Thanks. I have fixed the issues which came from the fact we never built it with the double type and it was not working as expected. The patch is here <https://github.com/SimonRit/RTK/commit/78d7b39bf30c7567f18055137f18debf0383387a> and the python packages are available here <https://dev.azure.com/simonrit/RTK/_build/results?buildId=504&view=artifacts> .
On Wed, May 6, 2020 at 10:19 AM Jerome Lesaint <lesaint.jer...@gmail.com> wrote: > Ok, thanks! > > Just to let you know : I only added a rtkADMMTV[...].wrap file and built. > It recompiles ALL the wrappings but ONLY them. > As you suspected, the guy is a tricky one. It failed to compile with the > attached log. > > Should I log anything to put this in the development pipeline ? > Thanks again for your help, > Jérome > > > > Le mar. 5 mai 2020 à 21:51, Simon Rit <simon....@creatis.insa-lyon.fr> a > écrit : > >> In my experience, it does not recompile everything but it does recompile >> a significant part but that's very variable depending on how much you >> change. If you want a better answer, I think you can ask what to expect to >> ITK guys on discourse <http://discourse.itk.org/>. >> >> On Tue, May 5, 2020 at 3:48 PM Jerome Lesaint <lesaint.jer...@gmail.com> >> wrote: >> >>> Thanks Simon for the quick response. >>> One question before I give it a try : when re-building after creating >>> the .wrap file, does it only compiles the desired file or does it rebuild >>> the whole ITK again ? Sorry in advance for this probably stupid question, >>> but quite a different game... >>> In other words, can I try on one file or should I first check all the >>> missing filters that I need to recompile only once. >>> >>> Thanks, >>> Jérome >>> >>> Le mar. 5 mai 2020 à 15:28, Simon Rit <simon....@creatis.insa-lyon.fr> >>> a écrit : >>> >>>> Hi Jérôme, >>>> Not everything is wrapped in Python yet. To know if a filter should be >>>> wrapped, you can check if there is a corresponding file in the wrapping >>>> <https://github.com/SimonRit/RTK/tree/master/wrapping> subdirectory. >>>> You can try to create one >>>> rtkADMMTotalVariationConeBeamReconstructionFilter.wrap file by copying >>>> rtkADMMWaveletsConeBeamReconstructionFilter.wrap >>>> <https://github.com/SimonRit/RTK/blob/master/wrapping/rtkADMMWaveletsConeBeamReconstructionFilter.wrap> >>>> and to just change the filter name. This wrapping could be tricky so let us >>>> know if does not work directly. It it does, please create a pull request! >>>> Cheers, >>>> Simon >>>> >>>> >>>> On Tue, May 5, 2020 at 3:00 PM Jerome Lesaint <lesaint.jer...@gmail.com> >>>> wrote: >>>> >>>>> Dear all, >>>>> >>>>> I have build ITK with the ModuleRTK and ITK_WRAP_PYTHON flags set to >>>>> ON. >>>>> Everything seems to work fine except that some RTK filters are >>>>> missing. E.G ADMMTotalVariationConeBeamReconstructionFilter is not >>>>> available. >>>>> >>>>> Is it expected behavior ? Or have I missed something in the build >>>>> process ? >>>>> >>>>> Far less critical : once the submodule RTK is loaded in python, the >>>>> first call to any RTK object takes quiet a long time to execute. I also >>>>> have a few warnings (like the one below) when this first call runs. >>>>> >>>>> Template itk::PyBuffer<itk::Image<itk::CovariantVector<double,2>,2>> >>>>> >>>>> already defined as <class 'itkPyBufferPython.itkPyBufferICVD22'> >>>>> >>>>> is redefined as <class 'itkPyBufferPython.itkPyBufferICVD22'> >>>>> >>>>> >>>>> Thanks in advance for your help, >>>>> >>>>> >>>>> Jérome >>>>> _______________________________________________ >>>>> Rtk-users mailing list >>>>> Rtk-users@public.kitware.com >>>>> https://public.kitware.com/mailman/listinfo/rtk-users >>>>> >>>>
_______________________________________________ Rtk-users mailing list Rtk-users@public.kitware.com https://public.kitware.com/mailman/listinfo/rtk-users
