[Insight-developers] Wrong type in ConnectedComponentImageFilter

[Bill Lorensen]

I think this shows the value of warnings. It also shows it is hard to
see the new ones with so many lingering warnings.


On Wed, Jan 5, 2011 at 9:24 AM, Luis Ibanez <luis.ibanez at kitware.com> wrote:
> Hi Bill,
>
> Yeap, you are right.
>
> The variable (*I) is an OffsetValueType (which is a signed type),
> so the result of the addition should be stored in an OffsetValueType
> as well.
>
>
> Here is the patch that I come up with:
> http://review.source.kitware.com/#change,671
>
>
>
>     Luis
>
>
> ---------------
> On Wed, Jan 5, 2011 at 8:59 AM, Bill Lorensen <bill.lorensen at gmail.com> wrote:
>> Dave and Luis,
>>
>> I looked at this new warning today:
>> In file included from
>> /Users/builder/kitware/ITK/Testing/Code/BasicFilters/itkRelabelComponentImageFilterTest.cxx:22:
>> In file included from
>> /Users/builder/kitware/ITK/Code/BasicFilters/itkConnectedComponentImageFilter.h:263:
>> /Users/builder/kitware/ITK/Code/BasicFilters/itkConnectedComponentImageFilter.txx:273:23:
>> warning: comparison of unsigned expression >= 0 is always true
>> [-Wtautological-compare]
>>        if ( NeighIdx >= 0 && NeighIdx < linecount &&
>> !m_LineMap[NeighIdx].empty() )
>>             ~~~~~~~~ ^  ~
>>
>> It shows up on a couple of platforms.
>>
>> I looked at the code.
>>
>> the types of ThisIdx and NeighIdx used to be long. Now they are
>> SizeValueType (which is unsigned long). I believe they should be
>> IndexValueType (which is long).
>>
>> Bill
>>
>