Best practice for sharing workflows containing external (Python) dependencies

[wintron]

Hi there!

 

I've created a workflow that I want to share with my team, the workflow includes a Python script, which in turn has a dependency on a 3rd party library (the wonderful Requests[1]).

 

Are there any best practices around sharing this workflow with my team, whilst easing friction around the third party dependency? For example, currently I'm advocating dropping down to the command line and installing the Python dependency globally, ideally I'd like to avoid this, if at all possible.

 

Appreciate any pointers.

 

Kind regards!

Steve

 

 

[1] http://docs.python-requests.org/

[deanishe]

You bundle the dependency with your workflow. End of discussion, basically.

 

cd to your workflow in a shell and run: pip install --target . requests

 

 

Asking users to install libraries is a PITA, and is very likely to cause problems down the line as Python doesn't support versioned libraries. Anything that can't be bundled probably has C extensions that need compiling, which is a world of hurt for the person supporting that.

Edited March 3, 2016 by deanishe

[wintron]

Didn't know about pip install --target .

 

So easy now, many thanks!

[deanishe]

My favourite pip trick:

 

When you're not sure which Python (or pyenv etc.) the pip command is pointing to, you can use /path/to/the/python -m pip --pip --options pip args