676 protocol-name = token 1677 protocol-version = token 1679 Although protocol names are registered with a preferred case, 1680 recipients SHOULD use case-insensitive comparison when matching each 1681 protocol-name to supported protocols. 1683 A server that sends a 101 (Switching Protocols) response MUST send an 1684 Upgrade header field to indicate the new protocol(s) to which the 1685 connection is being switched; if multiple protocol layers are being 1686 switched, the sender MUST list the protocols in layer-ascending 1687 order. A server MUST NOT switch to a protocol that was not indicated 1688 by the client in the corresponding request's Upgrade header field. A 1689 server MAY choose to ignore the order of preference indicated by the 1690 client and select the new protocol(s) based on other factors, such as 1691 the nature of the request or the current load on the server. 1693 A server that sends a 426 (Upgrade Required) response MUST send an 1694 Upgrade header field to indicate the acceptable protocols, in order 1695 of descending preference. 1697 A server MAY send an Upgrade header field in any other response to 1698 advertise that it implements support for upgrading to the listed 1699 protocols, in order of descending preference, when appropriate for a 1700 future request. 1702 The following is a hypothetical example sent by a client: 1704 GET /hello HTTP/1.1 1705 Host: www.example.com 1706 Connection: upgrade 1707 Upgrade: websocket, IRC/6.9, RTA/x11 1709 The capabilities and nature of the application-level communication 1710 after the protocol change is entirely dependent upon the new 1711 protocol(s) chosen. However, immediately after sending the 101 1712 (Switching Protocols) response, the server is expected to continue 1713 responding to the original request as if it had received its 1714 equivalent within the new protocol (i.e., the server still has an 1715 outstanding request to satisfy after the protocol has been changed, 1716 and is expected to do so without requiring the request to be 1717 repeated). 1719 For example, if the Upgrade header field is received in a GET request 1720 and the server decides to switch protocols, it first responds with a 1721 101 (Switching Protocols) message in HTTP/1.1 and then immediately 1722 follows that with the new protocol's equivalent of a response to a 1723 GET on the target resource. This allows a connection to be upgraded 1724 to protocols with the same semantics as HTTP without the latency cost 1725 of an additional round trip. A server MUST NOT switch protocols 1726 unless the received message semantics can be honored by the new 1727 protocol; an OPTIONS request can be honored by any protocol. 1729 The following is an example response to the above hypothetical 1730 request: 1732 HTTP/1.1 101 Switching Protocols 1733 Connection: upgrade 1734 Upgrade: websocket 1736 [... data stream switches to websocket with an appropriate response 1737 (as defined by new protocol) to the "GET /hello" request ...] 1739 When Upgrade is sent, the sender MUST also send a Connection header 1740 field (Section 9.1) that contains an "upgrade" connection option, in 1741 order to prevent Upgrade from being accidentally forwarded by 1742 intermediaries that might not implement the listed protocols. A 1743 server MUST ignore an Upgrade header field that is received in an 1744 HTTP/1.0 request. 1746 A client cannot begin using an upgraded protocol on the connection 1747 until it has completely sent the request message (i.e., the client 1748 can't change the protocol it is sending in the middle of a message). 1749 If a server receives both an Upgrade and an Expect header field with 1750 the "100-continue" expectation (Section 8.1.1 of [Semantics]), the 1751 server MUST send a 100 (Continue) response before sending a 101 1752 (Switching Protocols) response. 1754 The Upgrade header field only applies to switching protocols on top 1755 of the existing connection; it cannot be used to switch the 1756 underlying connection (transport) protocol, nor to switch the 1757 existing communication to a different connection. For those 1758 purposes, it is more appropriate to use a 3xx (Redirection) response 1759 (Section 9.4 of [Semantics]). 1761 9.9.1. Upgrade Protocol Names 1763 This specification only defines the protocol name "HTTP" for use by 1764 the family of Hypertext Transfer Protocols, as defined by the HTTP 1765 version rules of Section 3.5 of [Semantics] and future updates to 1766 this specification. Additional protocol names ought to be registered 1767 using the registration procedure defined in Section 9.9.2. 1769 +------+-------------------+--------------------+-------------------+ 1770 | Name | Description | Expected Version | Reference | 1771 | | | Tokens | | 1772 +------+-------------------+--------------------+-------------------+ 1773 | HTTP | Hypertext | any DIGIT.DIGIT | Section 3.5 of | 1774 | | Transfer Protocol | (e.g, "2.0") | [Semantics] | 1775 +------+-------------------+--------------------+-------------------+ 1777 9.9.2. Upgrade Token Registry 1779 The "Hypertext Transfer Protocol (HTTP) Upgrade Token Registry" 1780 defines the namespace for protocol-name tokens used to identify 1781 protocols in the Upgrade header field. The registry is maintained at 1782 . 