Glas List: Re: [glas] value_type

From: Peter Gottschling (pgottsch_at_[hidden])
Date: 2005-10-04 21:06:56

On 04.10.2005, at 02:26, Karl Meerbergen wrote:

> Matthias Troyer wrote:
>
>> On Oct 3, 2005, at 11:14 PM, Toon Knapen wrote:
>>
>>
>>> Basically the difference is:
>>> 1) or we specify a set of requirements the value_type's need to
>>> fullfill
>>> so that you can use any algorithm in the glas-library
>>> 2) or we specify the set of requiments on the value_type's per
>>> algorithm
>>>
>>>
>>
>> Since concepts should specify the _minimal_ requirements on each
>> algorithm, clearly 2) is the way to go.
>>
>> Matthias
>>
>>
>
> I too prefer the 2nd option.
>
> Karl
>
Me too.  For instance, 2x2 matrices can be useful as value_type for 
vectors and matrices but they are not invertible in general. Requesting 
division would therefor forbid them as value_type.

In general, applicability can always be extended in two directions:
1) minimizing the requirements of an algorithm to maximize the set of 
applicable data types
2) maximizing the requirements that a given data type fulfills to 
maximize the set of applicable algorithms.

Peter
> _______________________________________________
> glas mailing list
> glas_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/glas
>
------------
Peter Gottschling
Research Associate
Open Systems Laboratory
Indiana University
301i Lindley Hall
Bloomington, IN 47405
Tel.: [phone]   Fax: [phone]
http://www.osl.iu.edu/~pgottsch