Irv Lustig

I tried this on Windows. Doing `python setup.py` caused the cythonize step to be done, but it was done serially. Can't that be done in parallel as when we do...

Well, I'm not sure I have the time nor understanding to get this working on Windows!

So now that I'm on Visual Studio 2019 build tools, there is an issue with the cmake version when doing `cmake .` . ``` CMake Error at CMakeLists.txt:1 (cmake_minimum_required): CMake...

I made the following changes to `CMakeLists.txt`, and got a lot further on Windows: ``` cmake_minimum_required(VERSION 3.15) set(CMAKE_SHARED_LIBRARY_SUFFIX .cp38-win_amd64.pyd) ``` Then I discovered that you have to specify the architecture...

> However, we today do in source builds of pandas. So want to make sure we emulate that as best as possible for initial lift and shift Once you figure...

> 1. The min required version is 3.17 (released March 2020). This is because the Python_SOABI variable was introduced then, which gives us the required platform identifier I think that...

I installed cmake into my conda environment on Windows. removed all cmakefiles from previous trials. Then pulled down the PR, and got this: ``` cmake --build . --target clean Error:...

> @Dr-Irv did you run `cmake .` first? That should configure everything and create CMakeCache.txt OK, now I did that, but had to run it twice. Here is the output:...

> You might need to clean and try again for the remaining errors. My guess is some of the debug / release profiles generated by VS got mixed up I...

> Yea I think the different build types got mixed up when running the build without specifying `--config Release` though I'm still stuck with those errors about `numpy.core.multiarray failed to...