Frequenz repository setup tools and common configuration for Python
Home Page: https://frequenz-floss.github.io/frequenz-repo-config-python/
License: MIT License
![dependabot[bot] avatar](https://avatars.githubusercontent.com/in/29110?v=4 "dependabot[bot]")
From https://github.com/frequenz-floss/frequenz-api-microgrid/blob/v0.x.x/CONTRIBUTING.md:
Upgrading dependencies
======================
If you want to update the dependency `frequenz-api-common`, then you need to
* update the submodule `frequenz-api-common`
* update the version of the `frequenz-api-common` package in `pyproject.toml`
* update the version of the `frequenz-api-common` package in
`minimum-requirements-ci.txt`
The version of `frequenz-api-common` used in all the three places mentioned
above should be the same.
Here is an example of upgrading the `frequenz-api-common` dependency to version
`v0.2.0`:
```sh
ver="0.2.0"
cd submodules/frequenz-api-common
git remote update
git checkout v${ver}
cd ../..
sed s/"frequenz-api-common == [0-9]\.[0-9]\.[0-9]"/"frequenz-api-common == ${ver}"/g -i pyproject.toml
sed s/"frequenz-api-common == .*"/"frequenz-api-common == ${ver}"/g -i minimum-requirements-ci.txt
```This could be automated by adding some command like frequenz-repo-config update-submodule frequenz-api-common v0.5.1
We need to remind contributors that pull request need release notes updates.
Use the workflow used in the SDK to check this:
No response
No response
We need to make sure to document that a label needs to be created to manually skip checking for release notes update when it is not necessary.
given the name fcr-balancing, the folder src/frequenz/actor/fcr-balancing was created,
but I believe we want to use _ here, at least my existing folder was fcr_balancing (migrating project).
No response
Affects only the configuration of actor repos (part:actor-only)
If this works as intended, feel free to close.
darglint is currently abandoned (the repo is archived).
There is the intention to implement darglint rules in ruff, so when that happens we should replace darglint with ruff.
There is also a python-based alternative that runs much faster, but it doesn't support Google-style comments yet (but it is also planned jsh9/pydoclint#3): https://github.com/jsh9/pydoclint
SDK example: https://github.com/frequenz-floss/frequenz-sdk-python/blob/v0.x.x/.github/RELEASE_NOTES.template.md
Probably we should also find a way to share this RELEASE_NOTES template to the projects using this library, so we have it only in one place. Worst case scenario it can be provided in the Python package and add a binary that will just reset the release notes using this template.
Reported by @cwasicki in https://github.com/frequenz-io/frequenz-actor-forecast/pull/4#issuecomment-1607848660
We are using ${{ env.DEFAULT_PYTHON_VERSION }} and other env vars when declaring the matrix but that's unfortunately not supported.
protolint is a tool to lint protocol buffer files.
We should add a nox session that runs protolint. One problem with protolint is that is not very popular yet, and it doesn't have an easy to install package in all distributions and also no pypi package for it. Because of this, the nox session should not be included in the default sessions, but it should be explicitly used when running the CI.
