- From: John C. Mallery <jcma@ai.mit.edu>
- Date: Thu, 29 May 1997 19:35:25 -0400
- To: Hallam-Baker <hallam@ai.mit.edu>
- Cc: Dylan Barrell <dbarrell@hotmail.com>, http-wg@cuckoo.hpl.hp.com
At 5:24 PM -0400 5/29/97, Hallam-Baker wrote: >> Why is there absolutely no mention of the CGI interaction between the HTTP >> daemon and the CGI programs? Surely this standard should specify which peices> the header are made available to CGI programs and how they are made available > > >Because CGI is not an IETF spec, it was a hack put together by Ari >Luotenen of CERN and Rob McCool at NCSA to allow people to plug >stuff into both their servers. While their Netscape server supports >CGI the prefered way to plug in a module is via NSAPI which is >much more efficient because it does not spawn a process for each >transaction. > >CGI is completely inappropriate to standardize in the IETF. It is >operating system specific and its an API and not a protocol. The >W3C might have an interest in producing a spec but I would not >count on it. It might be nice to have a document readily available >with Ari and Rob's names on it but it really isn't a technology >that should be considered leading edge at this point. NSAPI and >ISAPI have replaced it, there are many, many such APIs and there >is no particular reason to consider CGI as a special case even >if popular Web books have "CGI" on the cover in letters three >inch high. Actually, the Java server API or the earlier, and more general, the cl-http API are cleverer than the ns or ms APIs
Received on Thursday, 29 May 1997 16:42:46 UTC