[uf-dev] Human and machine readable data format

[Glenn Jones]

Toby A Inkster wrote:

> A problem which hasn't been raised with regards to this proposal is  
> that even though you are proposing a fixed date format, because it  
> *looks* like natural language, authors will assume that it *is*  
> natural language, and simply start including dates in whatever format

> they like. Then you get an NLP "arms race" between publishers and  
> parsers.

> If you don't believe that that will happen, take a look at what  
> happened with RFC 822 dates, which are simply a mess.

Very true the more you make the date look like natural language, the
less it looks like a fixed format. 

I really don't want us to get involved in any form of NLP, it just would
not work. I think it was Mike who said that dates have to be parsed
correctly, no level of error is acceptable. I don't want to travel to an
event on the wrong day because a parser got the date wrong. 
   
I had not come across the whole RFC 822 standard, you are right what a
mess. Dates do seem to be a recurring theme of pain for developers.


Glenn Jones