Here is how protolint is currently being used in the CI:
jobs:
protolint:
runs-on: ubuntu-20.04
steps:
- name: Fetch sources
uses: actions/checkout@v3
with:
submodules: true
- name: Get protolint
run: |
# checksum taken from: https://github.com/yoheimuta/protolint/releases/download/v0.38.1/checksums.txt
echo "c1a43f352da6f96f3e60efe33438eaca6148526e6145d9bbd77c40cebc9895a6 protolint_0.38.1_Linux_x86_64.tar.gz" > checksum.txt
curl -LO https://github.com/yoheimuta/protolint/releases/download/v0.38.1/protolint_0.38.1_Linux_x86_64.tar.gz \
&& sha256sum --check checksum.txt \
&& tar xf protolint_0.38.1_Linux_x86_64.tar.gz \
&& chmod +x protolint \
&& sudo mv protolint /usr/local/bin/protolint
- name: Run protolint
run: protolint lint protoThere are also a few alternative GitHub actions to use out of the box:
After running cookiecutter for a new actor, the suggested example printed after cookiecutter setup
cd frequenz-actor-forecast
python3.11 -m venv .venv
. .venv/bin/activate
pip install .[dev-noxfile]
nox
fails with error
Traceback (most recent call last):
File "/home/christoph/fqz/frequenz-actor-forecast/.venv/bin/nox", line 8, in <module>
sys.exit(main())
^^^^^^
File "/home/christoph/fqz/frequenz-actor-forecast/.venv/lib/python3.11/site-packages/nox/__main__.py", line 47, in main
exit_code = workflow.execute(
^^^^^^^^^^^^^^^^^
File "/home/christoph/fqz/frequenz-actor-forecast/.venv/lib/python3.11/site-packages/nox/workflow.py", line 51, in execute
return_value = function_(*args, global_config=global_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/christoph/fqz/frequenz-actor-forecast/.venv/lib/python3.11/site-packages/nox/tasks.py", line 113, in load_nox_module
return _load_and_exec_nox_module(global_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/christoph/fqz/frequenz-actor-forecast/.venv/lib/python3.11/site-packages/nox/tasks.py", line 63, in _load_and_exec_nox_module
loader.exec_module(module)
File "<frozen importlib._bootstrap_external>", line 940, in exec_module
File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
File "/home/christoph/fqz/frequenz-actor-forecast/noxfile.py", line 7, in <module>
from frequenz.repo.config import nox
File "/home/christoph/fqz/frequenz-actor-forecast/.venv/lib/python3.11/site-packages/frequenz/repo/config/__init__.py", line 257, in <module>
from . import nox, setuptools
File "/home/christoph/fqz/frequenz-actor-forecast/.venv/lib/python3.11/site-packages/frequenz/repo/config/setuptools/__init__.py", line 6, in <module>
from . import grpc_tools
File "/home/christoph/fqz/frequenz-actor-forecast/.venv/lib/python3.11/site-packages/frequenz/repo/config/setuptools/grpc_tools.py", line 19, in <module>
import setuptools.command.build as _build_command
ModuleNotFoundError: No module named 'setuptools.command.build'
The setuptools version that is used is in my case 59.6.0.
After adding the required version to the dependencies section the example works:
diff --git a/pyproject.toml b/pyproject.toml
index b056a9c..2fc035e 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -25,6 +25,7 @@ classifiers = [ # TODO(cookiecutter): Remove and add more if appropriate
]
requires-python = ">= 3.11, < 4"
dependencies = [ # TODO(cookiecutter): Remove and add more if appropriate
+ "setuptools == 67.7.2",
"typing-extensions == 4.5.0",
"frequenz-sdk == 0.20.0",
]
The license MIT is hardcoded here:
and here:
I expected it to be "Proprietary" as specified while creating the project.
No response
Cookiecutter template (part:cookiecutter)
No response
There were 2 new releases of the SDK since the cookiecutter template was created, we should encourage new projects to use the latest SDK.
Bump the SDK version in the cookiecutter template.
No response
No response
No response
Coverage reporting can be added via pytest-cov and then visualized via codecov.
In the pyproject.toml file we should add to the dependencies:
dev-pytest = [
"pytest == 7.3.1",
"pytest-cov == 4.0.0",
"...",
]Then basically just run pytest --cov python.package.to.test in the CI and use the codecov/codecov-action to upload the reports.
