ruancomelli / boiling-learning

Deep learning models for phase-change quantification and classification.

Boiling Learning

Image generated by DALL-E mini using the prompt "robot watching bubbles in water"

---

About

Project developed by Ruan Comelli at UFSC (Federal University of Santa Catarina) in order to obtain a Master's degree in Mechanical Engineering.

If you are curious about the evolution of this project, take a look at the changelog.

Versioning

This project uses ZeroVer, a versioning scheme in which software never leaves major version 0. This means that breaking changes are expected frequently.

Given a version number 0.x.y, the minor version x will be incremented whenever intentional breaking changes are introduced. If no new breaking changes are added, new releases will only increment the patch version y for both bug fixes and new features. This is similar to how SemVer treats the major and the minor version numbers, respectively.

Upcoming

• Improved README.

boiling-learning's Issues

Use `nitpick` to merge all of our configuration files into a single one

Nitpick can be used to replace the many configuration files that our project has. Perhaps it can be of use to us.

Remove `phantom` dependency

The library phantom does indeed look awesome, but it is used in very few places in Boiling Learning. Actually, only a single class depends on it, and it can be easily replaced with a metaclass containing an __isinstancecheck__ method.

By getting rid of another dependency, installing Boiling Learning becomes faster and less error-prone.

Update README.md

Currently, the README.md file describes an old version of the project. The main points to change are:

• Installation:
  • installation now should be done preferably using poetry
  • the files requirements.txt, requirements-dev.txt and requirements-scripts.txt no longer exist or are no longer used
  • the command python setup.py install will no longer work since there is no file setup.py
• Contributing: poetry install should be enough to install the necessary development dependencies
• Making changes: the preferable way of running tests and coverage check is running make test and make coverage
• Committing: the current way of committing is by running make commit
• Releasing new versions: to release a new version automatically, the recommended way is running make release
• Upcoming: I should probably include a "how to use" section

Clean up to-do list

Currently our to-do list contains many to-do items that we have already implemented or that no longer make sense adding to this project.

We should clean it up and probably move all remaining items to a section of the README or open corresponding GitHub issues, one for each.

Add `pdm-managed` badge to README

The PDM dependency manager provides a badge that we should include in our README in order to more easily inform our readers that we make use of it.

Replace references to SemVer with ZeroVer

ZeroVer describes best the current versioning scheme of this project since we are in version v0.X.Y, and it is unlikely that we will ever reach v1.

Disable Renovate

This project should be archived: it is fixed to a point in time and should allow for a complete reproduction of my Master's work. Hence, updating dependencies does not make sense since that could change behavior or reduce compatibility with the data I have collected.

Dependency Dashboard

This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.

Open

These updates have all been created already. Click a checkbox below to force a retry/rebase of any.

• chore(deps): update dependency tensorflow to <2.11.2,>=2.11.1 [security]
• chore(deps): update dependency autokeras to v1.1.0
• chore(deps): update dependency keras-tuner to v1.4.7
• chore(deps): update dependency python to v3.12.4
• chore(deps): update dependency autokeras to v2
• chore(deps): update pdm-project/setup-pdm action to v4
• chore(deps): update peter-evans/create-pull-request action to v6
• Click on this checkbox to rebase all open PRs at once

Detected dependencies

asdf

.tool-versions

• python 3.10.8

github-actions

.github/actions/setup-python-environment/action.yaml

• actions/checkout v4
• pdm-project/setup-pdm v3

.github/workflows/ci.yaml

• actions/checkout v4

.github/workflows/pre-commit-auto-update.yaml

• actions/checkout v4
• actions/setup-python v5
• browniebroke/pre-commit-autoupdate-action v1.0.0
• peter-evans/create-pull-request v5

.github/workflows/release.yaml

• actions/checkout v4
• actions/checkout v4
• actions/create-release v1
• actions/upload-release-asset v1

pep621

pyproject.toml

• tensorflow <2.11.0,>=2.10.0
• tensorflow-addons >=0.18.0
• typing-extensions >=4.3.0
• frozendict >=2.3.4
• funcy >=1.17
• more-itertools >=9.0.0
• numpy >=1.23.0
• parse <2.0.0,>=1.19.0
• pint >=0.20
• scikit-image >=0.19.1
• classes <1.0.0,>=0.4.1
• iteround <2.0.0,>=1.0.3
• loguru >=0.6.0
• decord ==0.6.0
• keras-tuner ==1.1.3
• autokeras ==1.0.20
• matplotlib >=3.0.0
• seaborn >=0.12.2
• imageio >=2.22.4
• Pillow >=7.1.2
• rich >=12.6.0
• typer >=0.7.0
• pandera >=0.13.4
• ffmpeg-python >=0.2.0
• scipy >=1.9.3
• pandas >=1.5.2
• h5py >=3.7.0
• uncertainties >=3.1.7
• pyyaml >=6.0
• test/pytest >=7.1.2
• test/coverage >=6.4.3
• test/tox >=4.0.0
• test/tox-gh-actions >=3.0.0
• test/tox-pdm >=0.6.0
• lint/black >=23.3.0
• lint/isort >=5.12.0
• lint/unimport >=0.16.0
• lint/pylint >=2.17.0
• lint/vulture >=2.7
• ci/commitizen >=2.29.5
• ci/pre-commit >=2.20.0
• build/pip >=22.2.2
• types/mypy >=1.2.0
• types/types-decorator >=5.1.8
• types/types-PyYAML >=6.0.11
• types/funcy-stubs >=0.1.1

---

• Check this box to trigger a request for Renovate to run again on this repository

Replace all logging with `loguru`

loguru is an awesome logging library. I've just incorporated in order to get easier logging level control (debug/info/critical) and centralized logging capabilities.

However, loguru has not been comprehensively applied in the project, and some print, print_verbose and print_header remain scattered across the codebase.

This issue is to track the task for replacing all logging with calls to loguru's logging functions.

Replace custom dependency updater script with PDM's `update-deps-action`

See pdm-project/update-deps-action.

Replace `BoolFlagged` with `returns.Success` or `returns.Failure`

The type BoolFlagged and its associated functionality can be replaced with returns.Result, returns.Success, returns.Failure and @returns.safe in a much cleaner way.