W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2014

Re: CRLF requirement

From: David Krauss <potswa@gmail.com>
Date: Wed, 2 Jul 2014 10:35:29 +0800
Cc: HTTP Working Group <ietf-http-wg@w3.org>
Message-Id: <104D8169-BA56-4565-8942-DF9F6E2314A4@gmail.com>
To: Glen Knowles <gknowles@ieee.org>

On 20140702, at 10:18 AM, Glen Knowles <gknowles@ieee.org> wrote:

> Changing "Connection: close" to "nnCoection: close" is something I have seen some "zero copy" proxies do when they wanted to remove the connection header. The theory is you just load a 32 bit "Conn", rotate it 16 bits to "nnCo", put it back, and rely on the upstream server to ignore it as an unrecognized header.

Hmm, if it was a convention among several implementations at the time, perhaps it should be recorded in the standard. /s

Also, such bitwise operations for string manipulation suggest unaligned memory access, which is a uniquely risky way of parsing. The same effect would be achieved by substituting the first character. Lemmings.
Received on Wednesday, 2 July 2014 02:36:10 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 30 March 2016 09:57:08 UTC