1784 Each registered protocol name is associated with contact information 1785 and an optional set of specifications that details how the connection 1786 will be processed after it has been upgraded. 1788 Registrations happen on a "First Come First Served" basis (see 1789 Section 4.4 of [RFC8126]) and are subject to the following rules: 1791 1. A protocol-name token, once registered, stays registered forever. 1793 2. A protocol-name token is case-insensitive and registered with the 1794 preferred case to be generated by senders. 1796 3. The registration MUST name a responsible party for the 1797 registration. 1799 4. The registration MUST name a point of contact. 1801 5. The registration MAY name a set of specifications associated with 1802 that token. Such specifications need not be publicly available. 1804 6. The registration SHOULD name a set of expected "protocol-version" 1805 tokens associated with that token at the time of registration. 1807 7. The responsible party MAY change the registration at any time. 1808 The IANA will keep a record of all such changes, and make them 1809 available upon request. 1811 8. The IESG MAY reassign responsibility for a protocol token. This 1812 will normally only be used in the case when a responsible party 1813 cannot be contacted. 1815 10. Enclosing Messages as Data 1817 10.1. Media Type message/http 1819 The message/http media type can be used to enclose a single HTTP 1820 request or response message, provided that it obeys the MIME 1821 restrictions for all "message" types regarding line length and 1822 encodings. 1824 Type name: message 1826 Subtype name: http 1828 Required parameters: N/A 1830 Optional parameters: version, msgtype 1832 version: The HTTP-version number of the enclosed message (e.g., 1833 "1.1"). If not present, the version can be determined from the 1834 first line of the body. 1836 msgtype: The message type -- "request" or "response". If not 1837 present, the type can be determined from the first line of the 1838 body. 1840 Encoding considerations: only "7bit", "8bit", or "binary" are 1841 permitted 1843 Security considerations: see Section 11 1845 Interoperability considerations: N/A 1847 Published specification: This specification (see Section 10.1). 1849 Applications that use this media type: N/A 1851 Fragment identifier considerations: N/A 1853 Additional information: 1855 Magic number(s): N/A 1857 Deprecated alias names for this type: N/A 1859 File extension(s): N/A 1861 Macintosh file type code(s): N/A 1863 Person and email address to contact for further information: 1864 See Authors' Addresses section. 1866 Intended usage: COMMON 1868 Restrictions on usage: N/A 1870 Author: See Authors' Addresses section. 1872 Change controller: IESG 1874 10.2. Media Type application/http 1876 The application/http media type can be used to enclose a pipeline of 1877 one or more HTTP request or response messages (not intermixed). 1879 Type name: application 1881 Subtype name: http 1883 Required parameters: N/A 1884 Optional parameters: version, msgtype 1886 version: The HTTP-version number of the enclosed messages (e.g., 1887 "1.1"). If not present, the version can be determined from the 1888 first line of the body. 1890 msgtype: The message type -- "request" or "response". If not 1891 present, the type can be determined from the first line of the 1892 body. 1894 Encoding considerations: HTTP messages enclosed by this type are in 1895 "binary" format; use of an appropriate Content-Transfer-Encoding 1896 is required when transmitted via email. 1898 Security considerations: see Section 11 1900 Interoperability considerations: N/A 1902 Published specification: This specification (see Section 10.2). 1904 Applications that use this media type: N/A 1906 Fragment identifier considerations: N/A 1908 Additional information: 1910 Deprecated alias names for this type: N/A 1912 Magic number(s): N/A 1914 File extension(s): N/A 1916 Macintosh file type code(s): N/A 1918 Person and email address to contact for further information: 1919 See Authors' Addresses section. 1921 Intended usage: COMMON 1923 Restrictions on usage: N/A 1925 Author: See Authors' Addresses section. 1927 Change controller: IESG 1929 11. Security Considerations 1931 This section is meant to inform developers, information providers, 1932 and users of known security considerations relevant to HTTP message 1933 syntax, parsing, and routing. Security considerations about HTTP 1934 semantics and payloads are addressed in [Semantics]. 1936 11.1. Response Splitting 1938 Response splitting (a.k.a, CRLF injection) is a common technique, 1939 used in various attacks on Web usage, that exploits the line-based 1940 nature of HTTP message framing and the ordered association of 1941 requests to responses on persistent connections [Klein]. This 1942 technique can be particularly damaging when the requests pass through 1943 a shared cache. 