[feat] initialize from an existing environment

[AlbertDeFusco]

In this PR

• implement conda-project init --from-environment <name or prefix>
• update to conda-lock 2.5.6
  • 2.3.0 removes the need for a temporary lockfile
  • 2.5.6 fixes input hash updates on re-lock
• timestamp now added to lockfile
• on conda-project install only re-lock if the current platform needs to be re-locked
  • previously the lockfile was re-solved if any platform needed it

Using the --from-environment <name or prefix path> flag the environment.yml and conda-lock.<env-name>yml files will be generated from an existing conda environment.

The conda-lock.<env-name>.yml for the current platform is constructed by performing a full conda env export so that the environment is captured as-is into the lockfile. Only packages for the current platform are written in to the lock. When someone attempts to install on another platform conda-project will automatically lock the missing platforms, but not modify the original platform, unless the environment.yml file was changed.

The environment.yml captures the requested conda and pip packages as follows. You'll see that conda packages are prefixed with its channel and the full version is captured in the spec even if a specific version was not originally requested. This helps keep the environment close to the original if the original platform is re-locked and when other platforms are locked.

You'll see that all four major platforms are listed in the environment.yml, but a full cross-platform solve is not attempted by default. The user could do conda project init --from-environment <env-or-path> --lock, which will capture the current platform according to the above procedure and then lock the remaining platforms.

name: data-science-work
channels:
  - defaults
  - anaconda-cloud
  - conda-forge
dependencies:
  - defaults::hvplot=0.9.2
  - defaults::pip=23.3.1
  - defaults::scikit-learn=1.3.0
  - defaults::pandas=2.1.4
  - conda-forge::wheel-filename=1.4.1
  - defaults::panel=1.3.8
  - defaults::python=3.11.5
  - defaults::python-libarchive-c=2.9
  - defaults::ipykernel=6.25.0
  - defaults::intake=0.6.8
  - conda-forge::ratelimit=2.2.1
  - pip:
    - anaconda-catalogs==0.2.0
    - torch==2.2.2
variables:
prefix: /Users/adefusco/Applications/miniconda3/envs/data-science-work
platforms:
  - osx-arm64
  - linux-64
  - osx-64
  - win-64

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 98.23%. Comparing base (8aab898) to head (1dabb92).
Report is 5 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #162      +/-   ##
==========================================
+ Coverage   97.89%   98.23%   +0.33%     
==========================================
  Files           9        9              
  Lines         996     1074      +78     
==========================================
+ Hits          975     1055      +80     
+ Misses         21       19       -2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jjhelmus]

LGTM