Re: HTTP response version, again

On Sat, 21 Dec 1996, Ben Laurie wrote:

> It also seems to me that the spec is not clear on this issue. There is clear
> intent in HTTP/1.1, in that the word "major" was added to the version of
> the response, but, AFAICS, no clear requirement to respond either 1.0 or 1.1 to
> a 1.0 request. Since a requirement of the spec is to be liberal in what is
> accepted, it seems to me that the correct interpretation of the spec is that
> a 1.1 reponse to a 1.0 request is permitted.
> I forget where we are wrt modifications to the spec. It seems to me that a
> modification should be made to clarify this point, whichever way it goes.

IMHO, I don't think it's particuarly neccessary for the spec to
specify how a HTTP/1.1 server has to respond to HTTP/1.0 requests. As
was pointed out, even if the spec did say that it had to respond with
HTTP/1.1, that doesn't mean anything, because in fact when a HTTP/1.0
request came in, the server could just use HTTP/1.0 semantics and
would be perfectly within its rights to respond with HTTP/1.0. I don't
think it's neccessary to specific one over the other - both can
coexist perfectly well. (that being said, I do favor responding to
HTTP/1.0 with HTTP/1.1).

However, it seems certainly desirable to add some language to the spec
along the lines of "if a message of a version with a known major
number, but an unknown minor number is received (e.g., if a HTTP/1.0
client recieves a HTTP/1.1 response, or a HTTP/1.2 server receives a
HTTP/1.7 request), it should be treated as if it was a request
equivilent to the highest minor number in that major version that is

Alexei Kosut <>      The Apache HTTP Server

Received on Saturday, 21 December 1996 12:04:19 UTC