Re: DTB Review

> 3.6.2 Predicates on Boolean Values
> 
> I think we need is-true and is-false predicates for boolean in order
> to tie them in to true and false in RIF.

I remember us talking about this at F2F13 and deciding people could just
use   ="true"^^xs:boolean and ="false"^^xs:boolean.

> 3.10.1 Functions on rdf:text
> 
> "he pair 〈s, l〉 in the value space of rdf:text"
> 
> The angle brackets come out as undefined characters in my browser.

Hmmm.  OWL stumbled on this too, deciding it was an IE7 problem.
http://lists.w3.org/Archives/Public/public-owl-wg/2009Apr/0398.html

> 3.11.1 Position Numbering
> 
> "Positions beyond the end of the list are adjusted by the builtins to
> point to the first position after the end of the list. "
> 
> That seems weird.  What's the use case for that?  Negative indexes are
> fine, I don't see they imply indexes beyond the end are OK.

It's not really connected to negative indexes.  The reason to have this
is that it allows you to append without first asking the length of the
list, as long as you know some number greater than the length of the
list:

   insert-before(List(0 1 2 3 4) 1000000000 99) = List(0 1 2 3 4 99)

This is, of course, a hack.  I'd support adding append, and then I'd be
okay with removing this feature.  (The one other place it comes into play
is in sublist; there, too, you can use some big number as a way to get
the 3-argument version to act like the 2-argument version.  Not terribly
useful.)

    -- Sandro

Received on Tuesday, 19 May 2009 12:17:35 UTC