Re: [boost] Terminology:Type Unification? (was Re: RFC Generic data structure traversal / manipulation library

From: Eric Niebler (eric_at_[hidden])
Date: 2008-01-16 11:03:19

Larry Evans wrote:
> On 01/16/2008 01:39 AM, dan marsden wrote:
>>
>> The operation is certainly not a fold, so I think it would be misleading
>> to associate the example with fold like terminology.
> 
> Agreed.  A fold operates with a binary op on a sequence of values where
> the result type of op is same type as operands.  The reason I was
> thinking fold might be applicable is that the example just chose
> the leaves in the tree of a certain type and added them.  Well, add
> is certainly a valid fold op; so, the difference was some filtering
> (to the given type) and then flattening the tree to conform to the
> structure needed by fold.  However, now I can see how a different
> name than fold is needed.

In Fusion, fold() operates on heterogeneous lists and does type 
unification using statically polymorphic function objects. Ditto for 
Proto's fold transform. So fold() isn't a bad name, IMHO.

-- 
Eric Niebler
Boost Consulting
www.boost-consulting.com