0

I’m new to Python and I’ve just downloaded Pycharm. I'm trying to install the Numpy package through the Project Interpreter, but I keep getting this error message

error: Microsoft Visual C++ 10.0 is required (Unable to find vcvarsall.bat).

I've spent the entire day checking for solutions, including installing C++ 10.0, playing around with the environment variables, and also updating everything I can update within Pycharm. But I’m still unable to get this to install.

Could someone please kindly provide a step by step guide to solving this issue?

Anthon
  • 69,918
  • 32
  • 186
  • 246
jake wong
  • 4,909
  • 12
  • 42
  • 85
  • I've followed this: http://stackoverflow.com/questions/2817869/error-unable-to-find-vcvarsall-bat and installed VS2010. It solved my initial problem of (Unableto find vcvarsall.bat). But now it throws a new problem when I'm trying to install the modules ValueError: ['path'] Any ideas? @dnlcrl Also, I'm trying to install Pandas. Not just Numpy so it seems that installing WinPython as you suggested doesnt help. – jake wong May 24 '15 at 10:36
  • It turns out that I am using 64 bit python, and its difficult to install Numpy on 64 bit... So I've reverted to 32 bit python and it solved the problem.. – jake wong Jun 01 '15 at 08:55

1 Answers1

0

Installing some modules in a windows environment is difficult. There is also a risk of affecting the underlying system or making a mess of things...

I recommend this:

Install the Anaconda distro of Python. This is more stable and comes with common packages preinstalled, including numpy.

Use conda package manager to create a virtual environment as your project interpreter.

Link: https://store.continuum.io/cshop/anaconda/

rask004
  • 552
  • 5
  • 15
  • In what sense is Anaconda more stable? More tests where can I find those? – Anthon May 24 '15 at 12:49
  • If I were to use Microsoft Visual Studio instead, would there be lesser trouble to add these packages? – jake wong May 24 '15 at 14:03
  • Anthon, What I meant to say was I found package installation under conda to be less frustrating and more likely to produce a working package i.e. it doesn't suddenly crash due to a missing dependency I hadn't identified or a compilation runtime bug that crept in. – rask004 May 25 '15 at 02:12