[Paraview] Re: Possible Endian Issues on Paraview OS X Intel

[David Cole]

CMakeCache.txt should never be moved from one machine to another, even 
with the *same* architecture. Within CMakeCache.txt are several full 
path references to files that exist on the machine where the CMakeCache 
was created/updated.

It's "invalid operating procedure" ever to move a CMake build tree from 
one machine to another. The only possible way that could ever work is if 
every reference to a header file or library or tool or path name in the 
CMakeCache is identical on both machines. (Something that's highly 
unlikely, given the number of settings in ParaView's CMakeCache...)

Hope this helps to clarify things a bit and avoid problems like these in 
the future,
David Cole


Mike Jackson wrote:

>    Well, to cut to the chase.. It isn't such a good idea to use a  
> CmakeCache file from a PPC OS X machine on an Intel OS X machine. And  
> to prove such a theory I started a clean build on the Intel machine  
> by using ccmake to make the changes, then moved the CMakeCache file  
> to the Powerbook G4 and reran ccmake. I the only thing I did change  
> on the Powerbook (PPC) machine was to set the Build Arch to PPC from  
> i386. I ran the build, waited...... and then finally ran Paraview.  
> Interestingly the problem that I had on the Intel MacBook Pro is now  
> showing up on the Powerbook.
>    So I started again with a fresh Paraview 2.4.4 and created build  
> directories on each machine. I ran ccmake on each machine separately  
> and made sure the settings were the same on each machine (Except for  
> Build Arch, which needs to be set correctly on each machine). I  
> changed the Linker flags to those mentioned in the wiki. Changed the  
> OpenGL include to use /usr/X11R6/include but left the OpenGL libs to  
> use the default frameworks in /System/Library/Frameworks. Ran the  
> build on each machine and now each machine can load the "troublesome"  
> data file.
>
> So what we can learn from this is that there are still some compile  
> decisions that are being made on each platform that make the transfer  
> of a CMakeCache file between OS X machines of different architectures  
> impossible.
>
> I have no idea what the problem is. I am just happy to have my  
> paraview back and working again.
>
> --------
> Mike Jackson
> imikejackson <at> gmail <dot> com
>
>
>
> On Jul 8, 2006, at 4:05 PM, Mike Jackson wrote:
>
>> So I finally got Paraview 2.4.4 compiled and running on my MacBook  
>> Pro (thanks Everyone). I attempted to load up a file written with  
>> vtkXMLImageDataWriter and I get an Error that essentially locks up  
>> Paraview.
>>
>> Here are the errors:
>>
>> ErrorMessage
>> # Error or warning: There was a VTK Error in file: /Users/Shared/ 
>> OpenSource/para
>> view-2.4.4/VTK/IO/vtkXMLStructuredDataReader.cxx (313)
>> vtkXMLImageDataReader (0x16e7b940): Error reading extent 0 255 0  255 
>> 0 255 from
>> piece 0
>> ErrorMessage end
>> ErrorMessage
>> # Error or warning: There was a VTK Error in file: /Users/Shared/ 
>> OpenSource/para
>> view-2.4.4/VTK/IO/vtkXMLDataReader.cxx (478)
>> vtkXMLImageDataReader (0x16e7b940): Cannot read point data array  
>> "RGB" from Poi
>> ntData in piece 0.  The data array in the element may be too short.
>> ErrorMessage end
>> ErrorMessage
>> # Error or warning: There was a VTK Error in file: /Users/Shared/ 
>> OpenSource/para
>> view-2.4.4/VTK/IO/vtkXMLStructuredDataReader.cxx (313)
>> vtkXMLImageDataReader (0x16e7b940): Error reading extent 0 255 0  255 
>> 0 255 from
>> piece 0
>> ErrorMessage end
>> ErrorMessage
>> # Error or warning: There was a VTK Error in file: /Users/Shared/ 
>> OpenSource/para
>> view-2.4.4/VTK/IO/vtkXMLDataReader.cxx (478)
>> vtkXMLImageDataReader (0x16e7b940): Cannot read point data array  
>> "RGB" from Poi
>> ntData in piece 0.  The data array in the element may be too short.
>> ErrorMessage end
>>
>> The same file loads and runs fine on a Powerbook G4 running the  same 
>> OS X 10.4.7, but Paraview 2.4.3.
>>
>> I scanned the source files but did not see anything that stuck out  
>> as testing for endianness. These same files also work on a Windows  
>> XP box if that helps.
>>
>> Is the compile process for paraview making assumptions about byte  
>> ordering based on Operating System instead of actual hardware?
>>
>> I _was_ able to load a legacy VTK Data file on the intel OS X  
>> machine just fine.
>>
>> Should I rebuild in Debug instead of Release to get more details in  
>> the error message?
>> Would a Newer version of Paraview work? Is there a known bug?
>>
>> I was using vtk 5.0.0 to write the xml base VTI file.
>>
>>
>> Respectfully?
>> --------
>> Mike Jackson
>> imikejackson <at> gmail <dot> com
>>
>>
>>
>
> _______________________________________________
> ParaView mailing list
> ParaView at paraview.org
> http://www.paraview.org/mailman/listinfo/paraview
>