Previous Releases

Table of Contents

• v0.2.0

v0.2.0

Overall aim: Set up repository and codebase to allow for collaboration. No new functionality.

Add a Changelog

• What: Add a CHANGELOG.MD to the root of the repository to communicate changes in the code.
  • Why: Human understandable changes over time, grouping changes (add features, changed functionality, bugfixes, etc). Please read https://keepachangelog.com/en/1.0.0/
  • Who: leifdenby
  • PR: https://github.com/joeloskarsson/neural-lam/pull/28

Add PR Template

• What: Ensure that we know for each contribution what the changes are.
  • Why: Having a template will ensure clarity.
  • Who: leifdenby

Turn Repository Fully into Package

• What: Move all *.py into neulam_lam folder, replace absolute imports by relative.
  • Why: All files in source folder enable installation of code as package (this is how Python packages are organized). Absolute imports should not be used inside package (PEP).
  • Who: leifdenby
  • In-support: leifdenby, khintz

Simplify Pre-commit Configuration and CI/CD Use

• What: Pin linter versions in pre-commit config, use GitHub action for running pre-commit.
  • Why: Without pinning versions, we rely on each user installing the correct version manually. CI/CD setup currently is less maintainable because we install pre-commit etc manually rather than using community GitHub action.
  • Who: leifdenby
  • Issue: https://github.com/joeloskarsson/neural-lam/pull/29
  • In-support: leifdenby

Replace neural_lam/constants.py with YAML-config File

• What: Currently constants.py is referenced many places in the code, but the definitions in here are actually specific to the dataset. We should generalize this so that constants can be defined for each dataset.
  • Why: Want to be tied.
  • Who: sadamov, review: thomas (Ireland), leif
  • Issue: https://github.com/mllam/neural-lam/pull/31

Add Tests for WeatherDataset(pytorch.Dataset)

• What: Add test that ensures WeatherDataset can be initiated and that return items have the correct shape.
  • Why: Currently the way that WeatherDataset is stored and what is returned is written in comments, but that doesn't check execution.
  • Who: SimonKamuk, leifdenby
  • Issue: https://github.com/mllam/neural-lam/issues/30
  • Relevant code: Test for MEPS

Handle Package Dependencies in pyproject.toml Instead of requirements.txt

• What: Put all dependencies in pyproject.toml with pdm and sync pyproject.toml to requirements.txt for pip-based installs.
  • Why: Handling dependencies by manually keeping requirements.txt updated is a tedious and error-prone task. requirements.txt don't fully describe a package (version, description) and are therefore no longer the recommended way of defining Python packages.
  • Who: leifdenby
  • Pull-request: https://github.com/mllam/neural-lam/pull/32