- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Tue, 30 May 2023 13:26:06 -0700
- To: fantasai <fantasai.lists@inkedblade.net>
- Cc: spec-prod <spec-prod@w3.org>
On Tue, May 30, 2023 at 12:56 PM fantasai <fantasai.lists@inkedblade.net> wrote: > - Should this markup be optional or required? Not all specs currently follow > this style: e.g. some specs just use regular casing, and define all instances > of these keywords to invoke RFC2119. Notably, the CSSWG has *never* used this markup convention, and in any case often uses slightly different wording that still makes assertions and should be interpreted in the same way as 2119 wording - see <https://drafts.csswg.org/css/#w3c-conventions>. (And also see <https://infra.spec.whatwg.org/#conformance> and <https://infra.spec.whatwg.org/#algorithms>, which covers WHATWG specs generally, and informally covers a bunch of W3C specs.) So, assuming the CSSWG/WHATWG/lots of other W3C specs usage isn't confusing anyone, that's a decent argument against not having to do this funky markup+formatting. And if people *are*, for whatever reason, depending on that formatting to tell where a normative requirement is, they'll be missing a ton anyway where we say something "is" X, or "To X: <list steps>", which are also MUST-equivalent. So I strongly recommend that we amend the Manual Of Style to drop the formatting requirement. In general I propose adopting the Infra spec's wording entirely; its subsetting of the 2119 keywords is good, and it contains good suggestions for alternate wording that avoids tripping over the keywords. ~TJ
Received on Tuesday, 30 May 2023 20:26:29 UTC