mahatmacane / conda-execute Goto Github PK

A tool for executing scripts in a defined temporary environment.

$ cat my_script.py 
#!/usr/bin/env python
"""
A script that uses numpy's random normal distribution to print 3 numbers.

"""

# conda execute
# env:
#  - python >=3
#  - numpy

import numpy as np

print(np.random.normal(size=3))

$ conda execute -v my_script.py 
Using specification: 
env: [python >=3, numpy]
run_with: [/usr/bin/env, python]

Prefix: /Users/pelson/miniconda/tmp_envs/ea977067a8fbeb21a594

[ 1.06976755  1.23957678  0.09308639]

Conda execute is installable into the root conda environment with:

conda install conda-execute --channel=conda-forge

Alternatively, conda-execute is a pure-python package, and is easy to install from source.

If you want to make a shell script that can be run with conda execute directly, rather than having to call conda execute each time, ensure that the shebang points at conda-execute and the run_with metadata is defined:

$ cat execute_directly.py 
#!/usr/bin/env conda-execute
"""
A script that uses numpy's random normal distribution to print 3 numbers.

"""

# conda execute
# env:
#  - python >=3
#  - numpy
# run_with: python

import numpy as np

print(np.random.normal(size=3))

$ ./execute_directly.py 
[-0.46194591  0.13287211 -0.10139428]

conda execute automatically cleans up all environments which are unused in the last N hours (configurable, default 25). However, to manually run the cleanup process, it is possible to inspect the temporary environments with:

$ conda tmpenv list

And any environments with 0 running processes can be removed with:

$ conda tmpenv clear

conda execute has been written to allow concurrent conda execute usage whilst at the same time sharing environments. This means that conda execute must make use of conda's locking machinery to avoid race-conditions. If you experience issues with the locking, please raise an issue with as much detail as possible.