@chfast do you have a preference on limiting to 2^64-1 vs 2^52-1?
I usually prefer 2^63-1 over 2^64-1, but here I see little difference between 2^64-1 and 2^52-1. Therefore, 2^52-1 looks better as it makes JavaScript happy and is the smaller one.
aside from char (16-bit), all Java primitives, including long (max 2^63 - 1), are signed, which can be annoying
- Can this please be implemented in two or more competing implementations before this is published as final status?
Implementations are not required for an EIP. It is generally a good idea when building a complex EIP to actually implement it before moving to final, but that is up to author discretion.
In this case, the EIP is just adding an assertion, so an “implementation” would be one line of code like assert(x <= 2^52), so I’m not concerned that additional complexity will pop up when implementing it.