Thread by @_benjaminowen

My first #NDCOslo session is about to start - @KodrAus will be speaking about combining C# and Rust

All languages have trade-off, and different components of a system have different requirements.

Rust offers high performance. C# is a more mature ecosystem which allows developers to be productive

Both C# and Rust are safe languages, but we need an unsafe layer in the middle to interop

High performance C# requires avoiding garbage collections - i.e. using value types wherever possible

High performance, value-type based C# increasingly looks like Rust

Rust references track lifetime and mutability.

C# manages lifetime of struct by only allocating on the stack.

Rust ownership rules help avoid race conditions that C# would not prevent

C# is much faster to compile

The Rust / C# interop layer is essentially defined as a C ABI

P/Invoke marks entry and return points to/from unmanaged code. This affects GC behaviour

P/Invoke is not free, and in some cases may cost more than the unmanaged code that we're calling

Not all concepts translate directly between languages, e.g Rust errors ! = C# exceptions

Rust references and C# managed pointers are not compatible. We need to interop using unmanaged pointers