When a new repository is created, we should optionally allow the user to:
While trying to update my actor project, I got this error:
❰marenz❙~/frequenz/fcr(git≠sdk-update)❱✘≻ cookiecutter gh:frequenz-floss/frequenz-repo-config-python --directory=cookiecutter --replay --replay-file .cookiecutter-replay.json --overwrite-if-exists
You've downloaded /home/marenz/.cookiecutters/frequenz-repo-config-python before. Is it okay to delete and re-download it? [yes]:
Traceback (most recent call last):
File "/usr/bin/cookiecutter", line 33, in <module>
sys.exit(load_entry_point('cookiecutter==2.1.1', 'console_scripts', 'cookiecutter')())
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/click/core.py", line 1130, in __call__
return self.main(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/click/core.py", line 1055, in main
rv = self.invoke(ctx)
^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/click/core.py", line 1404, in invoke
return ctx.invoke(self.callback, **ctx.params)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/click/core.py", line 760, in invoke
return __callback(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/cookiecutter/cli.py", line 194, in main
cookiecutter(
File "/usr/lib/python3.11/site-packages/cookiecutter/main.py", line 114, in cookiecutter
result = generate_files(
^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/cookiecutter/generate.py", line 373, in generate_files
generate_file(
File "/usr/lib/python3.11/site-packages/cookiecutter/generate.py", line 173, in generate_file
tmpl = env.get_template(infile_fwd_slashes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/jinja2/environment.py", line 1010, in get_template
return self._load_template(name, globals)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/jinja2/environment.py", line 969, in _load_template
template = self.loader.load(self, name, self.make_globals(globals))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/jinja2/loaders.py", line 138, in load
code = environment.compile(source, name, filename)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/site-packages/jinja2/environment.py", line 768, in compile
self.handle_exception(source=source_hint)
File "/usr/lib/python3.11/site-packages/jinja2/environment.py", line 936, in handle_exception
raise rewrite_traceback_stack(source=source)
File ".github/labeler.yml", line 13, in template
- "**/*.md"
^^^^^^^^^^^^^^
jinja2.exceptions.TemplateAssertionError: No filter named 'as_identifier'.
File ".github/labeler.yml", line 13
# - "src/frequenz/{{cookiecutter.type}}/{{cookiecutter.name | as_identifier}}/module/**"
no error.
No response
Cookiecutter template (part:cookiecutter)
No response
Examples in __init__.py files are not being linted, these files are completely skipped.
Examples in __init__.py files should be checked.
This happens because we are explicitly excluding those files because Sybil fails to import __init__.py files, raising an error like:
self = <sybil.document.PythonDocStringDocument object at 0x7f36e8644390>
example = <Example path=/home/luca/devel/repo-config/src/frequenz/repo/config/nox/__init__.py line=12 column=1 using <bound meth...eBlockParser.evaluate of <frequenz.repo.config.pytest.examples._CustomPythonCodeBlockParser object at 0x7f36e8db7b50>>>
def evaluator(self, example: Example) -> Optional[str]:
"""
Imports the document's source file as a Python module when the first
:class:`~sybil.example.Example` from it is evaluated.
"""
> module = import_path(Path(self.path))
.nox/pytest_min/lib/python3.11/site-packages/sybil/document.py:153:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
path = PosixPath('/home/luca/devel/repo-config/src/frequenz/repo/config/nox/__init__.py')
def import_path(path: Path):
container = path
while True:
container = container.parent
if not (container / INIT_FILE).exists():
break
relative = path.relative_to(container)
if relative.name == INIT_FILE:
parts = tuple(relative.parts)[:-1]
else:
parts = tuple(relative.parts)[:-1]+(relative.stem,)
module = '.'.join(parts)
try:
return importlib.import_module(module)
except ImportError as e:
> raise ImportError(
f'{module!r} not importable from {path} as:\n{type(e).__name__}: {e}'
) from None
E ImportError: 'config.nox' not importable from /home/luca/devel/repo-config/src/frequenz/repo/config/nox/__init__.py as:
E ModuleNotFoundError: No module named 'config'
.nox/pytest_min/lib/python3.11/site-packages/sybil/python.py:40: ImportError
We need to investigate this and open an issue in Sybil if appropriate.
bug.yml
part field options should match all the part:xxx labels defined in the repoconfig.yml:
The generated docs use the chronologically latest gitub release, not the numerical latest.
This also affect GitHub releases.
I expected to see the highest stable version number as the default release
No response
Documentation (part:docs)
No response
SDK Example: https://github.com/frequenz-floss/frequenz-sdk-python/blob/v0.x.x/CONTRIBUTING.md
This also should live in a common repository, but I'm not sure if this should be the place, maybe we need a repository that is more about documentation and general policies. This repo seems too low level for this file.
