W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1997

Re: Common Gateway Interface

From: David W. Morris <dwm@xpasc.com>
Date: Fri, 30 May 1997 13:02:49 -0700 (PDT)
To: Dylan Barrell <dbarrell@hotmail.com>
Cc: lawrence@agranat.com, http-wg@cuckoo.hpl.hp.com
Message-Id: <Pine.SOL.3.95.970530124728.17541B-100000@shell1.aimnet.com>


On Fri, 30 May 1997, Dylan Barrell wrote:

> That depends on how you define the server. If the server constitutes all the 
> pieces which integrate with the daemon then this does form part of the
> protocol.

I would argue that the server does in fact constitute ALL elements of the
daemon and/or daemons required to respond to an HTTP request. Logically,
what you are asking for is that an internal intra daemon interface be
standardized as part of HTTP. How a server implementor chooses to
implement the inner workings of a server is totally out of scope for the
HTTP protocol. It has no direct bearing on whether an HTTP client
can successfully communicate with the server. To specify which bits of
HTTP protocol specific information should be made available to a 
particular portion of a server implemenation is unreasonable.

It is the responsiblity of the server implementor(s) to insure that
the composite of however they achieve their functionality will satisfy
the clients' expectations.

It that to argue that CGI shouldn't be standardized by the W3C or even the
IETF? Not my intent. I would even agree that it would be useful to
have it standardized but NOT as part of HTTP or by this working group.

If a particular implementation of HTTP chooses to provide a CGI interface
as a way to achieve its functionality and that instance of CGI is not
adequate for the composite server to conconform to the HTTP specification,
then that server is non-conformant with HTTP.

CGI authors have to evaluate their requirements and the capabilities of
the servers and operating systems the servers will run on and make the
appropriate implementation decisions to achieve their objectives.
It may mean not supporting a particular server or it may mean writing the
application as a dedicated special purpose HTTP server.

All out of scope for HTTP and this working group.

Dave Morris
Received on Friday, 30 May 1997 13:07:26 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:32:43 EDT