[microformats-discuss] International date formats

[Ian Hickson]

On Tue, 26 Jul 2005, Tantek Çelik wrote:
> 
> Parsing a "single normal space character ASCII 32" (rather than just 
> "white space") should be equivalently easy to parsing a "T", and thus 
> since they are equivalent from a computer point of view, you could make 
> the argument that the W3C Note should be extended to permit either since 
> the space is more human friendly.

Parsing a "T" is a LOT easier than parsing U+0020. Not because parsing 
U+0020 itself is hard, but because when authors think U+0020 is allowed, 
they think U+000A, U+0009, U+000D, U+000C are allowed as well, not to 
mention U+00A0 and a stream of other whitespace characters. And then they 
think that any number of them are acceptable, in any order. 
Implementations then end up all slightly disagreeing about what is 
whitespace and what isn't and pages end up being interpreted differently.

Whitespace should be avoided like the plague in attributes where 
interoperable results are expected from computers parsing the data.

Human consumption is not an issue here anyway. With a T or with 
whitespace, we should not be expecting ISO8601 dates to be the final 
presentation form.

   .date[title] { tooltip: datetime(attr(title)); }

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'