[uf-rest] "typed" microformats proposal

[Kevin Marks]

On Feb 13, 2006, at 4:49 PM, Dr. Ernie Prabhakar wrote:

> Excellent comments -- I agree with most of them.  For those who don't 
> know, the easiest way to see them is via history:
>
> http://microformats.org/wiki?title=rest%2Fdatatypes&diff=0&oldid=4945
>
> I'll wait to see if anyone else has a different perspective before 
> integrating Kevin's comments.
>
> The one area I'm most unsure about is "int" -- is there any harm in 
> just calling this "any integer" rather than restricting it to "int32"? 
>  How would the failure/parsing modes differ?

As we aren't specifying signed/unsigned, the interpretation varies 
anyway. I do like Python and Ruby's  'duck typing' approach, so 
specifying types somewhat loosely is OK. Insisting on int32 overflow 
behaviour may be expecting too much.

Is this a case for a SHOULD rather than a MUST ?
MUST support at least 32 bits of precision, and MAY support more ?