- From: Willy Tarreau <w@1wt.eu>
- Date: Fri, 20 Mar 2020 07:46:18 +0100
- To: Martin Thomson <mt@lowentropy.net>
- Cc: ietf-http-wg@w3.org
On Fri, Mar 20, 2020 at 11:42:07AM +1100, Martin Thomson wrote: > On Fri, Mar 20, 2020, at 11:22, Mark Nottingham wrote: > > > On 20 Mar 2020, at 11:02 am, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote: > > > In message <ebea7134-3da6-44be-b445-7f79d1717f2b@www.fastmail.com>, "Martin Tho > > > mson" writes: > > >> Sure, the numbers will be smaller, but why have the rounding at all? > > > > > > To make sure it is not precise enough for framing use. > > This > > And so when someone rounds up and expects that to be ample space, what then? > I'm not sure that I agree that this is a valid threat model. > > Seems like over-rotating to me. If implementations routinely encounter > values that don't match the actual length, then they will learn not to depend > on the two being the same. The request was to have a size indicator to provide a progress bar, which this seems to address fine. If the requirement is to have something more accurate, then of course it's not enough, but at the same time comes again the risk that it's misused (e.g. to size a malloc()). Willy
Received on Friday, 20 March 2020 06:46:36 UTC