1945 Response splitting exploits a vulnerability in servers (usually 1946 within an application server) where an attacker can send encoded data 1947 within some parameter of the request that is later decoded and echoed 1948 within any of the response header fields of the response. If the 1949 decoded data is crafted to look like the response has ended and a 1950 subsequent response has begun, the response has been split and the 1951 content within the apparent second response is controlled by the 1952 attacker. The attacker can then make any other request on the same 1953 persistent connection and trick the recipients (including 1954 intermediaries) into believing that the second half of the split is 1955 an authoritative answer to the second request. 1957 For example, a parameter within the request-target might be read by 1958 an application server and reused within a redirect, resulting in the 1959 same parameter being echoed in the Location header field of the 1960 response. If the parameter is decoded by the application and not 1961 properly encoded when placed in the response field, the attacker can 1962 send encoded CRLF octets and other content that will make the 1963 application's single response look like two or more responses. 1965 A common defense against response splitting is to filter requests for 1966 data that looks like encoded CR and LF (e.g., "%0D" and "%0A"). 1967 However, that assumes the application server is only performing URI 1968 decoding, rather than more obscure data transformations like charset 1969 transcoding, XML entity translation, base64 decoding, sprintf 1970 reformatting, etc. A more effective mitigation is to prevent 1971 anything other than the server's core protocol libraries from sending 1972 a CR or LF within the header section, which means restricting the 1973 output of header fields to APIs that filter for bad octets and not 1974 allowing application servers to write directly to the protocol 1975 stream. 1977 11.2. Request Smuggling 1979 Request smuggling ([Linhart]) is a technique that exploits 1980 differences in protocol parsing among various recipients to hide 1981 additional requests (which might otherwise be blocked or disabled by 1982 policy) within an apparently harmless request. Like response 1983 splitting, request smuggling can lead to a variety of attacks on HTTP 1984 usage. 1986 This specification has introduced new requirements on request 1987 parsing, particularly with regard to message framing in Section 6.3, 1988 to reduce the effectiveness of request smuggling. 1990 11.3. Message Integrity 1992 HTTP does not define a specific mechanism for ensuring message 1993 integrity, instead relying on the error-detection ability of 1994 underlying transport protocols and the use of length or chunk- 1995 delimited framing to detect completeness. Additional integrity 1996 mechanisms, such as hash functions or digital signatures applied to 1997 the content, can be selectively added to messages via extensible 1998 metadata fields. Historically, the lack of a single integrity 1999 mechanism has been justified by the informal nature of most HTTP 2000 communication. However, the prevalence of HTTP as an information 2001 access mechanism has resulted in its increasing use within 2002 environments where verification of message integrity is crucial. 2004 User agents are encouraged to implement configurable means for 2005 detecting and reporting failures of message integrity such that those 2006 means can be enabled within environments for which integrity is 2007 necessary. For example, a browser being used to view medical history 2008 or drug interaction information needs to indicate to the user when 2009 such information is detected by the protocol to be incomplete, 2010 expired, or corrupted during transfer. Such mechanisms might be 2011 selectively enabled via user agent extensions or the presence of 2012 message integrity metadata in a response. At a minimum, user agents 2013 ought to provide some indication that allows a user to distinguish 2014 between a complete and incomplete response message (Section 8) when 2015 such verification is desired. 2017 11.4. Message Confidentiality 2019 HTTP relies on underlying transport protocols to provide message 2020 confidentiality when that is desired. HTTP has been specifically 2021 designed to be independent of the transport protocol, such that it 2022 can be used over many different forms of encrypted connection, with 2023 the selection of such transports being identified by the choice of 2024 URI scheme or within user agent configuration. 2026 The "https" scheme can be used to identify resources that require a 2027 confidential connection, as described in Section 2.5.2 of 2028 [Semantics]. 2030 12. IANA Considerations 2032 The change controller for the following registrations is: "IETF 2033 (iesg@ietf.org) - Internet Engineering Task Force". 2035 12.1. Field Name Registration 2037 Please update the "Hypertext Transfer Protocol (HTTP) Field Name 2038 Registry" at with the 2039 field names listed in the two tables of Section 5. 2041 12.2. Media Type Registration 2043 Please update the "Media Types" registry at 2044 with the registration 2045 information in Section 10.1 and Section 10.2 for the media types 2046 "message/http" and "application/http", respectively. 2048 12.3. Transfer Coding Registration 2050 Please update the "HTTP Transfer Coding Registry" at 2051 with the 2052 registration procedure of Section 7.3 and the content coding names 2053 summarized in the table of Section 7. 