[Insight-developers] itkImageConstIteratorWithIndex.h

[Mark Foskey]

But still, people have gone to some effort to support reverse 
iteration, in two inconsistent ways.  We need to establish a consistent 
semantics.

Luis Ibanez wrote:
> 
> Hi Mark,
> 
> We considered to support reverse iteration in the early days
> of the toolkit. However, as algorithms were implemented, it
> simply happened that you rarely need reverse iteration for
> performing image processing.
> 
> The only close case was the DanielssonDistance map filter, and
> a special iterator was created for taking care of this case.
> (this is the Reflective Iterator).
> 
> Apart from that, we probably don't have any other instances of
> reverse iteration in the toolkit.
> 
> 
>   Luis
> 
> 
> --------------------------------------
> Mark Foskey wrote:
> 
>> Ah, my ignorance of the details of ITK iterators becomes more apparent.
>>
>> Most iterator types don't currently have Reverse versions.  It seems 
>> to me that they all should, except for those where a strong case can 
>> be made that one isn't necessary.  But doing that would increast the 
>> number of iterator classes greatly.
>>
>> On the other hand, the design should be consistent.  If you buy that, 
>> and you still don't want to write a bunch more reverse iterators, and 
>> you also want graceful reverse iteration, then they should all work 
>> something like itkImageConstIteratorWithIndex, perhaps with a 
>> PixelsRemaining() method.
>>
> 


-- 
Mark Foskey    (919) 843-5436  Computer-Aided Diagnosis and Display Lab
mark_foskey@unc.edu            Department of Radiology, CB 7515, UNC
http://www.cs.unc.edu/~foskey  Chapel Hill, NC  27599-7515