The following reply was made to PR general/4033; it has been noted by GNATS.
From: John-Mark Gurney <[EMAIL PROTECTED]> To: Marc Slemko <[EMAIL PROTECTED]> Cc: Apache bugs database <[EMAIL PROTECTED]> Subject: Re: general/4033: HTTP/0.9 requests yield invalid response Date: Thu, 11 Mar 1999 12:09:18 -0800 Marc Slemko scribbled this message on Mar 11: > On Thu, 11 Mar 1999, John-Mark Gurney wrote: > > > > Synopsis: HTTP/0.9 requests yield invalid response > > > > > > State-Changed-From-To: open-feedback > > > State-Changed-By: marc > > > State-Changed-When: Wed Mar 10 15:21:51 PST 1999 > > > State-Changed-Why: > > > Erm... a HTTP/0.9 request, by definition, doesn't have > > > a version number in the request but is just in the form > > > "GET /foo". Do you have some reason for thinking that > > > specifying it should work? > > > > yes, the rfc specificly states this (section 6.1): > > A Simple-Response should only be sent in response to an HTTP/0.9 > > Simple-Request or if the server only supports the more limited > > HTTP/0.9 protocol. > > > > also, earlier it is stated that you can't upgrade the major version > > of a http request only the minor, meaning you should AT LEAST return > > an error and not service the request... > > Except that a HTTP/0.9 request, by definition, is NOT a request that says > "GET /foo HTTP/0.9". > > It is arguable that perhaps you should special case things to allow the > invalid "GET /adf HTTP/0.9" request, but... that is NOT a HTTP/0.9 > request. oh well, you can close it, but it doesn't matter, the spec says that a difference of major version numbers isn't transport compatible, and apache is assuming that major version 0 of http is transport compatible with major version 1... it doesn't matter that it's a malformed HTTP/0.9 request, apache shouldn't return an HTTP/1.x response as it either should return a simple request or return an invalid request... -- John-Mark Gurney Eugene 4J School District
