[uf-discuss] RE: Microformats and RDFa not as far apart as previouslythought

[Michael MD]

> Guillaume Lebleu wrote:
>
>> <span class="dstart" lang="en-us">October 5, 2004</span>
>
>
> Cognition already supports this as a last ditch attempt at parsing 
>  dates - but I wouldn't recommend it get adopted widely. It's too 
> unreliable; too much work to deal with internationalisation; too much 
> work full-stop in languages that don't provide a handy library that  takes 
> care of most of the work.
>

even in those that do there are too many ambiguities...

Even in rare cases where you might know what country something is in you 
still can't be sure that date is written in a way that is common in that 
country.

eg .. here in Australia  dd/mm/yyyy is commonly used ... but I also see a 
significant minority of sites in Australia using US-style mm/dd/yyyy because 
it is the default setting in their CMS!

How would a parser be able to tell which part is the day and which is the 
month? ....

... getting it wrong would be worse than not getting it at all!

of course the ideal long-term solution would be to teach people to write 
dates more clearly (something like 12th June 2008 or July 8, 2008  is clear 
enough
but the masses need to learn not to use those awful "slash" formats and 
anything without the year or with 2 digit years)

until the use of ambiguous formats can be WIPED OUT we WILL need a version 
for machines!

The former is of course unlikely to happen any time soon ... so machine 
dates ARE needed .. without them you can say goodbye to hCalendar or 
anything else that relies on dates!