These are chat archives for spyder-ide/public
Spyder laughed at me
What does that mean? Please be clear if you want us help
@jpnoon I presume one of two things happened: You got an
ModuleNotFoundError, or you got the
ordinal 24n not found mkl issue; in the former case, that would be Python laughing at you; in the latter, that would be mkl/NumPy at fault; in neither case does Spyder have anything directly to do with the
import not working.
If the former, NumPy isn't installed in the environment you installed Spyder in; Spyder is an IDE, not a Python distribution like Anaconda (which we recommend installing it with, as it includes NumPy and many other useful Python tools, along with Spyder itself. To fix, you can either install NumPy into the same environment you're launching Spyder from, or install
spyder-kernels into the desired environment with NumPy and tell Spyder to launch kernels in that environment instead under the Preferences. You can read more about that in our guide to working with packages and environments with Spyder.
If the latter, that's a problem with mkl under Anaconda on Windows that they are looking to fix. If so, you should run
conda install mkl=2018.0.2 to fix it for now until they can resolve it on their end. Best of luck!
@jpnoon Again, if you can just share your actual error message with us, we can likely provide much more specific guidance on what your problem is and how to solve it.
There is some versions of Python that work with Numpy and others that don't.
NumPy supports Python 2.7 and 3.4+; any incompatibility with a specific Python version within that range due to Numpy itself is a bug in Numpy, and should be reported to them. However, thanks to continuous testing, the likelihood of it not importing at all or even very basic functionality working on a specific version due to a bug in either Numpy or Python is very low indeed.
If you have problems with specific Python versions, that's likely due to either binary distribution formats (wheels or conda packages) not being built against your particular platform, bitness, NumPy and Python version, or some specific problem/incompatibility with the build and your particular machine. This has nothing to do with NumPy directly, but rather whomever packages, builds and distributes it and the libraries it depends on.
If you're not using Anaconda, you're welcome to try to get it working and we do create and maintain Spyder builds for
pip installs, but these sorts of platform-specific issues are much more likely to crop up with
pip, and so unless you're able to fix them on your own (there isn't much we can do anyway) we recommend sticking to Anaconda. These sorts of issues are generally minimized in Anaconda since they work hard to ensure wide compatibility, but there is a current issue with the Intel
mkl library on Windows which causes a DLL error to appear when trying to
import numpy on certain platforms and configurations. They are working to fix it upstream, but for now the fix is to run
conda install mkl=2018.0.2 if you do see an "Ordinal 24N" error pop up.
Best of luck!