- From: Robert J Burns <rob@robburns.com>
- Date: Mon, 16 Mar 2009 06:32:20 -0500
- To: Aryeh Gregor <Simetrical+w3c@gmail.com>
- Cc: Tom Duhamel <tom420.duhamel@gmail.com>, whatwg@lists.whatwg.org, "public-html@w3.org" <public-html@w3.org>
Hi Aryeh, On Mar 15, 2009, at 6:59 PM, Aryeh Gregor wrote: > On Sun, Mar 15, 2009 at 6:30 PM, Tom Duhamel > <tom420.duhamel@gmail.com> wrote: >> What about those 'other calendars of concern'? Are they reasonably >> compatible with Gregorian, or so much different that my example of >> the Mayan >> long count becomes a good one? I know nothing about those, but I >> fear it's >> the later. > > You're correct. For instance, the Jewish calendar has 12 months of 29 > or 30 days, except that sometimes it has 13 months, and it only syncs > up with the Julian calendar every 19 years, except sometimes when it's > off by a day, and the algorithm for calculating whether months have 29 > or 30 days depends on when the new moon will hypothetically first be > visible in Jerusalem according to an astronomical model that was > finalized early in the first millennium CE, plus also on when certain > holidays would fall out in the coming year. > > I don't know much about the other calendars, but they're similarly > unrelated to the Gregorian calendar, for sure. Apparently, some > variants of the Islamic calendar aren't even planned out in advance -- > you literally have no idea whether the current month is going to be 29 > days or 30 until it's actually announced based on observation of the > new moon, and different Muslim countries decide it differently. There > are variants that allow the start of Islamic months to be calculated > in advance, but there seem to be several different ones. > However, in terms of the question Tom poses – which is much more about calendar date representation – the dissimilarity of the Mayan Long Count calendar is not shared by most other calendars. Most calendars – in contrast to the Mayan long count – can be represented in terms of years, months and ordinal days of the month in exactly the way that ISO 8601 handles Gregorian date representation. Even the Mayan Long Count calendar can be represented in something resembling ISO 8601 by allowing more than the three fields for the date (e.g., "12-19-16-03-03t19:57+00:00"). To say that calendars do not match up except at certain times is the same as saying they are different calendars (like saying that a metric and US measuring tape only match at the origin). Many calendars try to match their year to the tropical solar cycle but even the Gregorian does not lead to the tropical events occurring precisely at the same time each year (I've heard arguments about whether the equinox is on the 21st or the 20th of the month when the point is it drifts within a certain range). Astronomy also tells us that the length of the tropical cycle itself changes over time so treating the Gregorian as the more accurate (or calculation as superior to observation) is a little like celebrating the amazing atomic-clock-like precision of a broken clock that occurs each day. On the other hand, though the Islamic calendar is not a tropical solar calendar, in terms of representation it still involves years (potentially more than 4 digits), months (two digits), and days (two digits) just like the Gregorian and these other calendars (note the Julian calendar requires something like 445 days for one particular year, but that still doesn't push the limits of the ISO 8601 date representation approach as none of the months exceed 99 days and the months also remained below 99). Again, HTML UAs can easily allow the representation of these various calendar dates without imposing any requirement burden on UAs to support conversion and extraction of these alternative calendar dates (merely a requirement to recognize calendar keywords and ignore unsupported calendars; in the case of Julian the ISO-8601-style representation translates to the exact same localized month names from 44 BC onward). So support for representing dates in multiple calendars does not have to include support for converting among dates. Specialized UAs (or even specialized web services) might provide facilities to perform the conversion. In terms of HTML we only need to concern ourselves with date representation. In addition I think many UAs will need to support extraction of Gregorian dates (though again other UAs might allow extraction and conversion of alternative calendar dates). There's a reluctance to force implementors to perform conversion between alternative calendar dates. OK fine. But let's not make authors convert those dates either simply to make use of a feature to unambiguously encode the date in a machine readable form. Let's allow that feature to support dates even without conversion (especially since we're merely looking for an unambiguous uniform approach to encoding dates and not conversion, extraction, comparison, etc.). Take care, Rob
Received on Monday, 16 March 2009 11:33:24 UTC