- From: Michael Sweet <msweet@apple.com>
- Date: Thu, 05 Dec 2013 13:13:37 -0500
- To: Martin Thomson <martin.thomson@gmail.com>
- Cc: Mark Nottingham <mnot@mnot.net>, Ilari Liusvaara <ilari.liusvaara@elisanet.fi>, HTTP Working Group <ietf-http-wg@w3.org>
Martin, On Dec 5, 2013, at 1:06 PM, Martin Thomson <martin.thomson@gmail.com> wrote: > On 5 December 2013 08:04, Michael Sweet <msweet@apple.com> wrote: >> Keep in mind that for POST and PUT, servers should not send successful status codes (2xx) until the client has provided all of the message body for the request. > > Should not? I disagree. If the response does not depend on the rest > of the body, it's safe to send the response. True, but I've never seen a PUT or POST that does not depend on the whole request body to (safely) determine success. (I probably should have said "generally should not" or something along those lines, but that felt a bit too wishy-washy at the time...) > In HTTP/1.1, that might be inadvisable due to the bugs you describe. > However, in HTTP/2.0 a client can take the response, cancel the stream > and save the bytes without hurting anything else. Right, that's one of the many reasons why I am excited about HTTP/2.0. _________________________________________________________ Michael Sweet, Senior Printing System Engineer, PWG Chair
Received on Thursday, 5 December 2013 18:14:24 UTC