Roadmap¶

Let’s have a look at the ideas and goals for the future versions.

This list shall evolve through time, especially as we get closer to each revision its specifications should get more precise.

0.1¶

[x] Basic functionality — Being able to run Node code from Python while getting dependencies installed automatically
[x] Not too slow boot time — Implementation of basic functionality showed that the boot time is quite high (0.5s) which is impractical for many applications. This needs to go down.
[x] Garbage collection — When Python proxies get collected, the corresponding resources on the JS side should be freed as well.

[ ] Install NPM deps through pip — Today you need to declare Node dependencies when calling Node Edge. While this works, this requires to install dependencies at runtime and also prevents you from locking the dependencies the way you would for a normal Python package. NPM packages need to be proxied for Pip/Poetry/etc.
[ ] Increased security — Right now the binding happens through a socket. It would make sens to protect the communication between Python and Node better to avoid having a local attacker “stealing” the socket.

[ ] Node binding instead of external process — The initial implementation relies on Node being already installed on the system. It would be more interesting to ship Node directly with the Node Edge package and use a C binding to call it from the same process instead of running it in an outside process.
[ ] No more context manager — We’ll drop the requirement to be used through a context manager. All should be transparent.

[ ] Async support — At this stage the requirements for an API will be well-understood. It will be the time to refactor everything and offer both a sync and an async API.

[ ] TypeScript-based typed proxies — TypeScript packages should be accessible through generated proxies with valid type annotations.

[ ] Comprehensive documentation — Any gaps in the documentation should be filled.
[ ] Stable API — The API should be stable and not change anymore.