[Paraview] Data set Size limit in ParaView/HDF5

[Michael Jackson]

Not recent enough. I think that was the version where I found the bug for Darwin systems which limited it to 4GB data sets. That version is also about 2-3 years old. HDF5 1.8.13 is the latest. 1.8.14 is coming out in few weeks.

Mike Jackson

PS: We are just trying to get everyone to build ParaView on their systems. The issue with the SuperBuild is that you can not seem to set an "external" hdf5 build. It will always use the bundled version.

On Nov 5, 2014, at 3:13 PM, Utkarsh Ayachit <utkarsh.ayachit at kitware.com> wrote:

> Yes, to create "redistributable" binaries, Superbuild would be the way
> to go. The ParaVIew superbuild already uses hdf5 1.8.9 Do you know if
> that recent enough?
> 
> Utkarsh
> 
> On Wed, Nov 5, 2014 at 2:03 PM, Michael Jackson
> <mike.jackson at bluequartz.net> wrote:
>> I was trying to directly update the source codes then compile my own. That didn't work. I was able to build ParaView 4.2.0 against my installed HDF5 (1.8.13) and have it create the package in the /Applications BUT not all the libraries got "fixed up". The Qt libraries and my HDF5 libraries did not get their "install_name" fixed.
>> 
>> Do I need to build from the "SuperBuild" to actually create a new package I can give to my employees?
>> 
>> Thanks
>> Mike Jackson
>> 
>> On Nov 5, 2014, at 12:13 PM, Utkarsh Ayachit <utkarsh.ayachit at kitware.com> wrote:
>> 
>>> Mike,
>>> 
>>> Are you talking about updating the HDF5 in ParaView source itself or
>>> in the binaries we distribute?
>>> 
>>> Utkarsh
>>> 
>>> On Wed, Nov 5, 2014 at 12:06 PM, Michael Jackson
>>> <mike.jackson at bluequartz.net> wrote:
>>>> Thanks for getting the bug report in. I was poking around the ParaView source and it does not look straight forward at all to substitute in a prebuilt HDF5 library (system hdf5). I can not even find the CMake Variables that might let me do this. I tried so simply substitute the latest stable HDF5 into the source tree but as I figured CMake threw lots of errors. If you give me a gentle nudge I might be able to help a bit. We are really being held back by this with our current data sets.
>>>> 
>>>> Thanks
>>>> Mike Jackson
>>>> 
>>>> On Nov 5, 2014, at 11:38 AM, Utkarsh Ayachit <utkarsh.ayachit at kitware.com> wrote:
>>>> 
>>>>> Sounds good. I'll work on updating the HDF5 version for next release.
>>>>> 
>>>>> http://www.paraview.org/Bug/view.php?id=15106
>>>>> 
>>>>> Utkarsh
>>>>> 
>>>>> On Wed, Nov 5, 2014 at 11:30 AM, Michael Jackson
>>>>> <mike.jackson at bluequartz.net> wrote:
>>>>>> The current version of ParaView (4.2.1) uses a fairly old version of HDF5 (API version 1.8.9). This version has a nasty bug on OS X that will not allow HDF5 to read a data set that is larger than 4GB (2^32). This is due to a bug in the configuration process that incorrectly identifies a "size_t" as being unsigned 32 bit integer on Darwin instead of a 64 bit unsigned integer. What is the possibility of getting a newer HDF5 included in ParaView for the next release?
>>>>>> 
>>>>>> BTW. a new release of HDF5 is coming in a few weeks. This would make a great opportunity to get this fixed with the latest version of HDF5.
>>>>>> 
>>>>>> Thanks
>>>>>> Mike Jackson
>>>>>> _______________________________________________
>>>>>> Powered by www.kitware.com
>>>>>> 
>>>>>> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>>>>>> 
>>>>>> Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
>>>>>> 
>>>>>> Follow this link to subscribe/unsubscribe:
>>>>>> http://public.kitware.com/mailman/listinfo/paraview
>>>> 
>>