Re: draft response for LC comment 51 RM1 and 62 JM1

On 3 Mar 2009, at 19:37, Evan Wallace wrote:

> All,
[snip]
> However, named user defined datatypes are a critical capability for  
> OWL DL that I
> had expected to see in OWL 2.  I think we should support them, if  
> we can.   We rather
> rushed past this discussion at the face-to-face on Monday afternoon/ 
> evening as
> people were understandably tired of datatype discussions at that  
> point.  Let's please
> discuss this a bit before rejecting such a basic feature.

This was an accident. In OWL 1, the idea was that you'd build your  
datatypes externally in an XML Schema file then use a "standard" URI  
to access that definition. This never got settled (although e.g.,  
Pellet supported one variant). When we moved datatype definitions  
inside OWL we sorta kinda forgot to bring the naming with it.

While it's important to have anonymous type expressions, named ones  
are pretty important too. Not just for terseness, but for intention- 
revealingness.

So, I think it's a bug, not an intention (and, really, my fault).

Technically, as Evan pointed out from Boris's and Ian's paper, the  
main thing is to avoid cycles, so that the names really are macros.

Cheers,
Bijan.

Received on Tuesday, 3 March 2009 20:16:37 UTC