- From: Yves Lafon <ylafon@w3.org>
- Date: Thu, 5 Jun 2008 05:01:48 -0400 (EDT)
- To: Henrik Nordstrom <henrik@henriknordstrom.net>
- Cc: Robert Siemer <Robert.Siemer-httpwg@backsla.sh>, Julian Reschke <julian.reschke@gmx.de>, Jamie Lokier <jamie@shareable.org>, Joe Orton <joe@manyfish.co.uk>, Mark Nottingham <mnot@mnot.net>, Brian Smith <brian@briansmith.org>, ietf-http-wg@w3.org
On Wed, 4 Jun 2008, Henrik Nordstrom wrote:
> tis 2008-06-03 klockan 23:29 +0200 skrev Robert Siemer:
>> Exactly. - The parties who care about message truncation should not use
>> a connection closure as message end indication. Likewise
>> should intermediaries not change the message end indication method if
>> they don't have the full message yet.
>
> Fully agreed.
>
> The thing I oppose is texts indicating that chunked encoding provides a
> guaranteed integrity. It does not. If you want something which provides
> a reasonable guarantee then Content-Length must be used.
>
>> But one question stays: should a client/proxy retry if it detects a
>> truncated message? - As I read RFC2616: yes (especially if the method is
>> safe).
>
> Perfectly fine for user agents, but not so sure about proxies. But the
> partial response MAY be cached and completed using range requests.
Ok, so if the proxy gets a reply using connection closure to signal the
end of the reply, should the proxy always treat this as an incomplete
transaction ?
> Detected truncated messages really SHOULD be considered invalid. This
> includes both chunked encoding closed without seeing the end chunk, or
> connection closed without seeing all of the response as indicated by
> Content-Length. Both cases indicates without any doubt that something
> went very wrong with the message.
>
> It's clearly wrong to process them as if they were completely valid
> messages.
>
> Regards
> Henrik
>
--
Baroula que barouleras, au tiéu toujou t'entourneras.
~~Yves
Received on Thursday, 5 June 2008 09:02:23 UTC