# vtkpython 3.7 error with windows debug build
**URL:** https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692
**Category:** Development
**Created:** [April 5, 2019, 11:11pm UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692 "2019-04-05T23:11:59Z")
**Posts on this page:** 9
**Page:** 1
### Author:  [@Aron\_Helser](https://discourse.vtk.org/u/Aron_Helser)
#### Post date: [April 5, 2019, 11:11pm UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692/1 "2019-04-05T23:11:59Z")
I’m seeing an error when trying to use vtkpython with python 3.7 and a debug build:
Python 3.7.3 (v3.7.3:ef4ec6ed12, Mar 25 2019, 22:22:05) [MSC v.1916 64 bit (AMD64)] on win32
Type “help”, “copyright”, “credits” or “license” for more information.
> > > import vtk
> > > Traceback (most recent call last):
> > > File “”, line 1, in
> > > File “D:/akit/vtk/build\bin/Lib/site-packages\vtk.py”, line 32, in
> > > all\_spec.loader.exec\_module(all\_m)
> > > File “D:/akit/vtk/build\bin/Lib/site-packages\vtkmodules\all.py”, line 7, in
> > > from .vtkCommonCore import \*
> > > File “D:/akit/vtk/build\bin/Lib/site-packages\vtkmodules\vtkCommonCore.py”, line 2, in
> > > from .vtkCommonCorePython import \*
> > > ModuleNotFoundError: No module named ‘vtkmodules.vtkCommonCorePython’
It appears to be because a “\_d” is appended to all the compiled libraries in the vtkmodules directory.
A ‘RelWithDebInfo’ build works fine.
---
### Author:  [@efahl](https://discourse.vtk.org/u/efahl)
#### Post date: [April 5, 2019, 11:37pm UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692/2 "2019-04-05T23:37:52Z")
I believe that you must use python\_d.exe to run using the full debug VTK build, which then forces you to use full debug of all other packages, which, well, good luck…
If you only need the .pdb files so you can get meaningful tracebacks in the debugger, then the RelWithDebInfo build is the way to go.
---
### Author:  [@Aron\_Helser](https://discourse.vtk.org/u/Aron_Helser)
#### Post date: [April 6, 2019, 12:37am UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692/3 "2019-04-06T00:37:14Z")
That makes no sense - why would python force all packages to be debug?
Anyway, it’s something to try, when I get some time. I can make progress using ‘RelWithDebInfo’, so I mostly wanted to report something in case someone else got stuck.
---
### Author:  [@dgobbi](https://discourse.vtk.org/u/dgobbi)
#### Post date: [April 6, 2019, 2:28pm UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692/4 "2019-04-06T14:28:07Z")
Related MR: [https://gitlab.kitware.com/vtk/vtk/merge\_requests/5316](https://gitlab.kitware.com/vtk/vtk/merge_requests/5316)
---
### Author:  [@efahl](https://discourse.vtk.org/u/efahl)
#### Post date: [April 6, 2019, 11:15pm UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692/5 "2019-04-06T23:15:43Z")
Python doesn’t force it, and really has nothing to do with it. This is how Microsoft names debug libraries when you turn on various debug compiler/linker flags.
---
### Author:  [@ben.boeckel](https://discourse.vtk.org/u/ben.boeckel)
#### Post date: [April 8, 2019, 1:29pm UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692/6 "2019-04-08T13:29:04Z")
Python, in debug mode, expects `_d` suffixes on its module library names. See [the code](https://github.com/python/cpython/blob/master/Python/dynload_win.c#L21). Windows is just a world where everything is either debug or not debug. The differing runtime libraries between the two configurations requires it. Now, there could certainly be an option to use a release Python with a Debug VTK. That would basically just remove the `_d` suffix, but if `IMPORTED_LOCATION_DEBUG` and such are set on the `Python3::Python` target, we’re going to end up with the debug version anyways.
> This is how Microsoft names debug libraries when you turn on various debug compiler/linker flags.
It’s just convention because they are incompatible with each other. There’s nothing forcing a `d` (or `_d` suffix) on debug libraries except for it being the common solution to shipping debug and release libraries side-by-side.
---
### Author:  [@Aron\_Helser](https://discourse.vtk.org/u/Aron_Helser)
#### Post date: [April 8, 2019, 2:00pm UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692/7 "2019-04-08T14:00:27Z")
What is the recommendation to someone doing a debug VTK+python build on windows, for running scripts? They have to build with a debug version of python? Does it work? (I haven’t had a chance to try it…)
Thanks,
Aron
---
### Author:  [@dgobbi](https://discourse.vtk.org/u/dgobbi)
#### Post date: [April 8, 2019, 3:06pm UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692/8 "2019-04-08T15:06:06Z")
My approach to debugging on Windows is to use a RelWithDebInfo build, as Eric suggested earlier. But I’m not a Visual Studio afficionado and prefer not to do side-by-side builds, I switched to Ninja many years ago.
---
### Author:  [@ben.boeckel](https://discourse.vtk.org/u/ben.boeckel)
#### Post date: [April 8, 2019, 3:11pm UTC](https://discourse.vtk.org/t/vtkpython-3-7-error-with-windows-debug-build/692/9 "2019-04-08T15:11:27Z")
I use `RelWithDebInfo` by default pretty much everywhere due to performance degradation in general. The `_d` suffix on Windows does require a debug Python right now. It should be shipped with official `python.org` packages. Not sure about Anaconda. It’s not something that’s been extensively tested though.