From: Simon Grant
Sent: Tuesday, November 09, 2010 9:28 AM
> We have "gregorian" because there is a consensus about what Gregorian

> dates mean.

> We don't appear (yet) to have a full consensus on the interpretation of

> any other calendars.

No, I don't see it that way, the Julian calendar is sufficiently well-specified so I don't think there is an issue of consensus of its interpretation.


As I see it there is no requirement -  or I should say, no requirement has been expressed - to represent the Julian calendar within this spec. If you (or anyone) feels this is a requirement, please say so and then this becomes a different discussion.


So, assuming for the moment that it is not a requirement, the question was, why not at least add 'julian' to the list of calendars in our controlled list (which at the moment has a single member, 'gregorian')  and say "if  calendar=julian then it is a Julian date, which is not further specified by this specification".


And my objection to that is this: As soon as our list of calendars expands beyond one (and I explained why I am confortable with 'gregorian'), we put ourselves in the position of maintaining a vocabulary of calendars. If we don't maintain it, then we would need to point to an established vocabulary of calendars, and it doesn't seen that one exists. And the reason it doesn't exist is probably because calendars are complicated, which is why we would want to avoid them.


I'm not ruling this out, just saying maintaining a vocabulary of calendars wouldn't be as easy as it might sound and I don't want to do it unless we can establish that there really is a requirement.