[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: more srfi-12 rationale?



To my:
> * Why do we need a new condition type, disjoint from other Scheme values,
> with 5 new operations, some of which are quite non-trivial?  Why can't
> we just use standard lists (perhaps association lists)?  It seems rather
> non-Schemey (non-orthoginal) to me, to add a new data-stype, with new
> operations, that is *similar* to existing data types, but which is
> designed for one very specific applications (exceptions and handlers).
Shriram Krishnamurthi <shriram@xxxxxxxxxxx> responded:
> All of this is also true of continuations.

I know I'm out of the official "Scheme experts" loop.  Still, as the
author of a frequently-praised Scheme implementation, I thought I knew
something about Scheme.  Thus I am rather surprised that no one else
agrees with me that the whole "conditions" concept requires some
justification.  I thought one goal for Scheme is maximal expressiveness
while minimizing primitive concepts.  Instead, the only response I get
is something that seems completely beside the point.  Continuations
give you a lot of expressive power using a mechanism that is very
difficult for users to build using other constructs.  I don't see how
that is true for conditions, which from what I can tell give *no*
expressive power or even convenience over using plain association or
property lists.  Clearly, there is some background I'm missing.  I am
aware of the Common Lisp condition system, but have never used it or
really studied it.  Has the collective wisdom of the "Lisp community"
decided that CL conditions is "good"?
-- 
	--Per Bothner
bothner@xxxxxxxxxxx  per@xxxxxxxxxxx   http://www.bothner.com/~per/