[Bug 24569] Least common types and lattices

https://www.w3.org/Bugs/Public/show_bug.cgi?id=24569

C. M. Sperberg-McQueen <cmsmcq@blackmesatech.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED

--- Comment #6 from C. M. Sperberg-McQueen <cmsmcq@blackmesatech.com> ---
We discussed this at our face-to-face meeting in Prague.

We concluded that whether we have a type lattice depends on exactly what we
wish to regard as a "type" for purposes of this analysis.

One solution would be to specify that the points in the lattice are groups of
types equivalent in the sense that subtype-itemtype(A,B) and
subtype-itemtype(B,A).

Another would be to ignore whether itemtypes are or are not a lattice, and
simply define an ad hoc type inferencing scheme for use here.  We are
interested primarily in two bits of information:  (a) can there be children?
(b) can members of the set be numeric?  It might be possible to capture this
with a very simple type hierarchy, or a simple 2x2 diagram.  

A 2x2 matrix could give us better information.  For example it would allow us
to infer that the static type of (element(), xs:string) is non-numeric, whereas
a type hierarchy or lattice that looks anything like what's described in our
specs must end up with a least common supertype of item() or something similar.

We concluded that this issue needs further study.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.

Received on Thursday, 13 February 2014 11:21:34 UTC