[uf-discuss] changing abbr-design-pattern to title-design-pattern?

[Jeremy Keith]

James Craig wrote:
> Due to opening up the pattern a bit more, there will also need to  
> be a flag to indicate when to use title attribute versus contents.  
> Something like this "useTitle" class:

No, this smells like a really bad idea. That class is now an  
instruction for machines.

One of the fundamental tenets of microformats is that ease of  
publishing should trump ease of parsing ("people first, machines  
second"). A class like "useTitle" would make parsing easier but it's  
a pain in the butt for publishers. I think we can come up with a rule  
for parsers that doesn't require any superfluous classes (even if it  
means that the parsers have to work harder).

The problem:
Right now it is perfectly valid to encode a datetime like this:

<span class="dtstart">2007-05-01</span>

see http://microformats.org/wiki/abbr-design-pattern

Under the proposed title design pattern, this would also be a valid  
way of encoding a datetime:

<span class="dtstart" title="2007-05-01">May 1st 2007</span>

Now a parser must figure out where the machine-readable datetime is.  
In this example, it's pretty straightforward: look in the title  
attribute first; if there is no title attribute, look at the value  
enclose between the tags.

The tricky bit is when this occurs:

<span class="dtstart" title="a day of fun on the streets of  
Berlin">2007-05-01</span>

If a parser looks at the title attribute first, it won't find the  
machine-readable data. This is where something like James's class  
suggestion would clarify the situation. But I believe this to be an  
edge case. I think we'll get more than 80% of cases by simply saying  
"for dtstart, dtend and geo values, look in the title attribute first  
and between the tags second." And to cover the edge cases, I think  
the onus is on the parsers to figure out whether a string looks like  
a machine-readable piece of data. For datetime patterns, that's  
simply a case of using a regular expression already documented on the  
wiki:
http://microformats.org/wiki/datetime-design-pattern#Code

In short, I don't think any kind of flag is necessary to indicate  
where the machine-readable data is being stored. I think that any  
such flag would unnecessary complexity for humans. The patterns are  
enough.

Bye,

Jeremy

-- 
Jeremy Keith

a d a c t i o

http://adactio.com/