alloy-rs
[Feature] Make UInt panic on overflow if overflow_checks is enabled
Component
primitives
Describe the feature you would like
Currently UInt wraps on overflow. Please, make it panic on overflow if cfg(overflow_checks) is enabled, i. e. exactly same thing rustc does with standard numbers. Here is why.
First and foremost, UInt is used mainly for representing monetary values. Wrapping behavior is unnatural for monetary values. So, UInt should panic on overflow. But still some users may want to skip checks for speed reasons. So, we should provide way to configure behavior. The most natural thing to do is to make UInt panic when standard numbers panic and wrap when standard types wrap. So, we should use cfg(overflow_checks).
Of course, this is breaking change, so it should be done in next major version.
I use alloy for managing my personal cryptocurrency, i. e. for transferring it around. Unnoticed overflows may cause loss of money, I absolutely don't want this. So, I plan to create my wrapper around U256, which will panic on overflow. And I also will use clippy::arithmetic_side_effects to make sure that I will not use arithmetic directly on ruint/alloy's U256. But ideally this should be fixed in alloy itself.
Also, currently Solidity throws exception on integer overflow by default ( https://docs.soliditylang.org/en/v0.8.28/types.html#integers ), so currently alloy is less safe than Solidity!
I already reported this problem to ruint ( https://github.com/recmo/uint/issues/408 ), but they rejected my request. So, I propose to create alloy's wrapper, which will panic on overflow (if cfg(overflow_checks)).
If you don't want to replace UInt everywhere, then at least create panicking wrapper and use it in places, where we clearly know that we are dealing with money. For example, Provider::get_balance. This method clearly returns monetary value, there is no any sense for wrapping behavior for this value (but I still suggest checking cfg(overflow_checks) for consistency with standard types)
Additional context
No response
@jenpaff , as I already said in my report, ruint rejected my request. But feel free to close this, if you want