Martin Holmes wrote:
> So, where I used to use something like this:
> <date notBefore="1598" notAfter="1615" precision="low"></date>
> meaning "we're just taking a guess really, but our best guess would be 
> that it's between 1598 and 1615", how would I express that with the 
> current attributes?

Isn't that certainty rather than precision? I mean are you saying it is 
notBefore 1598 and notAfter 1615, but you're not actually certain about 
ascribing it to that period?  Or are you saying that the precision with 
which you are saying that is not really on the granularity of a year 
(rather than month/day/hour/minute/second)?

> You're of course welcome to say that such a formulation is so vague as 
> to be ridiculous and not worth encoding, but it's the sort of thing I 
> often get from content experts, and I think it's better than nothing; 
> and it's certainly different from
> <date notBefore="1598" notAfter="1615"></date>
> i.e. without the @precision attribute. Would @cert do the job here? I 
> guess this is, really, more of a matter of certainty than it is of 
> precision, strictly speaking.

I think the point is if I say:

<date notBefore="1598" notAfter="1615">foo</date>

then my precision is based in a granularity of years, if I say:

<date notBefore="1598-04" notAfter="1615-08">foo</date>

then obviously I have a precision of a granularity of at least months. (and 
so on).

I think in your case it really is certainty about ascribing it to that date 
range at all that you are expressing, not precision with which the date 
range is expressed.  This doesn't mean there isn't a place for recording 
precision, but I think the argument is claiming that W3C date formats have 
a basic level of precision measurement built into their format.  (However, 
I would be open to an argument that is is in some cases a false precision 
and that people might like to record precision explicitly. Or more 
accurately perhaps, how the precision of that one element differs from the 
standard consistently imposed across the whole document.)

My two pence,


> Cheers,
> Martin
> Lou's Laptop wrote:
>> I think the point is that the W3C or ISO datatypes come with a defined
>> precision which is explicit eg in the number  of digits you supply, and
>> that there is no need therefore to add an extra attribute.
>> Note that (as James, I think, pointed out first) "precision" is not the
>> same as "reliability".
>>  Martin Holmes wrote:
>>> Hi there,
>>> Garces, Juan wrote:
>>>> Please forgive should I not see the obvious, but how do 
>>>> @notBefore/@notAfter and @from/@to express precision or lack thereof?
>>> Having looked at the references and the discussion, I'm also puzzled by
>>> this. It seems to me that the job done by @precision is no longer done
>>> by any other attribute or combination of attributes. @cert perhaps comes
>>> close:
>>> cert  (certainty) signifies the degree of certainty associated with the
>>> intervention or interpretation.
>>> but I'm not sure "intervention or interpretation" extends to assigning a
>>> date, does it?
>>> Cheers,
>>> Martin

Dr James Cummings, Oxford Text Archive, University of Oxford
James dot Cummings at oucs dot ox dot ac dot uk