RFC 2310 (rfc2310) - Page 2 of 5

The Safe Response Header Field

RFC 2310             The Safe Response Header Field           April 1998


   It is sometimes necessary for a user agent to repeat a POST request.
   Examples of such cases are

     - when retrying a POST request which gave an indeterminate error
       result in the previous attempt
     - when the user presses the RELOAD button while a POST result is
       displayed
     - when the history function is used to redisplay a POST result
       which is no longer in the history buffer.

   If the POST request is unsafe, HTTP requires explicit user
   confirmation is before the request is repeated.  The confirmation
   dialog often takes the form of a `repost form data?'  dialog box.
   This dialog is confusing to many users, and slows down navigation in
   any case.

   If the repeated POST request is safe, the user-unfriendly
   confirmation dialog can be omitted.  However plain HTTP/1.1 [1] has
   no mechanism by which agents can tell if POST requests are safe, and
   they must be assumed unsafe by default.  This document adds a
   mechanism to HTTP, the Safe header field, for telling if a POST
   request is safe.

   Using the Safe header field, web applications which require the use
   of a safe POST request, rather than a GET request, for the submission
   of web forms, can be made more user-friendly.  The use of a POST
   request may be required for a number of reasons, including

     - the contents of the form are potentially very large
     - the form is used to upload a file (see [2])
     - the application needs some internationalization features
       (see [3]) which are only available if the form contents are
       transmitted in a request body the information in the form cannot
       be encoded in a GET request URL because of security
       considerations.

4 The Safe response header field

   The Safe response header field is defined as an addition to the
   HTTP/1.x protocol suite.

   The Safe response header field is used by origin servers to indicate
   whether repeating the received HTTP request is safe in the sense of
   Section 9.1.1 (Safe Methods) of the HTTP/1.1 specification [1].  For
   the purpose of this specification, a HTTP request is considered to be
   a repetition of a previous request if both requests





Holtman                       Experimental