- From: Daniel Stenberg <daniel@haxx.se>
- Date: Wed, 4 Sep 2013 10:39:58 +0200 (CEST)
- To: Martin Thomson <martin.thomson@gmail.com>
- cc: HTTP Working Group <ietf-http-wg@w3.org>
On Tue, 3 Sep 2013, Martin Thomson wrote: > I think that the direction we were heading was something like: 100-continue > + HTTP/2.0 = no. Yes. I don't have any big problem with that other than the fact that when I introduce http2 support to curl, it will make the request act a little different if the user asks for it to go over http2 rather than using vanilla http - for the rare servers that actually does something sensible with 100-continue... A command line example to a server that rejects a POST would be: curl --http2.0 -d @largishfile http://example.org/ vs curl --http1.1 -d @largishfile http://example.org/ This client can't have any prior knowledge about HTTP2 support and this is a POST done as first request with large body... The effect seems to risk a little more data waste in the HTTP+upgrade case. I won't lose any sleep over it. =) -- / daniel.haxx.se
Received on Wednesday, 4 September 2013 08:40:26 UTC