Fwd: linear and non-linear terms

[Carl Pollard]

Hi again, Jonathan,

I found the 8-page abstract at the EDILOG 2002 websitem but no appendix.
Is there a longer version?

At first glance your ML-based treatment of records appears similar to
the treatment in my draft called "Cleaning out the HPSG garage"
(available at the website of the 2001 Trodheim summer school, and a
better source for this then the (misleadingly titled) invited talk
from HPSG 2001). I follow Barr and Wells in treating (static) record
types as just indexed (cartesian) products, so that records themselves
are just terms (more specifically, indexed tuples). Is this also the
case in Betarte/Betarte & Tasistro's ML-based setup? Does the
construction actually depend on the additional power of dependent types?

Carl
----------------------------------------------------------------------
>
(a preliminary
version is in an appendix to our paper from this year's Edilog,
`Clarification Ellipsis in Dependent Type Theory'.).
>>