W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > January to April 1998

new editorial(?) issue HEAD_WITH_IMS

From: Jeffrey Mogul <mogul@pa.dec.com>
Date: Wed, 04 Mar 98 14:37:57 PST
Message-Id: <9803042237.AA00826@acetes.pa.dec.com>
To: http-wg@cuckoo.hpl.hp.com
Dave Kristol and I have been working in private to nail down
a precise description of how the various conditional headers
(If-Modified-Since, If-None-Match, etc.) interact.

DESCRIPTION OF PROBLEM:

One new question came up: is it "legal" to send an
If-Modified-Since header with a HEAD method?

The current revision of the specification does not address
this question directly.  However, it makes two somewhat
contradictory statements:

Section 9.4 (HEAD) says:
    The HEAD method is identical to GET except that the server MUST NOT
    return a message-body in the response.

"identical" implies that HEAD+IMS should be the same as GET+IMS,
except that a 200 response would carry no body (a 304 response
never has a body, of course).

On the other hand, section 14.24 (or 14.25, depending on whether
the alphabetical ordering has been fixed in your copy of -rev-02)
says
	The If-Modified-Since request-header field is used with the
	GET method to make it conditional ...
but does not mention the possibility of using this header with
any other method.

(The specifications of the other four If-* headers do not mention
a specific method, although If-Range is introduced as a way
to avoid an extra GET request.)

PROPOSED SOLUTION:

Our intuition is that, based on the language in 9.4, any headers
allowed with GET should be allowed with HEAD (and interpreted
the same way by the server).  I.e., the language in 14.24
should be changed to
	The If-Modified-Since request-header field is used with the
	GET (or HEAD) method to make it conditional ...
or even
	The If-Modified-Since request-header field is used with a
	method to make it conditional ...

WHAT SERVERS ALREADY DO:

Although I'm not a fan of testing existing implementations to
see what the specification should say, I'll note that Apache/1.2.5
seems to allow and support HEAD+IMS:

    % telnet www.freebsd.org 80
    Trying 204.216.27.18...
    Connected to hub.freebsd.org.
    Escape character is '^]'.
    HEAD / HTTP/1.0
    
    HTTP/1.1 200 OK
    Date: Wed, 04 Mar 1998 22:20:33 GMT
    Server: Apache/1.2.5
    Last-Modified: Wed, 04 Mar 1998 12:02:49 GMT
    ETag: "2742a6-2fa4-34fd42e9"
    Content-Length: 12196
    Accept-Ranges: bytes
    Connection: close
    Content-Type: text/html
    
    Connection closed by foreign host.
    % telnet www.freebsd.org 80 
    Trying 204.216.27.18...
    Connected to hub.freebsd.org.
    Escape character is '^]'.
    HEAD / HTTP/1.0
    If-Modified-Since: Wed, 04 Mar 1998 12:02:49 GMT
    
    HTTP/1.1 304 Not Modified
    Date: Wed, 04 Mar 1998 22:20:44 GMT
    Server: Apache/1.2.5
    Connection: close
    ETag: "2742a6-2fa4-34fd42e9"
    
    Connection closed by foreign host.
    %
    
By the way, Apache/1.2.5 also returns "304" when the request includes
    If-None-Match: "2742a6-2fa4-34fd42e9"
instead of
    If-Modified-Since: Wed, 04 Mar 1998 12:02:49 GMT

-Jeff
Received on Wednesday, 4 March 1998 14:40:55 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:33:13 EDT