[Haskell] Proposal for a Standard of Abstract Collections (with Reference Implementation)

Christian Maeder maeder at tzi.de
Fri Mar 19 15:20:47 EST 2004

Robert Will wrote:
>     http://www.stud.tu-ilmenau.de/~robertw/dessy/fun/

I've already preferred DData over Edison 
also because DData is Haskell 98 compliant (without glasgow extensions), 
thus "portable" (in terms of the haddock documentation of the library).

Viewing maps as collections of pairs indeed suggests to change some 
names (like member and filter) and some types, i.e. using a pair as 
input instead of curried arguments. (i.e. "member" could become 
"isDefined" when maps are viewed as finite functions.)

I would not mind, if a collection class/interface is provided on top of 
the given DData modules/types (that one may or may not use) using 
glasgow extensions.

And I would also not mind if there are other bag, set and map 
implementations (i.e. based on common tree implementation), so that a 
simple change of the import gives you a different performance behaviour 
(i.e. modules {Bag, Set, Map}{ByOrderedSeq,ByBalancedTree}.hs)

However the (non-portable) instances for a big collection class should 
be in separate modules.


More information about the Libraries mailing list