2055 12.4. Upgrade Token Registration 2057 Please update the "Hypertext Transfer Protocol (HTTP) Upgrade Token 2058 Registry" at 2059 with the registration procedure of Section 9.9.2 and the upgrade 2060 token names summarized in the table of Section 9.9.1. 2062 12.5. ALPN Protocol ID Registration 2064 Please update the "TLS Application-Layer Protocol Negotiation (ALPN) 2065 Protocol IDs" registry at with the 2067 registration below: 2069 +----------+--------------------------------------+-----------------+ 2070 | Protocol | Identification Sequence | Reference | 2071 +----------+--------------------------------------+-----------------+ 2072 | HTTP/1.1 | 0x68 0x74 0x74 0x70 0x2f 0x31 0x2e | (this | 2073 | | 0x31 ("http/1.1") | specification) | 2074 +----------+--------------------------------------+-----------------+ 2076 13. References 2078 13.1. Normative References 2080 [Caching] Fielding, R., Ed., Nottingham, M., Ed., and J. Reschke, 2081 Ed., "HTTP Caching", draft-ietf-httpbis-cache-08 (work in 2082 progress), May 2020. 2084 [RFC1950] Deutsch, L. and J-L. Gailly, "ZLIB Compressed Data Format 2085 Specification version 3.3", RFC 1950, 2086 DOI 10.17487/RFC1950, May 1996, 2087 . 2089 [RFC1951] Deutsch, P., "DEFLATE Compressed Data Format Specification 2090 version 1.3", RFC 1951, DOI 10.17487/RFC1951, May 1996, 2091 . 2093 [RFC1952] Deutsch, P., Gailly, J-L., Adler, M., Deutsch, L., and G. 2094 Randers-Pehrson, "GZIP file format specification version 2095 4.3", RFC 1952, DOI 10.17487/RFC1952, May 1996, 2096 . 2098 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 2099 Requirement Levels", BCP 14, RFC 2119, 2100 DOI 10.17487/RFC2119, March 1997, 2101 . 2103 [RFC3986] Berners-Lee, T., Fielding, R., and L. Masinter, "Uniform 2104 Resource Identifier (URI): Generic Syntax", STD 66, 2105 RFC 3986, DOI 10.17487/RFC3986, January 2005, 2106 . 2108 [RFC5234] Crocker, D., Ed. and P. Overell, "Augmented BNF for Syntax 2109 Specifications: ABNF", STD 68, RFC 5234, 2110 DOI 10.17487/RFC5234, January 2008, 2111 . 2113 [RFC7405] Kyzivat, P., "Case-Sensitive String Support in ABNF", 2114 RFC 7405, DOI 10.17487/RFC7405, December 2014, 2115 . 2117 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2118 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 2119 May 2017, . 2121 [RFC8446] Rescorla, E., "The Transport Layer Security (TLS) Protocol 2122 Version 1.3", RFC 8446, DOI 10.17487/RFC8446, August 2018, 2123 . 2125 [Semantics] 2126 Fielding, R., Ed., Nottingham, M., Ed., and J. Reschke, 2127 Ed., "HTTP Semantics", draft-ietf-httpbis-semantics-08 2128 (work in progress), May 2020. 2130 [USASCII] American National Standards Institute, "Coded Character 2131 Set -- 7-bit American Standard Code for Information 2132 Interchange", ANSI X3.4, 1986. 2134 [Welch] Welch, T., "A Technique for High-Performance Data 2135 Compression", IEEE Computer 17(6), June 1984. 2137 13.2. Informative References 2139 [Err4667] RFC Errata, Erratum ID 4667, RFC 7230, 2140 . 2142 [Klein] Klein, A., "Divide and Conquer - HTTP Response Splitting, 2143 Web Cache Poisoning Attacks, and Related Topics", March 2144 2004, . 2147 [Linhart] Linhart, C., Klein, A., Heled, R., and S. Orrin, "HTTP 2148 Request Smuggling", June 2005, 2149 . 2151 [RFC1945] Berners-Lee, T., Fielding, R., and H. Nielsen, "Hypertext 2152 Transfer Protocol -- HTTP/1.0", RFC 1945, 2153 DOI 10.17487/RFC1945, May 1996, 2154 . 2156 [RFC2045] Freed, N. and N. Borenstein, "Multipurpose Internet Mail 2157 Extensions (MIME) Part One: Format of Internet Message 2158 Bodies", RFC 2045, DOI 10.17487/RFC2045, November 1996, 2159 . 2161 [RFC2046] Freed, N. and N. Borenstein, "Multipurpose Internet Mail 2162 Extensions (MIME) Part Two: Media Types", RFC 2046, 2163 DOI 10.17487/RFC2046, November 1996, 2164 . 2166 [RFC2049] Freed, N. and N. Borenstein, "Multipurpose Internet Mail 2167 Extensions (MIME) Part Five: Conformance Criteria and 2168 Examples", RFC 2049, DOI 10.17487/RFC2049, November 1996, 2169 . 2171 [RFC2068] Fielding, R., Gettys, J., Mogul, J., Nielsen, H., and T. 2172 Berners-Lee, "Hypertext Transfer Protocol -- HTTP/1.1", 2173 RFC 2068, DOI 10.17487/RFC2068, January 1997, 2174 . 2176 [RFC2557] Palme, F., Hopmann, A., Shelness, N., and E. Stefferud, 2177 "MIME Encapsulation of Aggregate Documents, such as HTML 2178 (MHTML)", RFC 2557, DOI 10.17487/RFC2557, March 1999, 2179 . 2181 [RFC5322] Resnick, P., "Internet Message Format", RFC 5322, 2182 DOI 10.17487/RFC5322, October 2008, 2183 . 2185 [RFC7230] Fielding, R., Ed. and J. Reschke, Ed., "Hypertext Transfer 2186 Protocol (HTTP/1.1): Message Syntax and Routing", 2187 RFC 7230, DOI 10.17487/RFC7230, June 2014, 2188 . 2190 [RFC7231] Fielding, R., Ed. and J. Reschke, Ed., "Hypertext Transfer 2191 Protocol (HTTP/1.1): Semantics and Content", RFC 7231, 2192 DOI 10.17487/RFC7231, June 2014, 2193 . 2195 [RFC8126] Cotton, M., Leiba, B., and T. Narten, "Guidelines for 2196 Writing an IANA Considerations Section in RFCs", BCP 26, 2197 RFC 8126, DOI 10.17487/RFC8126, June 2017, 2198 . 2200 Appendix A. Collected ABNF 2202 In the collected ABNF below, list rules are expanded as per 2203 Section 4.5 of [Semantics]. 