W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 1995

Server Hacking

From: Roger Gonzalez <rg@server.net>
Date: Mon, 16 Oct 1995 10:07:25 -0400
Message-Id: <199510161407.KAA00539@caffeine.server.net>
To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com

Okay, you wise guys.  There's a crucial issue on the table:

I've gotten several connections with "User-agent" headers that contain
things like:

  User-agent: SomeGuyTyping/1234.1234 (ha ha)

and

  User-agent: TelnetHacker/1.1

As you can see, there is vast potential for screwing up vital client s/w
statistics-gathering.  We simply must standardize how we snoop each
others servers.  :-)

obhttp:

Can we clarify in the spec what "Accept" is used for?  I'm of the
opinion that if all you do with something is save it to disk, then
you shouldn't say you accept it.  -Anything- can just be shoved to
disk.  "Accept: */*" conveys absolutely no useful information, whereas
if we knew that "image/gif" was accepted but "image/jpeg" wasn't, the
information provider could tailor the returned resource accordingly.

I also think that we perhaps should consider adding a version
identifier to the media types accepted.  i.e. "image/gif;89a", or
"text/html;2.0". This would allow a CGI script to perhaps decide to
send out a <PRE> tabbed table instead of an HTML2.0 table.  (I
currently do this anyway, but base off the user agent, which is much
less cool.)

-Roger

Roger Gonzalez                    NetCentric Corporation
rg@server.net                     56 Rogers Street
home   (617) 646-0028             Cambridge, MA 02142
mobile (617) 755-0635             work (617) 868-8600



60 09 3A EE FE 6A 1E CC   -pgp-   B7 F7 6B 0F 00 1D 01 C7 
Received on Monday, 16 October 1995 07:02:42 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:34 EDT