From @brettcannon on September 20, 2017 20:42

So here is my thought of what a "create environment" command could do ...

Choose an interpreter
1. If one already selected, great!
2. Otherwise prompt user to select an interpreter
Anaconda
1. Create environment: conda create –name $ENV –yes –file environment.yml
  - Should we match on *.yml and prompt to disambiguate?
2. Create activated terminal
  - Windows: activate $ENV
  - UNIX: source activate $ENV
Pip
1. Create environment
  - Python < 3.3: virtualenv --python $PYTHON $ENV
  - Python >= 3.3: python3 –m venv $ENV
2. Create activated terminal
  - PS
  - Cmd.exe
  - Bash
  - csh
  - fish
3. Install dependencies
  1. If multiple matches for requirements.txt, prompt user to choose
  2. In activated terminal: pip install –upgrade –r $REQUIREMENTS
Make the created environment the selected interpreter for the project

IOW, simplify the following down to a command with at most 2 prompts (environment name and requirements/environment file):

$ python3 –m venv env 
$ source env/bin/activate 
$ pip install –r requirements.txt

End up with an activated terminal
Set the interpreter for the project set to the newly-created environment.

Not sure if this is common enough to warrant implementing, though. Would this be useful in a teaching environment with heterogeneous OSs where people might mess up one of these steps? Or in that instance do we just say "use a shell script" if they are all on the same kind of system or "just explain what you're doing" when it's a mixed set of OSs?

Uh oh!

Provide a "Create environment" command #160

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions