Re: LYNX-DEV Older HTTP Versions

[David Woolley]

> request, and the server sends back something like a HTTP/1.0 error 400 (Bad
> Request), Lynx used to try to fetch the document again with "HTTP0" - a
> simple request.  Now, it doesn't do that.

What we are talking about here is broken server heuristics, or just possibly
a broken request from Lynx.  The server has returned a message claiming that
it is an HTTP1 server (an HTTP0 server doesn't return headers).  The chances
are that a minimal HTTP1 request would be as effective as an HTTP0 one.
(HTTP1 = HTTP/1.0 (GET /xxxx HTTP/1.0\nheaders); HTTP0 = HTTP/0.9 (GET /xxxx)

The most commonly reported broken server problem is servers which don't like
large lists of acceptable content types.  Pruning mailcap files is one
way of testing for this (I'm on metered call dial-up, so it's not convenient
to actually test your URL).

You didn't include the Server header, so we can't tell, offline, whether it
is one of the products known to be broken in this way.

> 
> It looks to me like Lynx has fooled itself into thinking that everything's
> OK, when in fact the document couldn't be fetched.

If Lynx were to fall back to HTTP0, it would have to accept any response as
OK, as the server can only indicate a failure in the content, so asking for
an HTTP0 attempt is asking for Lynx to fool itself.