Chris Wilson: So, we’re having a fun thread on the HTML WG. I’m explaining what Microsoft must do, given the half-billion or so users that rely on us to not break their web experience. A certain contingent disagrees with me (okay) and seems to want to beat it out of me (not okay).
The discussion is focused on nailing down a syntax for versioning the web (e.g.
<!DOCTYPE html5>) before agreeing to a set of specific semantics associated with that syntax (i.e., things like foo will behave as foo1 if X is specified, otherwise it will behave as foo2).
There are all sorts of ways to version the web, varying from Content-Type to User-Agent, to DOCTYPE, to version attribute, to meta tag to structured comments...
IMHO, the way forward is for the group to provisionally agree that one or more of these may ultimately be needed to be implemented by one or more vendors, and then move forward trying to triage specific proposals for semantics. The root issue is a difference in perception as to the relative sizes of these three buckets: things IE can change without breaking the web, things that won’t require changes to IE, and changes which would break the web.
Chris’s main point boils down to “the set of items that can be safely placed in the first bucket isn’t as large as you might think”, whereas one of the primary considerations of the WHATWG so far has been to maximize the second bucket to the extent that it is sane to do so. While nobody seems interested in placing things intentionally in that third bucket, if due to some oversight some specific items which were intended for the second bucket missed their mark and either landed into the first or the third, then building and working an issues list of precisely those items would in turn shed some significant light onto what versioning, if any, will be required.