19.6 Compatibility with Previous Versions

It is beyond the scope of a protocol specification to mandate compliance with previous versions. HTTP/1.1 was deliberately designed, however, to make supporting previous versions easy. It is worth noting that, at the time of composing this specification (1996), we would expect commercial HTTP/1.1 servers to:

- recognize the format of the Request-Line for HTTP/0.9, 1.0, and 1.1 requests;

- understand any valid request in the format of HTTP/0.9, 1.0, or 1.1;

- respond appropriately with a message in the same major version used by the client.

And we would expect HTTP/1.1 clients to:

- recognize the format of the Status-Line for HTTP/1.0 and 1.1 responses;

- understand any valid response in the format of HTTP/0.9, 1.0, or 1.1.

For most implementations of HTTP/1.0, each connection is established by the client prior to the request and closed by the server after sending the response. Some implementations implement the Keep-Alive version of persistent connections described in section 19.7.1 of RFC 2068 [33].

19.6.1 Changes from HTTP/1.0

This section summarizes major differences between versions HTTP/1.0 and HTTP/1.1. Changes to Simplify Multi-homed Web Servers and Conserve IP Addresses

The requirements that clients and servers support the Host request- header, report an error if the Host request-header (Section 14.23) is missing from an HTTP/1.1 request, and accept absolute URIs (Section 5.1.2) are among the most important changes defined by this specification.

Older HTTP/1.0 clients assumed a one-to-one relationship of IP addresses and servers; there was no other established mechanism for distinguishing the intended server of a request than the IP address to which that request was directed. The changes outlined above will allow the Internet, once older HTTP clients are no longer common, to support multiple Web sites from a single IP address, greatly simplifying large operational Web servers, where allocation of many IP addresses to a single host has created serious problems. The Internet will also be able to recover the IP addresses that have been allocated for the sole purpose of allowing special-purpose domain names to be used in root-level HTTP URLs. Given the rate of growth of the Web, and the number of servers already deployed, it is extremely important that all implementations of HTTP (including updates to existing HTTP/1.0 applications) correctly implement these requirements:

- Both clients and servers MUST support the Host request-header.

- A client that sends an HTTP/1.1 request MUST send a Host header.

- Servers MUST report a 400 (Bad Request) error if an HTTP/1.1 request does not include a Host request-header.

- Servers MUST accept absolute URIs.

19.6.2 Compatibility with HTTP/1.0 Persistent Connections

Some clients and servers might wish to be compatible with some previous implementations of persistent connections in HTTP/1.0 clients and servers. Persistent connections in HTTP/1.0 are explicitly negotiated as they are not the default behavior. HTTP/1.0 experimental implementations of persistent connections are faulty, and the new facilities in HTTP/1.1 are designed to rectify these problems. The problem was that some existing 1.0 clients may be sending Keep-Alive to a proxy server that doesn't understand Connection, which would then erroneously forward it to the next inbound server, which would establish the Keep-Alive connection and result in a hung HTTP/1.0 proxy waiting for the close on the response. The result is that HTTP/1.0 clients must be prevented from using Keep-Alive when talking to proxies.

However, talking to proxies is the most important use of persistent connections, so that prohibition is clearly unacceptable. Therefore, we need some other mechanism for indicating a persistent connection is desired, which is safe to use even when talking to an old proxy that ignores Connection. Persistent connections are the default for HTTP/1.1 messages; we introduce a new keyword (Connection: close) for declaring non-persistence. See Section 14.10.

The original HTTP/1.0 form of persistent connections (the Connection: Keep-Alive and Keep-Alive header) is documented in RFC 2068. [33]

19.6.3 Changes from RFC 2068

This specification has been carefully audited to correct and disambiguate key word usage; RFC 2068 had many problems in respect to the conventions laid out in RFC 2119 [34].

Clarified which error code should be used for inbound server failures (e.g. DNS failures). (Section 10.5.5).

CREATE had a race that required an Etag be sent when a resource is first created. (Section 10.2.2).

Content-Base was deleted from the specification: it was not implemented widely, and there is no simple, safe way to introduce it without a robust extension mechanism. In addition, it is used in a similar, but not identical fashion in MHTML [45].