Now that we have the new https://github.com/frequenz-floss/frequenz-api-common/ repository, we should add it as a submodule and a python dependency (and include paths for building proto files) to the api repositories.
We need to make sure all markdown files have a consistent style.
Use https://github.com/markdownlint/markdownlint to lint markdown files.
No response
No response
No response
EditorConfig is an easy way to configure basic styling (like indentation style, whitespace trimming and end of line style) for different file types that is supported by most modern editors.
We should include max_line_length (undocumented, so we should try if it works editorconfig/editorconfig#280) to 88 for Python and 72 for commit messages for example. More on this property: https://github.com/editorconfig/editorconfig/wiki/Property-research:-maximum-line-length
Example:
# Each line is a file pattern followed by one or more owners.
# Owners will be requested for review when someone opens a pull request.
# Fallback owner.
# These are the default owners for everything in the repo, unless a later match
# takes precedence.
* @frequenz-floss/SOME-team
The example we are including in src/frequenz/{type}/{name}/__init__.py is not great, specially for apps, where we can have an example that is actually some sort of Hello World app using the SDK.
The same as we check and format python files automatically, it would be good to do the same with pyproject.toml (and any other toml file) to make sure we have a consistent formatting across projects and can detect errors early.
taplo is a nice tool written in rust that can do both (and even check a toml semantically against a json schema).
Generate all the boilerplate for new projects using Cookiecutter.
pyupgrade can automatically upgrade code to newer python syntax.
We should add at least unit tests, but it would be also nice to add integration tests, include some example repository and making sure things work as expected.
The example we are including in src/frequenz/{type}/{name}/__init__.py is not great, specially for actors, where we can have an example that is actually some sort of Hello World actor using the SDK.
Once the workflows are configured, we should configure branch protection to required them.
Add a .github/workflow/labeler.yml workflow:
name: Pull Request Labeler
# XXX: !!! SECURITY WARNING !!!
# pull_request_target has write access to the repo, and can read secrets. We
# need to audit any external actions executed in this workflow and make sure no
# checked out code is run (not even installing dependencies, as installing
# dependencies usually can execute pre/post-install scripts). We should also
# only use hashes to pick the action to execute (instead of tags or branches).
# For more details read:
# https://securitylab.github.com/research/github-actions-preventing-pwn-requests/
on: [pull_request_target]
jobs:
Label:
permissions:
contents: read
pull-requests: write
runs-on: ubuntu-latest
steps:
- name: Labeler
# Only use hashes, see the security comment above
uses: actions/labeler@e54e5b338fbd6e6cdb5d60f51c22335fc57c401e # 4.0.1
with:
repo-token: "${{ secrets.GITHUB_TOKEN }}"Configure the workflow creating a .github/labeler.yml.
For example:
# Configuration for the Labeler GitHub action, executed by
# .github/workflows/labeler.yml.
#
# The basic syntax is [label]: [path patterns].
#
# For more details on the configuration please see:
# https://github.com/marketplace/actions/labeler
"part:docs":
- "**/*.md"
- LICENSE
"part:tests":
- "tests/**"
"part:tooling":
- ".git*"
- ".git*/**"
- "**/*.toml"
- "**/*.ini"
- CODEOWNERS
- MANIFEST.in
- "*requirements*.txt"
- setup.py
- setup.cfg
"part:BLAH":
- "src/frequenz/BLAH/**"part:xxx labels defined in the repo are present.SDK Example:
Use seedir to print the directory structure instead of tree after the project was generated, as it is a Python package, so we don't depend on the users to have an external tool to show the tree.
Add [https://github.com/ZeWaka/KeywordLabeler] app configuration in .github/keylabeler.yml.
