I maintain the Home Assistant integration for BLUETTI power stations (bluetti/bluetti-home-assistant), built on an independent library, pybluetti/pybluetti. The integration recently reached Platinum tier on the quality scale (async-dependency, inject-websession, mypy --strict), 100% test coverage, ruff clean. Goal: get to a state where a core submission is realistic.
I’m now discussing next steps with BLUETTI (the manufacturer), and they’ve proposed:
Official repo stays BLUETTI-recognized, but deliberately slower-paced (internal audit/release constraints).
A separate community fork would get the active development — pending PRs, and likely the Platinum-tier work itself.
BLUETTI links to the community repo from their official page, and selectively backports features over time.
My concern: this inverts what I’d expect core to want — the repo most likely to meet the technical bar (tests, typing, quality tier) is the community one, while the repo BLUETTI actually recognizes stays intentionally behind.
Has anyone been through this with a manufacturer-backed integration? Specifically:
Does this kind of split rule out core consideration, or has it worked before?
Does core expect formal manufacturer endorsement of the specific repo submitted, or does technical quality + active maintenance carry more weight?
Any precedent of a “community” fork becoming the de facto path to core?
Happy to share more detail. Trying to settle the right structure before more work goes into either repo.
The vendor provides a robust API, well documented. Hobbyists pick it up, and develop solutions the vendor has not considered. Product range and capability are expanded, benefiting both.
Sales and reputation soar. The vendor profits improve.
Expert users evaluate prototypes, providing valuable real world feedback. The released products are robust, requiring little support. The enthusiasts provide viral advertising that money cannot buy, hitting the target audience precisely. Cost savings on advertising, marketing and support are huge.
See what happened with Arduino, originally a small science project, or Espressif with their Chinese ESP8266 SOC, Android, a flavor of Unux adopted by Google, or even the Raspberry Pi, that had examples and documentation correctly translated into readable English the hobby market could enthusiastically embrace. Where are they now? Embedded widely, the professionals also using that pool of knowledge and experience.
That other vendors do not follow this path is probably due to inertia. IP and trade secrets vs Open Source. The lawyers and accountants are aghast. In this fast paced world of ongoing development, you just have to be in front and leave everybody else chasing. Whatever you can do to achieve this, including harnessing Open Source and a huge database of user input - go for it!
Show this post to the CEO, not the CFO. They understand market forces and profits. Money doesn’t talk, it shouts!