Transfer-coding and message lengths all interact in ways that required fixing exactly when chunked encoding is used (to allow for transfer encoding that may not be self delimiting); it was important to straighten out exactly how message lengths are computed. (Sections 3.6, 4.4, 7.2.2, 13.5.2, 14.13, 14.16)

A content-coding of "identity" was introduced, to solve problems discovered in caching. (Section 3.5)

Quality Values of zero should indicate that "I don't want something" to allow clients to refuse a representation. (Section 3.9)

The use and interpretation of HTTP version numbers has been clarified by RFC 2145. Require proxies to upgrade requests to highest protocol version they support to deal with problems discovered in HTTP/1.0 implementations (Section 3.1)

Charset wildcarding is introduced to avoid explosion of character set names in accept headers. (Section 14.2)

A case was missed in the Cache-Control model of HTTP/1.1; s-maxage was introduced to add this missing case. (Sections 13.4, 14.8, 14.9, 14.9.3)

The Cache-Control: max-age directive was not properly defined for responses. (Section 14.9.3)

There are situations where a server (especially a proxy) does not know the full length of a response but is capable of serving a byterange request. We therefore need a mechanism to allow byteranges with a content-range not indicating the full length of the message. (Section 14.16)

Range request responses would become very verbose if all meta-data were always returned; by allowing the server to only send needed headers in a 206 response, this problem can be avoided. (Section 10.2.7, 13.5.3, and 14.27)

Fix problem with unsatisfiable range requests; there are two cases: syntactic problems, and range doesn't exist in the document. The 416 status code was needed to resolve this ambiguity needed to indicate an error for a byte range request that falls outside of the actual contents of a document. (Section 10.4.17, 14.16)

Rewrite of message transmission requirements to make it much harder for implementors to get it wrong, as the consequences of errors here can have significant impact on the Internet, and to deal with the following problems:

  1. Changing "HTTP/1.1 or later" to "HTTP/1.1", in contexts where this was incorrectly placing a requirement on the behavior of an implementation of a future version of HTTP/1.x

  2. Made it clear that user-agents should retry requests, not "clients" in general.

  3. Converted requirements for clients to ignore unexpected 100 (Continue) responses, and for proxies to forward 100 responses, into a general requirement for 1xx responses.

  4. Modified some TCP-specific language, to make it clearer that non-TCP transports are possible for HTTP.

  5. Require that the origin server MUST NOT wait for the request body before it sends a required 100 (Continue) response.

  6. Allow, rather than require, a server to omit 100 (Continue) if it has already seen some of the request body.

  7. Allow servers to defend against denial-of-service attacks and broken clients.

This change adds the Expect header and 417 status code. The message transmission requirements fixes are in Sections 8.2, 10.4.18,, 13.11, and 14.20.

Proxies should be able to add Content-Length when appropriate. (Section 13.5.2)

Clean up confusion between 403 and 404 responses. (Section 10.4.4, 10.4.5, and 10.4.11)

Warnings could be cached incorrectly, or not updated appropriately. (Section 13.1.2, 13.2.4, 13.5.2, 13.5.3, 14.9.3, and 14.46) Warning also needed to be a general header, as PUT or other methods may have need for it in requests.

Transfer-coding had significant problems, particularly with interactions with chunked encoding. The solution is that transfer- codings become as full fledged as content-codings. This involves adding an IANA registry for transfer-codings (separate from content codings), a new header field (TE) and enabling trailer headers in the future. Transfer encoding is a major performance benefit, so it was worth fixing [39]. TE also solves another, obscure, downward interoperability problem that could have occurred due to interactions between authentication trailers, chunked encoding and HTTP/1.0 clients.(Section 3.6, 3.6.1, and 14.39)

The PATCH, LINK, UNLINK methods were defined but not commonly implemented in previous versions of this specification. See RFC 2068 [33].

The Alternates, Content-Version, Derived-From, Link, URI, Public and Content-Base header fields were defined in previous versions of this specification, but not commonly implemented. See RFC 2068 [33].