On 11-Dec-1997, Paul Hudak <[EMAIL PROTECTED]> wrote:
> Having participated in many previous Haskell design efforts, I must say
> that John's WWW-based system is MUCH better than straight email. With
> email you have 16 different threads that are really hard to keep track
> of; the tree-based approach keeps things better organized. A newsgroup
> isn't as good either, even if threaded.
Fair enough.
With regard to the delay, one small and relatively simple improvement
would be to provide a way to download the entire directory tree
(tarred & gzipped) for local browsing. Just a suggestion...
> In any case, the committee certainly did not "deliberately discourage
> the participation of those not on the committee"; indeed I'd say the
> opposite strategy was taken. I think it's very unusual for a committee
> to open its dialogue to the world.
The committee are to be commended for that.
Sorry if my message was a bit grumpy.
ObHaskell:
With regard to type classes, one feature that would be really
nifty would be dynamic (run-time checked) type class membership
tests / type conversion. This would let you do things like
using `show x' if x is showable, and using some reasonable
default if it is not. In order for this to be efficiently
implementable, I think it is necessary to restrict instance
declarations so that for any given type and class there is at
most one instance declaration that could apply.
This feature would be another rationale for prohibiting
overlapping instance declarations... partly because it requires
that prohibition, in order to work efficiently, but also because
it would actually let you do most of the useful things you can
do with overlapping instance declarations, in an arguably
clearer way.
--
Fergus Henderson <[EMAIL PROTECTED]> | "I have always known that the pursuit
WWW: <http://www.cs.mu.oz.au/~fjh> | of excellence is a lethal habit"
PGP: finger [EMAIL PROTECTED] | -- the last words of T. S. Garp.
