- From: Willy Tarreau <w@1wt.eu>
- Date: Thu, 29 Oct 2020 05:04:40 +0100
- To: Cory Benfield <cory@lukasa.co.uk>
- Cc: Mark Nottingham <mnot@mnot.net>, HTTP Working Group <ietf-http-wg@w3.org>, Poul-Henning Kamp <phk@varnish-cache.org>
On Wed, Oct 28, 2020 at 10:12:37AM +0000, Cory Benfield wrote: > I don't love this change. It's a frustrating additional complexity in > the parser, and it seems unfortunate that we're adding it for reasons > that have nothing to do with whether we want it and everything to do > with limitations of the format we publish the spec in. I think we > shouldn't do it because if we wrote down in the RFC _why_ SP is > allowed, we'd feel fairly silly. I agree. We don't adjust on-wire protocols to match a pretty-printer. And doing just this could easily encourage lazy implementations to bring line folding back in force after we've deprecated it and added the "MUST NOT send" rule. Better adopt RFC8792 as mentioned somewhere else in this thread! I've seen Mark's patch making use of it and it looks safe and very clear to me. Willy
Received on Thursday, 29 October 2020 04:05:01 UTC