[vtk-developers] Re: [vtkusers] Unconnected filter(s) causing seg faults

[Charl P. Botha]

Mathieu Malaterre wrote:
> Charl P. Botha wrote:
>> N Smethurst wrote:
>>> How should the checks be implemented? I noticed that in the methods 
>>> that do check, some do this
>>>
>>>   if (input == NULL)
>>>     {
>>>     vtkErrorMacro(<<"Input is NULL");
>>>     return;
>>>     }
>>>
>>> whilst others just do a
>>>   if (input == NULL)
>>>     {
>>>     return;
>>>     }

>> Personally I prefer the first form.  At the very least we should keep 
>> the error message consistent so that it can be filtered if the 
>> programmer requires that.
> 
> 
> One day ago, the second form was prefered:
> http://vtk.org/cgi-bin/cvsweb.cgi/VTK/Graphics/vtkMaskPoints.cxx.diff?r1=1.41&r2=1.42 
> 
> 
> I guess sometime it can happen to have an empty input. So there is no 
> reason to produce an error.
> 
> I'll go for the second form, or a vtkWarningMacro instead.

The diff you refer to wasn't checking for GetInput() like the examples 
Nick has found.  I agree though, sometimes having no input is not an 
error, but the user should definitely be notified as this can be the 
source of many program errors.

I do like your suggestion of vtkWarningMacro as a compromise.

-- 
charl p. botha http://cpbotha.net/ http://visualisation.tudelft.nl/