[uf-discuss] class names and uniqueness - hAtom

[Ryan King]

On Oct 25, 2005, at 9:24 PM, Dr. Ernie Prabhakar wrote:

> Hi Tantek,
>
> On Oct 25, 2005, at 8:54 PM, Tantek Çelik wrote:
>
>> And as for making things more generic, I'd say, at first, don't.
>
> I can accept that as a good starting point.  But, that raises the  
> question: what comes after "at first?"
>
> For example, I think Danny's done a great job of solving the  
> specific problem in question.  But now that he's done it, that same  
> pattern:
> * title(s)
> * author
> * content
> * summary/description
> * link(s)
>
>  appears to recur in:
>
> * S5 slides
> * Classified ads
> * Card catalogs
> * Abstracts
>
> It sure seems like there's something ur-pattern behind all these  
> things which is being articulated in slightly different ways.    
> Does it really make sense to create five different microformats?

We'll know after we've created 4 of them. :D

Seriously, I think the best model is:

0. Research
1. Build a µf
2. Get it adopted
3. GOTO 0

In later iterations, you'll likely find that there's refactoring that  
can be done. But until you're there, you won't actually be able to  
make the right design decisions.

> I realize there's a danger here of forcing dissimilar objects into  
> the same mold, and maybe this discussion is premature without more  
> in-depth research. But, given at least the possibility of an ur- 
> pattern, what is the optimal process for discovering it?

-ryan
--
Ryan King
ryan at technorati.com