search

jorgenkg / python-neural-network

This is an efficient implementation of a fully connected neural network in NumPy. The network can be trained by a variety of learning algorithms: backpropagation, resilient backpropagation and scaled conjugate gradient learning. The network has been developed with PYPY in mind.
BSD 2-Clause "Simplified" License
297 stars 98 forks source link

Your package is py2.x and it installs on py3.x #17

Open KamodaP opened 7 years ago

KamodaP commented 7 years ago

Your package has syntax of standard python 2.x version, but probably has a fault in pypi configuration, since it downloads on python3.x, installs, and fails ungraciously.

jorgenkg commented 7 years ago

Thank you for making me aware of this issue. If this is indeed something I can configure in the pypi config, I will take a look into it

KamodaP commented 7 years ago

https://packaging.python.org/distributing/#setup-args Classifiers section. I believe that if you specify there python2 it should not allow accidentally installing on python3. I’ve forked your project and prepared for compliance with python3, I will set up a pull request, but I’ve read that you wish to stay with python2, in that case please just leave pull request hanging, it will allow users to more easily find proper distribution. But if you want to have python3.x compliance I can help with setting up packages on pypi

jorgenkg commented 7 years ago

Unfortunately, that configuration have already been specified in the project's setup.py:

classifiers      = [
          'Programming Language :: Python :: 2.7',
          'Topic :: Scientific/Engineering :: Artificial Intelligence',
        ],

You are correct. I've decided to stick to Python 2.7 for now since the old version still rivals the performance of the new interpreter, and is the targeted python version of PYPY.

KamodaP commented 7 years ago

You can have both version simultaniously. Like you can download numpy for python2.x and for python3.x. I don't know what kind of configuration it would require, but I'm certain it works. when you create the package you have something like pkg_name-any-any-win32.tgz? I think that this name is miningfull, like when you look at pypi numpy they have separate files with the same naming convention. Like numpy-1.12.0b1-cp27-none-win_amd64.whl