For example:
# KeywordLabeler app configuration. For more information check:
# https://github.com/ZeWaka/KeywordLabeler#readme
# Determines if we search the title (optional). Defaults to true.
matchTitle: true
# Determines if we search the body (optional). Defaults to true.
matchBody: true
# Determines if label matching is case sensitive (optional). Defaults to true.
caseSensitive: true
# Explicit keyword mappings to labels. Form of match:label. Required.
labelMappings:
"part:docs": "part:docs"
"part:tests": "part:tests"
"part:tooling": "part:tooling"
"part:❓": "part:❓"part:xxx labels defined in the repo are present.labelMappingsSDK example: https://github.com/frequenz-floss/frequenz-sdk-python/blob/v0.x.x/.github/keylabeler.yml
pylint's unsubscriptable check is unreliable, having some false positives. For example:
"""Test pylint."""
class AClass:
"""Test pylint."""
a_map: dict[str, int] | None = None
def __str__(self) -> str:
"""Test pylint."""
return "None" if not self.a_map else self.a_map["a"]
class BClass(AClass):
"""Test pylint."""
a_map = {"a": 1}This produces an error E1136: Value 'a_map' is unsubscriptable (unsubscriptable-object) when using self.a_map["a"], which is wrong.
This check is also redundant, as it is a type check, which is already done (better) by mypy.
We should add this to the cookiecutter pyproject.toml template:
[tool.pylint.messages_control]
disable = [
# ...
"unsubscriptable-object",
]The SDK is already using this ignore:
A very easy way to make sure templates don't introduce regressions is to do golden testing.
This basically consists in testing the generated files manually and once they are verified to work well, store them as the "golden copy". Golden test consist in just comparing newly generated files with the golden copies.
This is an easy way to test complex outputs (like in this case, where testing involves running github actions, checking if owners are assigned to PRs, etc.) and making sure regressions are not introduced in refactoring for example.
There are 2 main golden tests libraries in Python:
Both compare similarly in terms of usage and sustainability according to Snyk, so we should compare the features and general feel to select one.
Update: It looks like pytest-goldie might be a fake. The source is not available in pypi, the home page gives a 404.
When using the setuptools.grpc tools, the user needs to specify recursive-include py *.pyi in the MANIFEST.in file, otherwise setuptools[scm] doesn't seem to realize that the files should be included (not sure why it does realize the .py files should).
It is supposed to be a way to include them automatically if the sub-command build_py reports the new *.pyi files:
This is probably also needed to support editable installs.
We need to make sure the dependencies we use have a compatible license. We should try fossa.
Example badge: https://github.com/codecov/example-python#license
We also should include any utilities needed to generate the docs by other projects.
SDK example:
API (protobuf) docs generation example:
This will generate the docs as a markdown file and put it in the GitHub job step summary, but we can put it in the mkdocs website instead.
gen-docs:
runs-on: ubuntu-20.04
steps:
- name: Fetch sources
uses: actions/checkout@v3
with:
submodules: true
- name: Generate documentation
run: |
docker run --rm -v $PWD:$PWD pseudomuto/protoc-gen-doc \
-I $PWD/proto \
-I $PWD/submodules/api-common-protos \
-I $PWD/submodules/frequenz-api-common/proto \
--doc_opt=markdown,gen-docs.md --doc_out=$PWD \
$(find $PWD/proto -name *.proto)
- name: Write summary
run: cat gen-docs.md > $GITHUB_STEP_SUMMARYRight now we are using a custom command to do a custom download of the protolint tool:
jobs:
protolint:
runs-on: ubuntu-20.04
steps:
- name: Fetch sources
uses: actions/checkout@v3
with:
submodules: true
- name: Get protolint
run: |
# checksum taken from: https://github.com/yoheimuta/protolint/releases/download/v0.44.0/checksums.txt
echo "33627c1fd4392edc9363b414651f60692286c27e54424fc535ebb373a47a3004 protolint_0.44.0_Linux_x86_64.tar.gz" > checksum.txt
curl -LO https://github.com/yoheimuta/protolint/releases/download/v0.44.0/protolint_0.44.0_Linux_x86_64.tar.gz \
&& sha256sum --check checksum.txt \
&& tar xf protolint_0.44.0_Linux_x86_64.tar.gz \
&& chmod +x protolint \
&& sudo mv protolint /usr/local/bin/protolint
- name: Run protolint
run: protolint lint protoThere are 2 main problems with this approach:
There are (at least) 2 github actions that we can look at to see if they solve these problems more nicely:
If these are not good enough, there is also a pypi package for protolint that could be used at least to get automatic updates:
Instead of having the workflow as a cookiecutter tempate, this repository could provide a reusable workflow that is used by the cookiecutter template.
When creating a new project using cookiecutter, it is not obvious at all what some variables are used for, and what should be the input format, etc.
Ideally show a help message before the variables are prompted (it might be doable via the pre build hook), but also show this information in the documentation website. We can create one markdown formatted file with this information, and display it in the hook and embed the same file (as a snippet) in the documentation so there is only one source of truth.
No response
No response
No response
api projects use Git submodules to manage dependencies for protobuf files.
We use dependabot to keep dependencies up to date and while dependabot does support git submdules as a dependency management system, it support is pretty basic, and submodules are always updated to the lastest default head and there is no way to configure it to update at least to tags, much less to configure updates following semver.
The current alternative (as some people did in that issue) is to implement our own workflow to check this (or send a PR to dependabot to improve support).
The most basic approach would be to schedule a workflow to run periodically that could just be looking for new tags in submodules (possibly following semver) and just failing the job if there are updates, so at least we get notified when something need updating. The update process would still be manual.
Please note that this is only needed (for now) for the google common proto files, as for internal dependencies we'll still get notified about new releases by the Python ecosystem via dependabot.
Many users reported problems while trying to generate a new project. The main mistakes were trying to modify the cookiecutter command, which should be typed literally: cookiecutter gh:frequenz-floss/frequenz-repo-config-python --directory=cookiecutter.
gh:frequenz-floss/my-project--directory=my-projectWe should make it more clear that this command needs to be typed as is, and all information about the project to be generated will be prompted.
No response
No response
No response
Projects normally have examples in the examples/ directory and in docstring documentation. The ones in examples/ are currently being linted, but never ran. The ones in docstrings are never linted nor ran. All of this very subject to break when there are updates.
In docstrings there are mainly 2 types of examples in documentation:
>>> in a docstringdocs/.We mainly use markdown examples, so we should focus on that first.
There are several tools to actually test those examples to make sure they work, so we should use one:
The example we are including in src/frequenz/{type}/{name}/__init__.py is not great, specially for models, where we can have an example that is actually some sort of Hello World model using the SDK.
We need tests to be fast.
Don't use nox anymore. Instead run each tool individually or ship our own tool to run all the linting and tests if necessary (for example: frequenz-repo-config lint and frequenz-repo-config test)
All projects, but specially this one, that in integration tests need to run nox for 5 different generated projects on top of running nox for itself. All of that is also done twice, once for the minimum supported dependencies and one for the currently available dependencies. So it adds up.
pre-commit might be an alternative to have a standarized way to invoke the tools and which tools should be used to test changes before creating a PR. We don't necessarily need to plug the pre-commit hook, pre-commit can also be invoked manually, but we take advantage of the well-known and adopted tool.
Tests for all repos are very slow with nox, because it needs to create a new virtual environment for each session and inside of it, install all dependencies over and over again. Right now we have 5 sessions and we need to add more (pyupgrade, protolint, etc.).
Also nox for our use case brings practically zero advantages. nox is supposed to be useful to test with multiple Python or libraries versions, where having a separate virtual environment is really needed, but we don't use that and we rely on the CI to test with different Python versions and we have a separate session for the min/max dependencies with tests.
Also for the CI, there is absolutely no need to create a venv at all, as it starts with a clean environment where we can pip install directly, adding more unnecessary extra time to the mix.
The test for this repository are currently taking ~50 minutes to complete 🐢
Add a CI workflow template.
While updating the .github/labeler.yml file I found myself needing to specify a pattern that excludes a path.
Adding an example to the existing two in the file that shows negative matching. e.g. taken from their website:
# Add 'frontend` label to any change to *.js files as long as the `main.js` hasn't changed
frontend:
- any: ['src/**/*.js']
all: ['!src/main.js']
No response
No response
No response
We can use the package pylint-google-style-guide-imports-enforcing to make pylint check the import style, as we want to follow that style but we hardly do because there is no enforcement.
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.