Skip to content

Latest commit

 

History

History
78 lines (53 loc) · 2.25 KB

CONTRIBUTING.md

File metadata and controls

78 lines (53 loc) · 2.25 KB
Raw

Contributors Guide

  • We support Python 3.5 to 3.10.

Setting up the dev environment

Just create a virtual environment and install the requirements.

python3 -m virtualenv venv
. venv/bin/activate
pip install -r requirements.txt

Testing

Run all tests:

python3 -m unittest discover

To run a specific test:

python3 -m unittest tests.integration.test_session.TestApiSession.test_good_response_is_parsed

For the integration tests, you'll need some environment variables (details).

Code standards

  • We lint the whole code using Flake8.

  • We format the whole code using Black.

  • We keep test coverage level high.

  • We prefer integration tests for the API clients and session classes.

  • We keep one class per file.

Architecture overview

The Python SDK exposes different "Api Client" classes, depending on the main resource in question (e.g. customer, usage, product plan, etc...).

These classes are simple wrappers around the requests.Session class.

API call errors are detected and turned into exceptions (ApiError).

High Throughput Ingestion

In order to allow high ingestion throughput without imposing significant latency on the caller, the SDK provides an ingestion client based on the producer-consumer pattern with an in-memory queue.

ingest.ThreadedProducer can be instantiated as a global variable and its send method can be called by multiple threads.

The producer instance will manage a a configurable amount of background threads (ingest.ThreadedConsumer). These will consume the items from the queue in batches and send them to Amberflo.

Making a new release

We follow semantic versioning.

Releasing a new version is mostly automated by a Github action. It does require a few manual steps:

  1. Update VERSION in metering/version.py to the new version.
  2. Add an entry to HISTORY.md describing what's new.
  3. Commit the changes
git commit -m "Release X.Y.Z."
  1. Create a new tag
git tag "vX.Y.Z"
  1. Push and then create a release in Github. Once you publish it, the Github action will publish the package to PyPI.