[PRL] Accumulator for container capacity

[Matthias Felleisen]

On Sep 30, 2005, at 11:58 AM, Felix S Klock II wrote:

> Karl-
>
> On Sep 30, 2005, at 11:41 AM, Karl Lieberherr wrote:
>
>> Adaptive Programming (AP) takes the principle of Structural Recursion 
>> (SR)
>> to a new level. In SR you organize your program following the 
>> structure of
>> your data. SR is an old principle and has recently been promoted
>> systematically in HtDP. AP applies SR to focusing on the difference 
>> between
>> the structural recursion template and the desired program. The 
>> information
>> in the structural recursion template (that is systematically defined 
>> by the
>> data type definition), is not repeated in the adaptive program.
>>
>
> What does AP give you that Polytypic programming does not?
>
> Polytypic programming does not repeat the information from the 
> template.
>
> In fact, in a sense polytypic programs are parameterized over the 
> template.  (More precisely, polytypic programs are parameterized over 
> the client's choice of type-constructor; the template is then derived 
> from the structure of the type constructor to yield a particular 
> instantiation of a polytypic program.)

Amen. If you look in HtDP IV, you will see a discussion of how to 
abstract the SR template. I don't ever teach this. But it comes from 
work in EoPL courses where I pushed students to develop a fully lazy 
abstraction because it's the only fully general way to get this right.

-- Matthias




> My memory is poor; I can imagine potential additions provided by AP 
> (e.g. the ability to inspect the name given to a class, which 
> polytypic programs in their purely structural form do not allow), but 
> I am not familiar enough with AP technology to be certain of what it 
> provides.  I just want to understand the reasons that one would select 
> AP over PolyP.