Subject: Re: Completeness and Consistency Checking
From: Craig Chambers (chambers@cs.washington.edu)
Date: Mon Apr 17 2000 - 12:05:52 PDT
If you were always going to do CC99, and you weren't going to extend CL95 to
handle parameterized types and methods, then I think you're right. But CC99 and
CL95 use different algorithms to do their work, and it's not clear to me which
is more efficient in practice. Also, CL95 should extend to parameterization,
and I'm less sure how CC99 would.
-- Craig Chambers
Mayur Naik wrote:
>
> Hello
>
> I have a query concerning your algorithm in [CL95] for static typechecking of multi-methods in a GF and your algorithm in [CC99] for generating the dispatcher for a GF.
>
> [CL95] involves checking Completeness and Consistency as part of implementation-side typechecking. But [CC99] can make these checks without additional effort: it can warn that the GF is incomplete or inconsistent if the MSAM computation at a leaf node of the lookup DAG results in the 'message not understood' or 'message ambiguous' error respectively. Isn't it redundant to perform Completeness and Consistency checks in [CL95] assuming that [CC99] is always executed?
>
> Thanks for your time.
>
> -- Mayur
>
> HotBot - Search smarter.
> http://www.hotbot.com
This archive was generated by hypermail 2b25 : Tue Oct 03 2000 - 15:21:29 PDT