- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Tue, 18 Mar 2014 22:16:18 +0100
- To: "Roy T. Fielding" <fielding@gbiv.com>, HTTP Working Group <ietf-http-wg@w3.org>
On 2014-03-18 22:01, Roy T. Fielding wrote: > It might help to understand that chunked requests are universally > supported by servers but not universally supported by resources. > > In particular, it is impossible to implement unlimited chunked > request bodies via a CGI script because that gateway protocol > requires the content-length be set in the environment variable > before the command is invoked. I think the same limitation is in > Servlets, because it was just a copy of CGI. In contrast, it is Nope. There's no problem with servlets with respect to this. You just obtain the ServletInputStream and start reading. > trivial to implement request streaming with an Apache module, since > the server's protocol filter handles the chunks automatically > if told to do so. It is also easy to implement limited request > buffers for a legacy back-end, configurable on a per-resource basis, > provided that sufficient protections against denial-of-sevice attacks > are in place. > > It would be a terrible mistake to limit HTTP/2 to the worst > of old implementations. That is the opposite of HTTP's design +1 > for flexible extensibility. There are hundreds (if not thousands) of > implementations of HTTP/1.1 that have no problem whatsoever with > compression, chunked encoding, or any of the other features of HTTP. Indeed. > ... Best regards. Julian
Received on Tuesday, 18 March 2014 21:16:52 UTC