2205 BWS = 2207 Connection = [ connection-option ] *( OWS "," OWS [ connection-option 2208 ] ) 2210 HTTP-message = start-line CRLF *( field-line CRLF ) CRLF [ 2211 message-body ] 2212 HTTP-name = %x48.54.54.50 ; HTTP 2213 HTTP-version = HTTP-name "/" DIGIT "." DIGIT 2215 OWS = 2217 RWS = 2219 TE = [ t-codings ] *( OWS "," OWS [ t-codings ] ) 2220 Transfer-Encoding = [ transfer-coding ] *( OWS "," OWS [ 2221 transfer-coding ] ) 2223 Upgrade = [ protocol ] *( OWS "," OWS [ protocol ] ) 2225 absolute-URI = 2226 absolute-form = absolute-URI 2227 absolute-path = 2228 asterisk-form = "*" 2229 authority = 2230 authority-form = authority 2232 chunk = chunk-size [ chunk-ext ] CRLF chunk-data CRLF 2233 chunk-data = 1*OCTET 2234 chunk-ext = *( BWS ";" BWS chunk-ext-name [ BWS "=" BWS chunk-ext-val 2235 ] ) 2236 chunk-ext-name = token 2237 chunk-ext-val = token / quoted-string 2238 chunk-size = 1*HEXDIG 2239 chunked-body = *chunk last-chunk trailer-section CRLF 2240 comment = 2241 connection-option = token 2243 field-line = field-name ":" OWS field-value OWS 2244 field-name = 2245 field-value = 2247 last-chunk = 1*"0" [ chunk-ext ] CRLF 2248 message-body = *OCTET 2249 method = token 2251 obs-fold = OWS CRLF RWS 2252 obs-text = 2253 origin-form = absolute-path [ "?" query ] 2255 port = 2256 protocol = protocol-name [ "/" protocol-version ] 2257 protocol-name = token 2258 protocol-version = token 2260 query = 2261 quoted-string = 2263 rank = ( "0" [ "." *3DIGIT ] ) / ( "1" [ "." *3"0" ] ) 2264 reason-phrase = 1*( HTAB / SP / VCHAR / obs-text ) 2265 request-line = method SP request-target SP HTTP-version 2266 request-target = origin-form / absolute-form / authority-form / 2267 asterisk-form 2269 start-line = request-line / status-line 2270 status-code = 3DIGIT 2271 status-line = HTTP-version SP status-code SP [ reason-phrase ] 2273 t-codings = "trailers" / ( transfer-coding [ t-ranking ] ) 2274 t-ranking = OWS ";" OWS "q=" rank 2275 token = 2276 trailer-section = *( field-line CRLF ) 2277 transfer-coding = token *( OWS ";" OWS transfer-parameter ) 2278 transfer-parameter = token BWS "=" BWS ( token / quoted-string ) 2280 uri-host = 2282 Appendix B. Differences between HTTP and MIME 2284 HTTP/1.1 uses many of the constructs defined for the Internet Message 2285 Format [RFC5322] and the Multipurpose Internet Mail Extensions (MIME) 2286 [RFC2045] to allow a message body to be transmitted in an open 2287 variety of representations and with extensible fields. However, RFC 2288 2045 is focused only on email; applications of HTTP have many 2289 characteristics that differ from email; hence, HTTP has features that 2290 differ from MIME. These differences were carefully chosen to 2291 optimize performance over binary connections, to allow greater 2292 freedom in the use of new media types, to make date comparisons 2293 easier, and to acknowledge the practice of some early HTTP servers 2294 and clients. 2296 This appendix describes specific areas where HTTP differs from MIME. 2297 Proxies and gateways to and from strict MIME environments need to be 2298 aware of these differences and provide the appropriate conversions 2299 where necessary. 2301 B.1. MIME-Version 2303 HTTP is not a MIME-compliant protocol. However, messages can include 2304 a single MIME-Version header field to indicate what version of the 2305 MIME protocol was used to construct the message. Use of the MIME- 2306 Version header field indicates that the message is in full 2307 conformance with the MIME protocol (as defined in [RFC2045]). 2308 Senders are responsible for ensuring full conformance (where 2309 possible) when exporting HTTP messages to strict MIME environments. 2311 B.2. Conversion to Canonical Form 2313 MIME requires that an Internet mail body part be converted to 2314 canonical form prior to being transferred, as described in Section 4 2315 of [RFC2049]. Section 6.1.1.2 of [Semantics] describes the forms 2316 allowed for subtypes of the "text" media type when transmitted over 2317 HTTP. [RFC2046] requires that content with a type of "text" 2318 represent line breaks as CRLF and forbids the use of CR or LF outside 2319 of line break sequences. HTTP allows CRLF, bare CR, and bare LF to 2320 indicate a line break within text content. 2322 A proxy or gateway from HTTP to a strict MIME environment ought to 2323 translate all line breaks within text media types to the RFC 2049 2324 canonical form of CRLF. Note, however, this might be complicated by 2325 the presence of a Content-Encoding and by the fact that HTTP allows 2326 the use of some charsets that do not use octets 13 and 10 to 2327 represent CR and LF, respectively. 2329 Conversion will break any cryptographic checksums applied to the 2330 original content unless the original content is already in canonical 2331 form. Therefore, the canonical form is recommended for any content 2332 that uses such checksums in HTTP. 2334 B.3. Conversion of Date Formats 2336 HTTP/1.1 uses a restricted set of date formats (Section 10.1.1.1 of 2337 [Semantics]) to simplify the process of date comparison. Proxies and 2338 gateways from other protocols ought to ensure that any Date header 2339 field present in a message conforms to one of the HTTP/1.1 formats 2340 and rewrite the date if necessary. 