On 23/10/05 2:06 AM, "Tim Bray" <[EMAIL PROTECTED]> wrote:
> At the moment, I am really against writing spec language trying to > micromanage who sees pub:control information and who doesn't. There will be > some pub:control with a defined semantic (this is a draft). There will be > other pub:control information that will be proprietary to sixapart or google > or whoever. > I too don't want to get into micro-management of pub:control. I'd prefer one top level major-management rule for pub:control: visible only via APP. If you have other meta data you want to let float on downstream put it outside of pub:control. > I agree that in most cases, this would only be involved in editing operations > and would be effectively one-way client->server. I can think of all sorts of > applications where you might want to pass this information back downstream to > a client. I can think of no scenarios where doing so would actually harm > interoperability. > Perhaps a SHOULD requirement would suffice then? I'd be happy with SHOULD. > The implementors will sort it out. So why should we be trying to write rules > and define terms? > We didn't think through other contexts for 'self' and now we're stuck with a sub-optimal attribute value. Ditto 'alternate' for auto-discovery. Why should we be repeating these mistakes? e.
