[tei-council] Dates and calendars

[Martin Holmes]

On 12-08-14 11:13 AM, Gabriel Bodard wrote:
> On 2012-08-14 16:31, Martin Holmes wrote:
>>> What is this schematron supposed to be constraining? No @calendar if
>>> <date> (etc) is empty?
>> I think so. That should be true, shouldn't it?
>
> I'm not clear what an empty <date> means, myself, so I'm not sure I can
> imagine what the danger of allowing @calendar on it is.

As Lou pointed out, @calendar "indicates the system or calendar to which 
the date represented by the content of this element belongs." If there 
is no content, then having @calendar is presumably wrong. However, I 
don't think it really should be; I think it's perfectly reasonable to 
encode something like this:

<date when-custom="1642-02-02" calendar="#julian"/>

when you want to encode a Julian date in a formal manner without 
transcribing any particular source from which it came. Consider this 
title of a modern document:

<title level="a">Management and Mismanagement at
               Bedlam, 1547 to 1633</title>

The two years in the title are presumably Julian, with the year running 
from March 25 to March 24. If you wanted to document the temporal 
coverage of this article, you might encode it like this:

<date from-custom="1547" to-custom="1633" calendar="#julian"/>

including the calendar information because those years are not 
equivalent to the proleptic Gregorian years with the same numbers.

> Is the idea that
> @calendar should only be used when the date is used in transcription of
> a primary text?

That would seem to be the intention of the attribute definition, 
although I think that's a bit too restrictive.

>>> No *-custom without @datingPoint|@datingMethod?
>> I had thought so, but then I realized I wouldn't want to enforce either
>> of those things. If @calendar points to a <calendar> with a detailed
>> description of the dating system, then there's no point in @datingMethod
>> (which also "supplies a pointer to a <calendarDesc> element or other
>> means of interpreting the values of the custom dating attributes".
>> (Incidentally, should that be <calendarDesc> or just <calendar>?)
>> Similarly, @datingPoint is obviously optional.
>>
>> However, I think it's true to say, isn't it, that if you have any of
>> @*-custom, you should have one of @calendar, @datingMethod or @datingPoint?
>
> Yes, that is almost certainly true. So is the rule, report @*-custom
> without @calendar|@datingPoint|@datingMethod? That seems responsible,
> although I'm one of those people who often uses sloppy behaviour in
> practice that would fall foul of this.
>
> I worry a bit about the two above rules being a bit contradictory,
> though. We don't complain about @*-custom on an empty element; but we do
> complain about @*-custom without @calendar|@dP|@dM, so I add @calendar
> to make it valid, and it then complains about @calendar on the empty
> element. (Maybe @calendar has no bearing on @*-custom at all, though?)

That would be worrying, because the purpose of @*-custom is to provide a 
formal way of recording a date in a non-gregorian calendar.

On balance, I think only the one rule is realistic: if you have any of 
@*-custom, you should have one of @calendar, @datingMethod or @datingPoint.

Cheers,
Martin

-- 
Martin Holmes
University of Victoria Humanities Computing and Media Centre
(mholmes at uvic.ca)