2342 B.4. Conversion of Content-Encoding 2344 MIME does not include any concept equivalent to HTTP/1.1's Content- 2345 Encoding header field. Since this acts as a modifier on the media 2346 type, proxies and gateways from HTTP to MIME-compliant protocols 2347 ought to either change the value of the Content-Type header field or 2348 decode the representation before forwarding the message. (Some 2349 experimental applications of Content-Type for Internet mail have used 2350 a media-type parameter of ";conversions=" to perform 2351 a function equivalent to Content-Encoding. However, this parameter 2352 is not part of the MIME standards). 2354 B.5. Conversion of Content-Transfer-Encoding 2356 HTTP does not use the Content-Transfer-Encoding field of MIME. 2357 Proxies and gateways from MIME-compliant protocols to HTTP need to 2358 remove any Content-Transfer-Encoding prior to delivering the response 2359 message to an HTTP client. 2361 Proxies and gateways from HTTP to MIME-compliant protocols are 2362 responsible for ensuring that the message is in the correct format 2363 and encoding for safe transport on that protocol, where "safe 2364 transport" is defined by the limitations of the protocol being used. 2365 Such a proxy or gateway ought to transform and label the data with an 2366 appropriate Content-Transfer-Encoding if doing so will improve the 2367 likelihood of safe transport over the destination protocol. 2369 B.6. MHTML and Line Length Limitations 2371 HTTP implementations that share code with MHTML [RFC2557] 2372 implementations need to be aware of MIME line length limitations. 2373 Since HTTP does not have this limitation, HTTP does not fold long 2374 lines. MHTML messages being transported by HTTP follow all 2375 conventions of MHTML, including line length limitations and folding, 2376 canonicalization, etc., since HTTP transfers message-bodies as 2377 payload and, aside from the "multipart/byteranges" type 2378 (Section 6.3.5 of [Semantics]), does not interpret the content or any 2379 MIME header lines that might be contained therein. 2381 Appendix C. HTTP Version History 2383 HTTP has been in use since 1990. The first version, later referred 2384 to as HTTP/0.9, was a simple protocol for hypertext data transfer 2385 across the Internet, using only a single request method (GET) and no 2386 metadata. HTTP/1.0, as defined by [RFC1945], added a range of 2387 request methods and MIME-like messaging, allowing for metadata to be 2388 transferred and modifiers placed on the request/response semantics. 2389 However, HTTP/1.0 did not sufficiently take into consideration the 2390 effects of hierarchical proxies, caching, the need for persistent 2391 connections, or name-based virtual hosts. The proliferation of 2392 incompletely implemented applications calling themselves "HTTP/1.0" 2393 further necessitated a protocol version change in order for two 2394 communicating applications to determine each other's true 2395 capabilities. 2397 HTTP/1.1 remains compatible with HTTP/1.0 by including more stringent 2398 requirements that enable reliable implementations, adding only those 2399 features that can either be safely ignored by an HTTP/1.0 recipient 2400 or only be sent when communicating with a party advertising 2401 conformance with HTTP/1.1. 2403 HTTP/1.1 has been designed to make supporting previous versions easy. 2404 A general-purpose HTTP/1.1 server ought to be able to understand any 2405 valid request in the format of HTTP/1.0, responding appropriately 2406 with an HTTP/1.1 message that only uses features understood (or 2407 safely ignored) by HTTP/1.0 clients. Likewise, an HTTP/1.1 client 2408 can be expected to understand any valid HTTP/1.0 response. 2410 Since HTTP/0.9 did not support header fields in a request, there is 2411 no mechanism for it to support name-based virtual hosts (selection of 2412 resource by inspection of the Host header field). Any server that 2413 implements name-based virtual hosts ought to disable support for 2414 HTTP/0.9. Most requests that appear to be HTTP/0.9 are, in fact, 2415 badly constructed HTTP/1.x requests caused by a client failing to 2416 properly encode the request-target. 2418 C.1. Changes from HTTP/1.0 2420 This section summarizes major differences between versions HTTP/1.0 2421 and HTTP/1.1. 2423 C.1.1. Multihomed Web Servers 2425 The requirements that clients and servers support the Host header 2426 field (Section 5.6 of [Semantics]), report an error if it is missing 2427 from an HTTP/1.1 request, and accept absolute URIs (Section 3.2) are 2428 among the most important changes defined by HTTP/1.1. 2430 Older HTTP/1.0 clients assumed a one-to-one relationship of IP 2431 addresses and servers; there was no other established mechanism for 2432 distinguishing the intended server of a request than the IP address 2433 to which that request was directed. The Host header field was 2434 introduced during the development of HTTP/1.1 and, though it was 2435 quickly implemented by most HTTP/1.0 browsers, additional 2436 requirements were placed on all HTTP/1.1 requests in order to ensure 2437 complete adoption. At the time of this writing, most HTTP-based 2438 services are dependent upon the Host header field for targeting 2439 requests. 