edickie / ciftify

The tools of the Human Connectome Project (HCP) adapted for working with non-HCP datasets
https://edickie.github.io/ciftify/
MIT License
111 stars 156 forks source link

Installation and read me updates #15

Closed DESm1th closed 7 years ago

DESm1th commented 7 years ago

The repo is now almost pip and conda installable! Only two steps left from you to officially set up.

  1. You have to make a github release for the pip install to work. Ciftify is too big for PyPi to host and PyPi doesnt allow external hosting anymore, so instead we have to pip install directly from a release on github. I updated the readme to show the correct pip command except that it points to my copy's unofficial release. So you'll have to create your official github release with the tag 0.1.0 and update the readme to point to the link to your .tar file instead

  2. You'll have to make an anaconda.org account and upload our conda build if you want to keep control of the conda distribution. I've made and tested the config files so all you really need to do is upload a build. So you'll need to have conda + conda-build and ensure that both are up to date, and then on your machine run conda build $path-to-your-ciftify-git-clone

Optionally, take the .tar.bz2 file that generates and then generate versions for other operating systems with (we're not windows compatible): conda convert -p osx-64 -p linux-32 $path-to-.tar.bz2 -o $your-output-dir

If the system you build on is not a 64 bit linux distribution you'll have to swap in linux-64 in place of the -p argument that matches your system.

And then upload all tar files that are generated by following the instructions here or possibly by the anaconda.org gui if it allows it.