Re: Pathname#walk for traversing path nodes (patch)

Hi --

On Tue, 23 Aug 2005, daz wrote:

>
> David A. Black
>
>>
>> On Tue, 23 Aug 2005, daz wrote:
>>
>>>
>>> Tanaka Akira wrote:
>>>
>>>> David A. Black writes:
>>>>
>>>>> Maybe "scan".
>>>>
>>>
>>> I've followed the recent protests that a Pathname is not
>>> a String and, here, the name of a String method is
>>> suggested.  An understanding of the operation of either
>>> #scan method would provide no clue to the operation of
>>> the other.
>>> This, IMHO, is what leads to dilution of meaning,
>>> then dilution of understanding.  I object strongly.
>>
>> I don't think any given class can monopolize a method name, unless
>> there's a really close connection.  (And in that case it would be
>> monopolization by default, since no other objects would want the
>> name.)
>>
>
> Perhaps - but I think the name "scan" should be reserved for
> methods which are doing some kind of scanning (i.e. discovery)
> rather that iterating over components which are invariant and
> well understood by the class.

My first choice would actually be split, but that's already in use for
this class.


David

-- 
David A. Black
[e-mail]