Re: POST premature connection close

I am still trying to pin down the problem where a client does not get an
answer to a POST request sent via Squid.

To identify whether Squid or the third-party client is responsible
for the wrong behavior, I read RFC 1945 (HTTP 1.0) and found

   Except for experimental applications, current practice requires that
   the connection be established by the client prior to each request and
   closed by the server after sending the response. [...] In
   any case, the closing of the connection by either or both parties
   always terminates the current request, regardless of its status.

This seems to indicate that the client is not allowed to shutdown the
sending part of the socket while it is waiting for a response.
Are there any different interpretations, or documentation to support it?

Regards

-- 
Peter Heimann