- From: Adrien de Croy <adrien@qbik.com>
- Date: Thu, 22 Nov 2007 10:40:12 +1300
- To: Bjoern Hoehrmann <derhoermi@gmx.net>
- CC: 'HTTP Working Group' <ietf-http-wg@w3.org>
ah, thanks. OK, this is an issue close to my heart, since it affects POST / PUT through a proxy which uses any sort of challenge/response (i.e NTLM) authentication mechanism. Our users have major problems with this. It would be great if the concept of chunking could be stretched to include notification of an abort of transfer. i.e. send a chunk with a length of "-1" or something. Obviously that's a protocol level change, but it would allow the connection to be maintained after an abort being signaled. I haven't seen a client that uses chunking for POST or PUT though - I guess in most if not all cases, the client knows the content length, and so specifies it. Should we be putting a stronger emphasis on clients to use chunking in all cases? That has down-sides as well, since then the client can't advertise apriori the length it will be sending (to allow for proxy / server policy evaluation based on upload sizes etc). This actually could be a good use-case to use chunking AND specify a content-length header (I hear the groans) - something specifically deprecated in the spec. People commonly want to limit download and upload sizes. The only way to limit upload sizes efficiently is if clients sent Content-Length headers. Otherwise you have to receive / process data until you hit the limit. But actually I believe in the end, the issue can't be properly addressed without a major protocol version change, one that then incorporates the concept of negotiating (i.e. obtaining mutual consent for) transfer of message bodies from client to server (as opposed to only having negotiation in the other direction). Regards Adrien Bjoern Hoehrmann wrote: > * Adrien de Croy wrote: > >> what sort of network error is being envisaged here? >> > > This is about the server telling you it won't handle the request while > it is being sent (e.g., you are trying to PUT something to a resource > that requires authentication, or after the 42nd chunk the server decides > the entity is longer than it is willing to handle). It's not about any > lower level network problems. > -- Adrien de Croy - WinGate Proxy Server - http://www.wingate.com
Received on Wednesday, 21 November 2007 21:39:42 UTC