Can't install packages

Hello,


I've just started working with python and with pycharm, and i hit a little problem...and the problem is that i can't install packages.

i tried installing numpy, matplotlib and beautifulsoup. and they all fail with "error unable to find vcvarsall.bat "

i run a clean install of windows 10 professional, python 3.5(64bit), and visual studio 2012. all the install paths are default.

i also tried the action that is being suggested, to run: pip install numpy, and it fails with the same error.

Could someone tell me what am i doing wrong ?

all the best,
Mihai.
3 comments
Comment actions Permalink
I believe that is usually an error regarding packages that want to compile C extensions, and for whatever reason, you aren't able to get binary packages installed.

Perhaps you could eliminate PyCharm from the equation and see if you can pip install those packages directly from the command line, using the same Python interpreter you are using in PyCharm?

If that also fails, do a little googling for vcvarsall.bat...I think you might have to install the free edition of Visual Studio.

As an alternative, I think the Anaconda distribution for Python is tailored to you...people that want NumPy etc. without a lot of fighting. PyCharm 5 now supports Anaconda environments. It will mean a little up-front work to learn about Anaconda and get it going, but it might pay off in the longer term.
0
Comment actions Permalink
can you elaborate on installing the free version of Visual Studio?
what version of VS is required for the install to work?
0
Comment actions Permalink
I'll admit, I'm neither a Windows guy nor a NumPy guy, but a little Googling turned up this: https://gehrcke.de/2015/02/how-to-set-up-a-64-bit-version-of-numpy-on-windows/

It mentions that NumPy does not provide pre-compiled binaries for 64 bit. That page then goes on to mention the alternatives, including Anaconda.

The NumPy download site says that it provides Windows installers for the binaries: http://www.scipy.org/scipylib/download.html

All in all, my guess is that you should head over to the NumPy universe and look for some tips. I suspect this is a very common occurrence. But hopefully the two links above provide enough. Good luck!
0

Please sign in to leave a comment.