Updated ``yml`` to include notebooks and other tools handy for development.


kwant.yml
```
name: kwant
channels:
  - conda-forge
dependencies:
  - python=3.5
  - gcc=4.8.5
  - lapack=3.6.1
  - mumps=5.0.2
  - numpy=1.11.2
  - scipy=0.18.1
  - cython=0.24.1
  - matplotlib=1.5.3
  - pytest=2.9.2
  - pytest-cov=2.4.0
  - sympy=1.0
  - notebook=4.3.0
  - holoviews=1.6.2

```

On 16 December 2016 at 13:12, Rafal Skolasinski <r.j.skolasin...@gmail.com>
wrote:

> A bit updated file to include notebook and holoviews.
>
> On 16 December 2016 at 13:07, Rafal Skolasinski <r.j.skolasin...@gmail.com
> > wrote:
>
>> Hi everyone,
>>
>> I was just setting up a minimal development environment (for work on
>> discretizer with conda) and I wanted to share my solution with you.
>>
>> The goal was to use conda to install kwant in develop mode to simplify
>> development and testing.
>>
>> Here is my recipe (I assume conda is installed):
>> ```
>> # create environment and activate it
>> conda env create -f kwant.yml
>> source activate kwant
>>
>>
>> # install master of tinyarray's
>> pip install git+ssh://g...@gitlab.kwant-project.org:443/kwant/tinyarray.g
>> it
>>
>>
>> # cd into project with your fork of kwant
>>
>>
>> # obtain and modify build.conf
>> wget https://raw.githubusercontent.com/conda-forge/kwant-feedstoc
>> k/master/recipe/build_linux.conf -O build.conf
>> sed -i -e "s:PREFIX:$CONDA_PREFIX:g" build.conf
>>
>>
>> # build and install kwant
>> python setup.py build
>> python setup.py develop
>> ```
>>
>> Now kwant should be installed into your new environment and any changes
>> you make to source files should be easy to test. Just select ``kwant``
>> kernel in any of your notebooks and you will be using your local copy of
>> kwant.
>>
>>
>> I hope it will be helpful, and I wait for your comments. The environment
>> is missing things like  jupyter notebook and holoviews but it can be added
>> with single line of
>> ```
>> conda install jupyter notebook holoviews
>> ```
>> etc.
>>
>> Cheers,
>> Rafal
>>
>>
>> P.S. Libraries version like numpy and scipy are relatively new and don't
>> reflect old versions that are minimum requirement for kwant. I believe that
>> this should not be a problem because CI is running tests against this
>> versions.
>>
>
>

Reply via email to