Closed anton-matosov closed 6 years ago
Great!We will be easy to install the multineat library without building the 'MultiNEAT-master' file,so it can make the install process more smooth!I will support and pay close attention to the library continuously.
Up.
It says these files are conflicting: src/NeuralNetwork.cpp src/NeuralNetwork.h src/PythonBindings.h
Ok, I will take a look
I have fixed all the conflicts
Thanks!
You are welcome.
BTW, I have more fixes and improvements in the https://github.com/MultiNEAT/MultiNEAT And MultiNEAT is now available on Conda-forge, and just in 23 hours reached 96 downloads. I am going to maintain that fork and you are welcome to join me there.
P.S. I have pulled all the fixes from this repo as well, so it is not behind by a single revision (pending CI validation and merge atm).
I have ported library to latest boost and implemented conda package build scripts to have fully contained distribution. I have also altered examples a little bit to make them suitable as smoke tests. Supported configurations:
I couldn't build library on Windows for python 2.7 because it uses super old compiler from VS2008 that doesn't support C++11 features. So this configuration is not supported.
This PR also includes complete CI setup for all these configurations. For Linux and OSX it uses Travis-CI For Windows - Appsveyor
Configuration environment variables:
CONDA_UPLOAD
- whether to upload package to anaconda.org. possible values:yes
orno
CONDA_LOGIN_USERNAME
- username on anaconda.org to upload package toCONDA_LOGIN_PASSWORD
- password on anaconda.org to upload package toIn future I want to submit it as feedstock to conda-forge to make re-distribution easy. Currently packages are available from my personal channel with boost from conda-forge:
conda install multineat -c anton.matosov -c conda-forge