Tooting the Rust horn in predictable ways:
Why would you want to restart a service frequently?
- Flush incorrect state
- Reset memory leaks
- ...others reasons?
Rust's borrow system makes it harder (but not impossible) to have memory leaks.
Rust's immutable-by-default semantics makes it harder (but not impossible) to end up where your long-running process'es state is incorrect.
You can still mark everything as `&mut` and litter the code with ref-counts. I've seen a lot of this "removing the safety rails because it's more convenient" among C/C++ developers who were forced into using Rust.
I think a bigger attack vector for anything dynamic, including a type-safe, memory-safe Axum back-end service, is denial-of-service because requests take too long to process. Application development is always vulnerable for being feature-driven: features sure do like resources! A service doesn't need bugs to get killed efficiently. It just needs to get slow over time.
