- From: Rowland Shaw <Rowland.Shaw@crystaldecisions.com>
- Date: Tue, 19 Nov 2002 06:41:02 -0800
- To: "Mikko Rantalainen" <mira@st.jyu.fi>, <www-html@w3.org>
Mikko Rantalainen wrote: Rowland Shaw wrote: > > Christoph Päper wrote: > >> > >> <proposal> > >> > >> Element | Attributes | Minimal Content Model > >> ---------+-------------------------+----------------------- > >> nr | Common, value (Number), | (PCDATA | Inline)* > >> | dim (Dimension) | > >> > >> X.Y The *nr* element > >> > >> The nr element indicates that a text fragment is of a numeric kind (e.g. > >> date, measurement, price). It may supply a standardized representation of > >> the enclosed term and may then be treated as a replaced element: the UA > >> replaces the contents by a form best matching the document language and > >> user preferences. > > > Seems like a reasonable suggestion, a few comments I'd make: > > > * I would see dates in a separate element > > > I'd prefer the date and time to be something more akin to: > > > <p>It's my birthday on <date month="11" day="23" year="2002" > > form="shortdate">23/11/02</date></p> > > Please, we have perfectly good standard for date and time presentation: > ISO-8601. I agree that we need both "generic numeric element" and > "datetime element". I deliberately steered clear of interpreting strings in that manner. Using separate numerical attributes for the components removes ambiguity, and simplifies parsing -- I intended the content to be an "alt" for non-conforming agents, and in the case of documents I write, this will generally by in English, as used in Suffolk, in the UK, hence the example used the UK short date form as it's alt. The information about that date remains semantically the same depending on the clients' locale. My other intention was to divorce dates and times. If I were to list the hours that I'm in the office, I'd say from 7:30 to 16:30, UK local time (so BST/GMT) I wouldn't say that I'm in the office from 19/11 7:56 -> 19/11 16:30, 21/11 7:30 -> 21/11 16:30 (etc) Point here being that there is effectively two types of time in regular use -- an instant in time, so 19-Nov-2002 14:31:10 GMT; and a "regular" time, so of the form, "I normally start work by 8:00 in the morning" Maybe a container that joins the two together, xor behaviour that defines whether the information was specified (hint: using "0" could imply midnight, amongst other flaws )
Received on Tuesday, 19 November 2002 09:43:54 UTC