Skip to content

Refactor logic around requirements generation #82

Description

@nealrichardson

#75 adds another condition to govern whether to (re)generate a requirements.txt from a uv.lock or pyproject.toml, based on app modes. This mostly works but would break down for R content that uses reticulate, and in general feels like we're making extra complicated logic with functions that have many arguments switching off behavior.

I wonder if there's a cleaner way we can solve this. As in, if the working directory contains a pyproject.toml or uv.lock but no requirements.txt, run the generate-requirements step. (I'm not sure what this would do for python content that contains no requirements files of any kind, we might need to confirm that it errors cleanly.)

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions