[vtk-developers] recent HDF5-related regression with installed file names...

[Dan Lipsa]

Hi Sean,
I pushed a fix for these issues to VTK master.

Thanks again for your report.
Dan

On Mon, Dec 1, 2014 at 4:07 PM, Dan Lipsa <dan.lipsa at kitware.com> wrote:

> Hi Sean,
> Thanks for your report. I am looking at this right now.
>
> Dan
>
>
> On Mon, Dec 1, 2014 at 3:53 PM, Sean McBride <sean at rogue-research.com>
> wrote:
>
>> Robert,
>>
>> Thanks for your reply.  I guess you mean this:
>>
>> ---------------
>> commit c76a183cb4bdfd255c60b006d84f82a259cd6de6
>> Merge: f0480df db8d3e2
>> Author: Dan Lipsa <dan.lipsa at kitware.com>
>> Date:   Sun Nov 23 16:49:15 2014 -0500
>>
>>     Merge topic 'update-hdf5' into master
>> ---------------
>>
>> I'm trying the commit before it to verify.
>>
>> Also, I retract my statement that in Release I get what I expect, even
>> there the name is wrong, it's missing the "-6.2" at the end (that every
>> other library file has).
>>
>> If I rename them manually, I hit another problem, a link error:
>>
>> duplicate symbol _H5MM_malloc
>> duplicate symbol _H5MM_calloc
>>
>> My app links to another HDF5, and so I presume those two functions are
>> not being mangled in vtk_libhdf5_mangle.h.
>>
>> Sean
>>
>>
>>
>> On Mon, 1 Dec 2014 14:39:01 -0500, Robert Maynard said:
>>
>> >I expect that this was caused by the upgrade to hdf5 1.8. This
>> >shouldn't be super hard to fix, as the library naming logic is all in
>> >ThirdParty/hdf5/vtkhdf5/config/cmake/HDFLibMacros.cmake
>> >
>> >On Mon, Dec 1, 2014 at 2:29 PM, Sean McBride <sean at rogue-research.com>
>> wrote:
>> >> Hi all,
>> >>
>> >> I've just now rebuilt & installed VTK master, and it seems there is a
>> >regression with HDF5-related install files.
>> >>
>> >> If building Release, I get what I expect, which is two files named:
>> >>
>> >> libvtkhdf5_hl.a
>> >> libvtkhdf5.a
>> >>
>> >> in the /lib install folder.  But building debug, I get instead:
>> >>
>> >> libvtkhdf5_debug.a
>> >> libvtkhdf5_hl_debug.a
>> >>
>> >> This was not the case previously, and seems incorrect.
>> >>
>> >> Anyone know about this breakage, and can hopefully fix it? :)
>> >>
>> >> Thanks,
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtk-developers/attachments/20141202/29a4753e/attachment-0001.html>