Cookies & Netscape

That's a dilemna that I don't know how to solve yet.

The only possible way out of date, as of today is to check the user
agent in the cookie support, which I am not really happy to do. I'll
wait until something is decided in the http working group. In the
meantime you may want to revert the cookie support to the old format
(pretty easy, check w3c.www.http.HttpCookieList, and make it not emit
the Version, etc fields)

Anselm.

Brian Millett writes:
 > Anselm,
 > 	I read the draft on http state management and yes the way to send
 > multiple Cookies is with a comma separated list of cookies.  However
 > Netscape 3.01 does not process this protocol correctly.  If the cgi
 > sends the following headers:
 > 
 >  
 > HTTP/1.0 200 OK
 > MIME-Version: 1.0
 > Set-Cookie: uid=fred; path=/,pwd=foo; path=/
 > Content-type: text/html
 > 
 > Then Netscape sends back to the server the following:
 > 
 > 11116 STDIN(REQ): POST /cgi-bin/test-cookie HTTP/1.0
 > 
 > 11116 STDIN(HEADERS): Referer: http://vlad:8080/cgi-bin/test-cookie
 > 11116 STDIN(HEADERS): Connection: Keep-Alive
 > 11116 STDIN(HEADERS): User-Agent: Mozilla/3.01Gold (X11; U; SunOS 5.4 sun4m)
 > 11116 STDIN(HEADERS): Host: vlad:8080
 > 11116 STDIN(HEADERS): Accept: image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, */*
 > 11116 STDIN(HEADERS): Cookie: uid=fred
 > 11116 STDIN(HEADERS): Content-type: application/x-www-form-urlencoded
 > 11116 STDIN(HEADERS): Content-length: 0
 > 11116 STDIN(HEADERS): 
 > 
 > The Cookie: uid=fred is not correct.
 > 
 > So my question is, should Jigsaw be backward compatible?  Or should I
 > just plan on it not working until Netscape fixes their browser?
 > 
 > Thanks
 > 
 > -- 
 > Brian Millett                    
 > Technology Applications Inc.     "Heaven can not exist,
 > (314) 530-1981                          If the family is not eternal"
 > bpm@techapp.com                   F. Ballard Washburn
 > 
 >

Received on Monday, 24 February 1997 03:11:34 UTC