On Fri, Mar 12, 2004 at 04:06:57PM +0100, Ketil Malde wrote:

"Daan Leijen" writes:

...

Simon Marlow wrote:

...

'insert map key x'

...

So we can do supposedly common things like: any (member s) xs map (lookup m) ks

...

the current design is nicer so I don't think we should get into long discussions about which partial application occurs more often unless we can back it up with data :-)

If you want a data point, I grepped quickly through my sources and found a handful of "map (lookupFM fm) xs"-type constructs.

...

more -- so if many others like it better too, I might be persuaded to adapt the implementation. However, this requires quite a bit of work, so I would like to leave this issue open for discussion for a while.

I've no really strong preference, but would vote for keeping the order defined by the old FiniteMap and Set implementations, if for no other reason than being more compatible.

Ack. the new, consistant, useful order and regular bias are some of my major reasons for prefering DData. If you want to do many things to the same Map, there are plenty of functions for dealing with lists of things to add or whatnot, the partially aplied functions with the map at the end give new, useful functionality orthogonal to the list dealing with functions. hope this makes sense. John -- --------------------------------------------------------------------------- John Meacham - California Institute of Technology, Alum. - john@foo.net ---------------------------------------------------------------------------