Abstract Collections vs. FunDeps

[Dylan Thurston]

On Fri, Apr 02, 2004 at 11:21:05AM +0200, Robert Will wrote:
> 1. Left-Single FunDeps are Syntactic Sugar for Constructor Classes

I think you should try writing programs a little before making such
broad claims...  I was convinced when I read this, but then someone in
haskell-cafe
(http://www.haskell.org//pipermail/haskell-cafe/2004-April/006014.html)
tried actually doing it, and ran into some trouble; see the second
half of the message.  I don't know how to make the "Constructor Class"
version of the program work without '-fallow-undecidable-instances'.
Do you?

Also:

> 3. Parametric Type Classes are Normalised FunDeps
> 
> ...
> Classes with only one FunDep are necessarily in normal form.  I don't
> want to consider transforming class declarations to normal form (no
> idea what that would mean...), just this: if we use a parametric type
> class of the form "class a \elem C b where" ('a' and 'b' sets of
> variables) then this corresponds just to a FunDep in normal form.
> Non-normal-form FunDeps can't be expressed with Parametric Type
> Classes.

I have no idea what you mean here.  "class a \elem C b where" doesn't
look like Haskell to me.  What is a parametric type class?  Could you
expand a little?

(And maybe we should move this discussion to haskell-cafe, since it's
very off-topic for the libraries list and other people might be
interested.  Maybe you could repost the parent there as an
introduction?)

Peace,
	Dylan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://www.haskell.org//pipermail/libraries/attachments/20040403/885eb342/attachment-0001.bin