httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marc Slemko <>
Subject Re: 1.2.4?
Date Thu, 21 Aug 1997 16:22:14 GMT
On Thu, 21 Aug 1997, Alexei Kosut wrote:

> On Thu, 21 Aug 1997, Marc Slemko wrote:
> > As a reminder, 1.2.3 works for very few servers because it sends things
> > like:
> > 
> >
> I should probably point out that this will work when proxying to any Apache
> 1.2+ server (at least, it should - I haven't tried since I wrote it, and it
> may have been broken by some of the virtual host stuff since), and anyone
> else who implemented a HTTP/1.1 server and actually read section 5.1.2
> ("all HTTP/1.1 servers MUST accept the absoluteURI form in requests, even
> though HTTP/1.1 clients will only generate them in requests to proxies.")
> But then, I doubt most of the supposedly "HTTP/1.1" servers support
> this. I have yet to see a supposedly HTTP/1.1 server other than Apache
> that implemented section 14.23: "All Internet-based HTTP/1.1 servers MUST
> respond with a 400 status code to any HTTP/1.1 request message which
> lacks a Host header field."

I have yet to see one that doesn't.

Hey, when did switch to Apache?

Oh, says Apache 1.2 will be out.
Whoever should be contacted to update that...

CL-HTTP/63.61 (written in Lisp): sends a Location: header in all
responses.  Running on  Gives a "400 Bad
request: missing host header" in response to requests without Host:
headers.  It properly handles full URIs.

DMKHTD/1.05j (Dave Kristol's server) runs on
It rejects requests without a Host: header.  It does not reject full URIs
without a Host: header.  That makes sense if you think about it, since it
can figure it out from the hostname, but I don't think it is legal.

Agranat-EmWeb/R3_0alpha6 runs on  It rejects
requests without a Host: header.  It accepts full URIs without a host

Jigsaw 1.0beta (hmm... what is the Server: header supposted to be?)
running at rejects requests without Host: headers
and doesn't even give an error; just closes the connection!  It also
accepts full URIs without Host: headers.

All of them except Agranat's and Jigsaw (and Apache...) give pretty error
messages when the Host:  header isn't there.

Hmm, IBM's IICS (remember, the one that used chunks of around 5 or 10
bytes?) does HTTP/1.1, but I can't find one right now.  Found half a dozen
other ones at IBM though.  Aha.  Returns
an error if no Host: header is sent.  Does send a content-location.  Still
does really stupid chunking.  Returns what looks like a bogus
Content-length for a HEAD request; hmm, perhaps it isn't bogus, I just
think it odd that they can return a Content-length for a parsed document
yet choose to chunk it in a very ugly way.

Hmm.  IBM-ICS/4.1.1 at returns "torsdag, 21-aug-97
16:22:34 GMT" as a date.  Didn't think that was allowed.

View raw message