[Insight-developers] Re: Failing Tests

[Luis Ibanez]

Bill,

I'm working on that...

We found the warning from the SGI pretty useful.
It happens that we have several virtual function "const" in one of the base
classes of the CostFunction hierarchy and some of the derived classes
have their version of these functions "non-const"

Most compilers just consider that the versions in the derived classes
are some other function (becuse the signature doesn't match).

The SGI compiler is kind enough to ask if we are intending to overload
the "const" function.      :-)

What is happening now at run-time is that the actual instantiatiations
are not being called (because their signature is different)....


    Luis

================================================

Lorensen, William E (CRD) wrote:

>Luis,
>	Getting closer...
>
>There are several failing tests that may be related to your recent changes:
>
>CovariantVectorGeometryTest
>ImageMomentsTest
>LevenbergMarquardtOptimizerTest
>PointGeometryTest
>VectorGeometryTest
>PrintSelf-Algorithms
>PrintSelf-Numerics
>
>There are others that are unrelated:
>
>ConfidenceConnectedImageFilterTest
>ConnectedThresholdImageFilterTest
>IsolatedConnectedmageFilterTest
>NeighborhoodConnectedImageFilterTest
>
>only crash on gcc with the -g flag set. Seem to be a compiler issue. We are looking into this.
>
>Thanks,
>
>Bill
>