2441 C.1.2. Keep-Alive Connections 2443 In HTTP/1.0, each connection is established by the client prior to 2444 the request and closed by the server after sending the response. 2445 However, some implementations implement the explicitly negotiated 2446 ("Keep-Alive") version of persistent connections described in 2447 Section 19.7.1 of [RFC2068]. 2449 Some clients and servers might wish to be compatible with these 2450 previous approaches to persistent connections, by explicitly 2451 negotiating for them with a "Connection: keep-alive" request header 2452 field. However, some experimental implementations of HTTP/1.0 2453 persistent connections are faulty; for example, if an HTTP/1.0 proxy 2454 server doesn't understand Connection, it will erroneously forward 2455 that header field to the next inbound server, which would result in a 2456 hung connection. 2458 One attempted solution was the introduction of a Proxy-Connection 2459 header field, targeted specifically at proxies. In practice, this 2460 was also unworkable, because proxies are often deployed in multiple 2461 layers, bringing about the same problem discussed above. 2463 As a result, clients are encouraged not to send the Proxy-Connection 2464 header field in any requests. 2466 Clients are also encouraged to consider the use of Connection: keep- 2467 alive in requests carefully; while they can enable persistent 2468 connections with HTTP/1.0 servers, clients using them will need to 2469 monitor the connection for "hung" requests (which indicate that the 2470 client ought stop sending the header field), and this mechanism ought 2471 not be used by clients at all when a proxy is being used. 2473 C.1.3. Introduction of Transfer-Encoding 2475 HTTP/1.1 introduces the Transfer-Encoding header field (Section 6.1). 2476 Transfer codings need to be decoded prior to forwarding an HTTP 2477 message over a MIME-compliant protocol. 2479 C.2. Changes from RFC 7230 2481 Most of the sections introducing HTTP's design goals, history, 2482 architecture, conformance criteria, protocol versioning, URIs, 2483 message routing, and header fields have been moved to [Semantics]. 2484 This document has been reduced to just the messaging syntax and 2485 connection management requirements specific to HTTP/1.1. 2487 In the ABNF for chunked extensions, re-introduced (bad) whitespace 2488 around ";" and "=". Whitespace was removed in [RFC7230], but that 2489 change was found to break existing implementations (see [Err4667]). 2490 (Section 7.1.1) 2492 Trailer field semantics now transcend the specifics of chunked 2493 encoding. The decoding algorithm for chunked (Section 7.1.3) has 2494 been updated to encourage storage/forwarding of trailer fields 2495 separately from the header section, to only allow merging into the 2496 header section if the recipient knows the corresponding field 2497 definition permits and defines how to merge, and otherwise to discard 2498 the trailer fields instead of merging. The trailer part is now 2499 called the trailer section to be more consistent with the header 2500 section and more distinct from a body part. (Section 7.1.2) 2502 Disallowed transfer coding parameters called "q" in order to avoid 2503 conflicts with the use of ranks in the TE header field. 2504 (Section 7.3) 2506 Appendix D. Change Log 2508 This section is to be removed before publishing as an RFC. 2510 D.1. Between RFC7230 and draft 00 2512 The changes were purely editorial: 2514 o Change boilerplate and abstract to indicate the "draft" status, 2515 and update references to ancestor specifications. 2517 o Adjust historical notes. 2519 o Update links to sibling specifications. 2521 o Replace sections listing changes from RFC 2616 by new empty 2522 sections referring to RFC 723x. 2524 o Remove acknowledgements specific to RFC 723x. 2526 o Move "Acknowledgements" to the very end and make them unnumbered. 2528 D.2. Since draft-ietf-httpbis-messaging-00 2530 The changes in this draft are editorial, with respect to HTTP as a 2531 whole, to move all core HTTP semantics into [Semantics]: 2533 o Moved introduction, architecture, conformance, and ABNF extensions 2534 from RFC 7230 (Messaging) to semantics [Semantics]. 2536 o Moved discussion of MIME differences from RFC 7231 (Semantics) to 2537 Appendix B since they mostly cover transforming 1.1 messages. 2539 o Moved all extensibility tips, registration procedures, and 2540 registry tables from the IANA considerations to normative 2541 sections, reducing the IANA considerations to just instructions 2542 that will be removed prior to publication as an RFC. 2544 D.3. Since draft-ietf-httpbis-messaging-01 2546 o Cite RFC 8126 instead of RFC 5226 () 2549 o Resolved erratum 4779, no change needed here 2550 (, 2551 ) 2553 o In Section 7, fixed prose claiming transfer parameters allow bare 2554 names (, 2555 ) 2557 o Resolved erratum 4225, no change needed here 2558 (, 2559 ) 2561 o Replace "response code" with "response status code" 2562 (, 2563 ) 2565 o In Section 9.4, clarify statement about HTTP/1.0 keep-alive 2566 (, 2567 ) 2569 o In Section 7.1.1, re-introduce (bad) whitespace around ";" and "=" 2570 (, 2571 , ) 2574 o In Section 7.3, state that transfer codings should not use 2575 parameters named "q" (, ) 2578 o In Section 7, mark coding name "trailers" as reserved in the IANA 2579 registry () 2581 D.4. Since draft-ietf-httpbis-messaging-02 2583 o In Section 4, explain why the reason phrase should be ignored by 2584 clients (). 