Re: revised CGI/1.0 specification (Rob McCool)
Message-id: <>
From: (Rob McCool)
Date: Tue, 23 Nov 1993 10:11:37 -0600
In-Reply-To: Markus Stumpf <>
       "Re: revised CGI/1.0 specification" (Nov 23,  3:27pm)
X-Mailer: Mail User's Shell (7.2.5 10/14/92)
To: Markus Stumpf <>,
Subject: Re: revised CGI/1.0 specification
 * Re: revised CGI/1.0 specification  by Markus Stumpf (
 *    written on Nov 23,  3:27pm.
 * Hoi folx,
 * I am really unhappy about the term "gateway" in the spec.
 * I cannot see any possibilities to use it as a *real* gateway.

It depends on how you define gateway, and what you're gatewaying to. It
seems our definitions are not in sync. We're referring to gateway as "an
interface to external server programs which allow you to interface with
services you may not normally have access to". This includes, finger, man,
wais, archie, and even all of the W3 supported ones (see below).

 * I'd like the possibility to receive something like
 *     GATE prot://hostname:port/some_more_info
 *     [ followed by some protocol specific info ]
 * The "gateway" could then make a connection to the host/port, send the
 * protocol specific info it received from the client and simply gate the
 * answer back to the requestor.
 * I cannot imagine how this would be possible with the current spec.

I don't see why not. I'm assuming you're referring to GATE as a new HTTP
method. The entry for SERVER_PROTOCOL is confusing since it mentions gopher.
I think we may want to change this so it's HTTP only since HTTP servers are
going to really use the spec.

The spec does not say that you must use one of the already defined methods.
If a server implemented GATE, and used CGI scripts to do so, these scripts
could easily implement any gatewaying you would want.

I'm not sure what exactly you're talking about by "followed by some protocol
specific info"... since you use a URL for the argument to GATE, this could
simply be a URL which should contain all the information you need. Thus,
ideas spring to mind of a CGI script which is linked with the W3 libraries
and does basically what CERN's server does when acting as a gateway. Since
the W3 conversion libraries all return HTML, the script could easily pass
results back to a Web browser.

If this was not what you meant, the specification allows for scripts which
blast data directly back to the client. This means scripts could gate the
answer without HTTP intervention (if that's what you want).

 * Therefor I'd prefer a name like CSI "Common Script Interface" or something
 * like that.
 * I am sorry to say that I don't have any "on the fly" proposals to be added
 * to the spec to allow for a "real gateway" behaviour as I understand it :-(
 * However, besides that I think the spec is really useful and needed.