> On Nov 6, 2014, at 5:04 PM, Patrick McManus <mcmanus@ducksong.com> wrote:
>
>
> On Wed, Nov 5, 2014 at 10:28 PM, Mark Nottingham <mnot@mnot.net <mailto:mnot@mnot.net>> wrote:
> Dare I say MAY?
>
> in this particular case I think {SHOULD NOT, MAY, SHOULD} are largely the same when implementing a conforming peer - its something to prepare for but that cannot be relied on. SHOULD is better advice, but I don't think it really will make a difference in practice so I would feel the same about both SHOULD and MAY.
What I generally hear from ADs is that SHOULD and SHOULD NOT have to come with and explanation of when you’re allowed to not follow them, for example, “SHOULD support at least 100 concurrent streams unless this is a constrained environment.” MAY can be used without guidance.
“SHOULD unless you use libraries with insufficient hooks” sounds like an implementation issue, a close relative of “SHOULD unless you really really don’t want to”. I think Mark should dare to say MAY.
Yoav
Received on Thursday, 6 November 2014 15:44:45 UTC