Open deven96 opened 2 months ago
It wouldn't be practical or possible to make every datastructure we use fallible, so how about we wrap critical client parts in an std::panic::catch_unwind
to ensure that if we ever do massive allocations, we are able to report errors from there back to the client
Vectors, Hashmaps, e.t.c all implicitly fail whenever allocations fail. We are already capping the global allocations using
cap
, fallible collections would then ensure we don't panic but rather surface the allocation errors