Some comments on http protocol draft

• Mail folder: WWW Talk Oct 93-present
• Next message: Rob McCool: "Re: revised CGI/1.0 specification"
• Previous message: Markus Stumpf: "Re: revised CGI/1.0 specification"

Subject: Some comments on http protocol draft
From: Markus Stumpf <stumpf@informatik.tu-muenchen.de>
To: www-talk@nxoc01.cern.ch
Date: 	Tue, 23 Nov 1993 16:28:24 +0100
X-Mailer: ELM [version 2.4 PL6]
Content-Type: text
Content-Length: 2094      
Message-id: <93Nov23.162827mesz.311438@hprbg5.informatik.tu-muenchen.de>

Hello,

I'd like to make some comments on the HTTP draft recently published by the
IETF (draft-ietf-iiir-http-00.{txt,ps} dated 5 Nov 1993.

1) on the METHODS:
   I'd really like to have a method called "GATE", which would be
   followed by a URL and any other method.
   This would allow servers to only open a connection to the specified
   host/port, send the "real" method and information from the requestor
   and send back the answer without having to further interpret the
   data. Using the URL with the GATE method allows for access restrictions
   on the server side based on protocol (i.e. allow all http requests
   regardless the port number they use).

2) I have thought on some mechanisms providing help in administering
   HTML documents.
   Using the "Referer:" field of the request a server may detect
   originators of bad links (as already stated in the draft).
   How about some protocol, at best UDP based I think, that would allow
   servers to exchange this kind of information.
   If a server detects such a bad link and has a requestor field it
   could send a datagram to the originating server. It could look like
      BADLINK  <requested URL>   <originators URL>
   Maybe we could get UDP port 80 for this (dunno whether this is already
   allocated by some other service). Server could then save this information
   to some logfile and/or notify the servers admin.
  
   Maybe we could have another HTML+ tag like <TITLE> that could be
   called <AUTHOR> and contain a email address of the author(s) of the
   document. This could be passed along with the request maybe in a
   "Author:" field or (if not present within the document) a server could
   send back the email address(es) of the Author of a document along
   with the response code or input it in the stream of a html document
   (which I think would be bad, as the server would have to interpret
   the document in some way).

Bye
	\Maex
-- 
______________________________________________________________________________
 Markus Stumpf                        Markus.Stumpf@Informatik.TU-Muenchen.DE