Masked Authenticated Messaging (MAM) Support

[todofixthis]

Implement support for Masked Authenticated Messages (refer to https://github.com/iotaledger/mam.client.js/ for a reference implementation).

[todofixthis]

Moving back to "scheduled"; we're not ready to work on this just yet. For now, we will implement a PoC that uses the JS lib to do the crypto stuff, but that's not part of this GitHub issue.

[miili]

We should tackle this soon, wrapping the MAM Rust library (https://github.com/iotaledger/MAM)!

There are two options:

• https://github.com/PyO3/pyo3
• https://github.com/dgrunwald/rust-cpython

The two libraries are discussed here PyO3/pyo3#55
PyO3's approach seems to be more native and flexible, further it is said to be 15% faster

@todofixthis Apparenltly there has been done some work on the foundation https://github.com/iotaledger/iota-bindings with PyO3... how do the bindings come into play?

[todofixthis]

Awesome news @miili !

I think for PyOTA we should focus on a pure-Python implementation (with optional C extension if performance is an issue).

A solution that relies on Rust bindings sounds interesting (and I'd love to see it!), but I think it would be better for the long-term direction of PyOTA if that was implemented in a separate project.

[miili]

Should MAM for Python come in a separate package or be included in PyOTA?

[todofixthis]

My preference is to follow any precedent set by the Javascript library. If MAM gets included directly in iota.lib.js, then I think PyOTA should follow suit. If not, well, I don't think PyOTA needs to be monolithic anyway (:

If we end up with MAM in a separate project, though, I would still be happy to include it as an extra in PyOTA's setup.py (e.g., so that developers can install it via pip install pyota[mam]).

[miili]

Looks like the Javascript library will be wrapped around the Rust MAM library, through WebAssembler. See https://github.com/iotaledger/MAM and https://github.com/iotaledger/iota-bindings for Python prep work.

If we implement the Rust realisation as well, we can rely on a solid and well tested library. There will be one point of responsibility for many implementations.
I hope to have spare time in February and would like to takle this @HerrMuellerluedenscheid ;)