search

ceholden / yatsm

Yet Another Time Series Model
https://yatsm.readthedocs.org/en/latest/
MIT License
63 stars 30 forks source link

Installing yatsm error #105

Open mlateb opened 7 years ago

mlateb commented 7 years ago

Dear, please I having an error message below when I'm trying to install yatsm on Ubuntu 17.04 Anaconda3. What does it mean? Thanks

Command "/home/user/anaconda3/envs/yatsm/bin/python -u -c "import setuptools, tokenize;file='/tmp/pip-kred0d4t-build/setup.py';f=getattr(tokenize, 'open', open)(file);code=f.read().replace('\r\n', '\n');f.close();exec(compile(code, file, 'exec'))" install --record /tmp/pip-qkvpe3vz-record/install-record.txt --single-version-externally-managed --compile" failed with error code 1 in /tmp/pip-kred0d4t-build/

CondaValueError: pip returned an error.

P.S. if it's not the right place to post such request please let me know

ceholden commented 7 years ago

Hello! I'm hoping we can get you past the installation error, but I will need some more information as I can't reproduce the issue locally (using another variety of Linux) or on the Travis-CI's test servers (which are also running Ubuntu).

The step that you're having an issue with installs glmnet-python, a Python around Fortran code for doing LASSO regression. This library, while faster than the alternative in Scikit-Learn, is an optional dependency so worst case we can just ignore it. Conda is using pip to install this library from a source on Github.

I have a few guesses or paths forward:

  1. Since the pip step is accessing a network resource, it might be worth retrying it in case there were network issues the day you tried it.
  2. If your "conda" installation is old, it might be worth trying to recreate the entire install (i.e., remove /home/user/anaconda3 and re-install from the Anaconda or Miniconda installers).
    • You can also try upgrading the core packages in the "root" (which handle things like installation) by running conda upgrade -n root --all.
    • You can check the version of conda using conda --version (I'm running conda 4.3.23 right now)
  3. The URL for the glmnet-python is using the git+git specification but there might be others that work (see pip Git docs). You might try editing the environment.yaml file and replacing the git+git with git+https or git+ssh.
    • Using SSH the URL would look like git+ssh://git@github.com/ceholden/glmnet-python.git@v2.0.1
    • Using HTTPS the URL would look like git+https://github.com/ceholden/glmnet-python.git@v2.0.1
  4. If it's still not working, you might consider removing the dependency from the environment.yaml file entirely

Let me know how it goes. Hoping some of these steps can resolve your issue!

Best Chris

mlateb commented 6 years ago

Hi Chris! The upgrading conda upgrade -n root --all worked for me. Thanks a lot!