Re: WG Last Call: HTTP over TLS (draft-ietf-tls-https-02.txt)

On Fri, Aug 06, 1999 at 12:20:15PM -0400, Rene Eberhard wrote:

> Possibly I really don't understand that topic... If so I'm sorry.
> The draft says that [ client implementations MUST treat any premature 
> closes as errors ] *EXCEPT* the two cases which deserve special note.

It's not "except"; quite in contrary the draft says "in particular".

>> I believe that servers
>> normally simply close the connection (with a FIN) if an error
>> occurs, so the answer for Q 1 is 1.2 as well.

> Here we should have some comments from Apache group, Microsoft
> or Netscape. I can't answer this question.

Some time ago I noticed that, for HTTPS, www.microsoft.com sends RSTs
(no closure alerts, obviously -- as I remarked in a previous message
those are considered "protocol errors" by various versions of MS
Internet Explorer) even after succesful completion.  TCP connections
were cleanly closed for HTTP, though; but I have doubts that they make
a fine distinction between succesful and erroneous end-of-connection.