2586 o Add Section 9.3 to explain how request/response correlation is 2587 performed () 2589 D.5. Since draft-ietf-httpbis-messaging-03 2591 o In Section 9.3, caution against treating data on a connection as 2592 part of a not-yet-issued request () 2595 o In Section 7, remove the predefined codings from the ABNF and make 2596 it generic instead () 2599 o Use RFC 7405 ABNF notation for case-sensitive string constants 2600 () 2602 D.6. Since draft-ietf-httpbis-messaging-04 2604 o In Section 9.9, clarify that protocol-name is to be matched case- 2605 insensitively () 2607 o In Section 5.2, add leading optional whitespace to obs-fold ABNF 2608 (, 2609 ) 2611 o In Section 4, add clarifications about empty reason phrases 2612 () 2614 o Move discussion of retries from Section 9.4.1 into [Semantics] 2615 () 2617 D.7. Since draft-ietf-httpbis-messaging-05 2619 o In Section 7.1.2, the trailer part has been renamed the trailer 2620 section (for consistency with the header section) and trailers are 2621 no longer merged as header fields by default, but rather can be 2622 discarded, kept separate from header fields, or merged with header 2623 fields only if understood and defined as being mergeable 2624 () 2626 o In Section 2.1 and related Sections, move the trailing CRLF from 2627 the line grammars into the message format 2628 () 2630 o Moved Section 2.3 down () 2633 o In Section 9.9, use 'websocket' instead of 'HTTP/2.0' in examples 2634 () 2636 o Move version non-specific text from Section 6 into semantics as 2637 "payload body" () 2639 o In Section 9.8, add text from RFC 2818 2640 () 2642 D.8. Since draft-ietf-httpbis-messaging-06 2644 o In Section 12.5, update the APLN protocol id for HTTP/1.1 2645 () 2647 o In Section 5, align with updates to field terminology in semantics 2648 () 2650 o In Section 9.1, clarify that new connection options indeed need to 2651 be registered () 2653 o In Section 1.1, reference RFC 8174 as well 2654 () 2656 D.9. Since draft-ietf-httpbis-messaging-07 2658 o Move TE: trailers into [Semantics] () 2661 o In Section 6.3, adjust requirements for handling multiple content- 2662 length values () 2664 o Throughout, replace "effective request URI" with "target URI" 2665 () 2667 o In Section 6.1, don't claim Transfer-Encoding is supported by 2668 HTTP/2 or later () 2670 Index 2672 A 2673 absolute-form (of request-target) 11 2674 application/http Media Type 40 2675 asterisk-form (of request-target) 12 2676 authority-form (of request-target) 12 2678 C 2679 Connection header field 28, 34 2680 Content-Length header field 19 2681 Content-Transfer-Encoding header field 51 2682 chunked (Coding Format) 17, 19 2683 chunked (transfer coding) 22 2684 close 28, 34 2685 compress (transfer coding) 25 2687 D 2688 deflate (transfer coding) 25 2690 F 2691 Fields 2692 Connection 28 2693 MIME-Version 50 2694 TE 26 2695 Transfer-Encoding 17 2696 Upgrade 36 2698 G 2699 Grammar 2700 absolute-form 10-11 2701 ALPHA 5 2702 asterisk-form 10, 12 2703 authority-form 10, 12 2704 chunk 23 2705 chunk-data 23 2706 chunk-ext 23 2707 chunk-ext-name 23 2708 chunk-ext-val 23 2709 chunk-size 23 2710 chunked-body 23 2711 Connection 29 2712 connection-option 29 2713 CR 5 2714 CRLF 5 2715 CTL 5 2716 DIGIT 5 2717 DQUOTE 5 2718 field-line 15, 24 2719 field-name 15 2720 field-value 15 2721 HEXDIG 5 2722 HTAB 5 2723 HTTP-message 6 2724 HTTP-name 8 2725 HTTP-version 8 2726 last-chunk 23 2727 LF 5 2728 message-body 17 2729 method 9 2730 obs-fold 16 2731 OCTET 5 2732 origin-form 10 2733 rank 26 2734 reason-phrase 15 2735 request-line 9 2736 request-target 10 2737 SP 5 2738 start-line 6 2739 status-code 14 2740 status-line 14 2741 t-codings 26 2742 t-ranking 26 2743 TE 26 2744 trailer-section 23-24 2745 transfer-coding 22 2746 Transfer-Encoding 18 2747 transfer-parameter 22 2748 Upgrade 36 2749 VCHAR 5 2750 gzip (transfer coding) 25 2752 H 2753 Header Fields 2754 Connection 28 2755 MIME-Version 50 2756 TE 26 2757 Transfer-Encoding 17 2758 Upgrade 36 2759 header line 6 2760 header section 6 2761 headers 6 2763 M 2764 MIME-Version header field 50 2765 Media Type 2766 application/http 40 2767 message/http 39 2768 message/http Media Type 39 2769 method 9 2771 O 2772 origin-form (of request-target) 10 2774 R 2775 request-target 10 2777 T 2778 TE header field 26 2779 Transfer-Encoding header field 17 2781 U 2782 Upgrade header field 36 2784 X 2785 x-compress (transfer coding) 25 2786 x-gzip (transfer coding) 25 2788 Acknowledgments 2790 See Appendix "Acknowledgments" of [Semantics]. 2792 Authors' Addresses 2794 Roy T. Fielding (editor) 2795 Adobe 2796 345 Park Ave 2797 San Jose, CA 95110 2798 United States of America 2800 EMail: fielding@gbiv.com 2801 URI: https://roy.gbiv.com/ 2803 Mark Nottingham (editor) 2804 Fastly 2806 EMail: mnot@mnot.net 2807 URI: https://www.mnot.net/ 2809 Julian F. Reschke (editor) 2810 greenbytes GmbH 2811 Hafenweg 16 2812 Muenster 48155 2813 Germany 2815 EMail: julian.reschke@greenbytes.de 2816 URI: https://greenbytes.de/tech/webdav/