There's not really any work on this yet, but it shouldn't be too hard! I whipped up this branch for converting the parser, compiler etc to no_std, but I imagine the vm would be the main hurdle.

Looks like bincode would be the main blocker: bincode-org/bincode#265

This item is still a stretch goal, but I would love to see this happen!

Things to note / to be done:

• Be able to use rustpython in a no_std env (like @coolreader18 mentioned)
• Performance / code size: we need to reduce code size and improve performance overall. This has many benefits: faster runtime on desktop use, faster usage on microcontrollers, less power usage, less memory usage. Good thing all the way!
• The standard library. This boils down to being able to run rustpython with or without the standard library, and as such being able to run it on a microcontroller where there might be a flash memory restriction on amount of python files you can embed.
• The VM is where most of the action happens, and is I guess the hardest part to get going. I would start by porting the parser crate to no_std and see if this can be used on a microcontroller. Next up would be the compiler, and final step (and most difficult) the vm crate. I have no clue about code size at this point, but I'm guessing the vm is using the biggest amount of memory.

Thank you for starting the no_std development again! My team is building a system which uses a RISC-V microcontroller, with a Tock kernel and a hardware accelerator, for accelerating machine learning applications. For now, we might proceed with Rust applications, but our end goal is to support Python applications. Hopefully, RustPython can help us achieve that goal.

Any progress towards RustPython supporting no_std would be amazing tbh!

@coolreader18 is there anything I can do to help out?

The bincode blocker is unfortunate but I'm not sure where its needed (I'm not familiar with the RP codebase at all) there are other alternatives that support no_std for instance capnproto but again I'm not sure how viable they are to replace bincode since I dont know how or what bincode is being used for...

@coolreader18 regarding bincode being the main blocker, it seems that there is a repository in the same organization for a no_std-compatible bincode: bincode-org/bincode-core. Could this be a viable solution?

Possibly, though I don't think bincode-core is published yet. I think my current plan would be to rewrite rustpython-bytecode to use deku instead of bincode; it's moreso designed for custom binary formats, and might be easier to keep stable.

bincode 2 should support no_std by disabling the default features or disabling the std feature

@leapofazzam123 what is the current relationship between pravega/bincode2 and bincode-org/bincode? At a glance, the Cargo.toml in bincode-org/bincode seems to indeed support a no_std configuration.

It looks like bincode-v2 has been merged back to bincode-org but it's still in beta? That's what I glean.