[vtkusers] Weird link error - possibly 4.7.1??

[Andrew Parker]

This is the output of the following code inside common/cmakelists.txt

IF (HAVE_FLOAT_H)
  CHECK_SYMBOL_EXISTS(isnan "float.h" VTK_HAS_ISNAN)
  IF (NOT VTK_HAS_ISNAN)
    CHECK_SYMBOL_EXISTS(_isnan "float.h" VTK_HAS__ISNAN)
  ENDIF (NOT VTK_HAS_ISNAN)
  CHECK_SYMBOL_EXISTS(isinf "float.h" VTK_HAS_ISINF)
  IF (NOT VTK_HAS_ISINF)
    CHECK_FUNCTION_EXISTS(_isinf "float.h" VTK_HAS__ISINF)
  ENDIF (NOT VTK_HAS_ISINF)
ENDIF (HAVE_FLOAT_H)

message(STATUS "have_float_h " ${HAVE_FLOAT_H})
message(STATUS "VTK_HAS_ISNAN " ${VTK_HAS_ISNAN})
message(STATUS "VTK_HAS__ISNAN " ${VTK_HAS__ISNAN})
message(STATUS "VTK_HAS_ISINF " ${VTK_HAS_ISINF})
message(STATUS "VTK_HAS__ISINF " ${VTK_HAS__ISINF})

#### gives =>

-- have_float_h 1
-- VTK_HAS_ISNAN
-- VTK_HAS__ISNAN
-- VTK_HAS_ISINF
-- VTK_HAS__ISINF

the ifdef is unset, and hence the code inside vtkMath is not being
compiled, any thoughts??

Andy

On 25 June 2012 09:40, Andrew Parker <andy.john.parker at googlemail.com>wrote:

> IF (NOT HAVE_FLOAT_H)
>   CHECK_INCLUDE_FILE("float.h" HAVE_FLOAT_H)
> ENDIF(NOT HAVE_FLOAT_H)
>
> IF (HAVE_FLOAT_H)
>   CHECK_SYMBOL_EXISTS(isnan "float.h" VTK_HAS_ISNAN)
>   IF (NOT VTK_HAS_ISNAN)
>     CHECK_SYMBOL_EXISTS(_isnan "float.h" VTK_HAS__ISNAN)
>   ENDIF (NOT VTK_HAS_ISNAN)
>   CHECK_SYMBOL_EXISTS(isinf "float.h" VTK_HAS_ISINF)
>   IF (NOT VTK_HAS_ISINF)
>     CHECK_FUNCTION_EXISTS(_isinf "float.h" VTK_HAS__ISINF)
>   ENDIF (NOT VTK_HAS_ISINF)
> ENDIF (HAVE_FLOAT_H)
>
> It must be this section in the code and the use of 4.7.1.  The only link
> error I now have is this:
>
> vtk-5.10.0/bin/libvtkCharts.so.5.10.0: undefined reference to
> `vtkMath::IsNan(double)'
>
> That has nothing to do with info vis, and if you look at the source it's
> clear that the header is unguarded and the method exists, but that the
> implementation is guarded by the above, which clearly must be active since
> the symbol is missing.  Any suggestions why the above is incorrect or not
> working for 4.7.1??
>
> Cheers,
> Andy
>
>
>
>
> On 22 June 2012 16:16, Andrew Parker <andy.john.parker at googlemail.com>wrote:
>
>> Surely the error message screams that there is just a problem with a
>> missing compile of vtkMath???  c++ file not compiled or symbols missing due
>> to missing library, why does this have to do with infovis?
>>
>> Andy
>>
>>
>> On 22 June 2012 16:14, Andrew Parker <andy.john.parker at googlemail.com>wrote:
>>
>>> Well that's all the the case, just double checked, and it still
>>> fails....II take it all I have to do is have vtk_use_infovis set to on for
>>> it to be enabled?
>>>
>>> Anymore for anymore?
>>>
>>>  Andy
>>>
>>>
>>> On 22 June 2012 16:11, David Doria <daviddoria at gmail.com> wrote:
>>>
>>>> On Fri, Jun 22, 2012 at 11:09 AM, Andrew Parker
>>>> <andy.john.parker at googlemail.com> wrote:
>>>> > I already have  VTK_USE_INFOVIS  ON  set, and therefore it's built
>>>> like
>>>> > that.  What do I do now, is there something else I should be linking
>>>> or
>>>> > including if this is the case?
>>>> >
>>>> > Andy
>>>>
>>>> With VTK5.8+, you should definitely just link against:
>>>>
>>>>  TARGET_LINK_LIBRARIES(EXE ${VTK_LIBRARIES})
>>>>
>>>> and not specify the libraries explicitly. If you are doing that, and
>>>> VTK was built with Infovis enabled as you said, I'm not sure what else
>>>> the problem could be.
>>>>
>>>> David
>>>>
>>>
>>>
>>>
>>> --
>>>
>>> __________________________________
>>>
>>>    Dr Andrew Parker
>>>
>>>    Em at il:  andrew.parker at cantab.net
>>>
>>>
>>
>>
>> --
>>
>> __________________________________
>>
>>    Dr Andrew Parker
>>
>>    Em at il:  andrew.parker at cantab.net
>>
>>
>
>
> --
>
> __________________________________
>
>    Dr Andrew Parker
>
>    Em at il:  andrew.parker at cantab.net
>
>


-- 

__________________________________

   Dr Andrew Parker

   Em at il:  andrew.parker at cantab.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20120625/1dc104ae/attachment.htm>