underflow

Despite the fact that signed overflow/underflow is undefined behavior I'm pretty sure that many more bugs have resulted from unsigned underflow than signed overflow or underflow.

No matter if you use signed or unsigned types, be sure you handle both underflow and overflow.

Its primary goal is to follow the Go spec correctly, including simulating overflow/underflow on int64 type.

Having to check for overflow/underflow after every addition is what is impractical in my book.

Since the computer can't be infinite, it will inevitably fail with overflow, underflow, and such.

When they're comparing Really Small Numbers multiplied together, to avoid underflow, they instead calculate the logs of the pieces and add them.

With signed integers it is much more difficult to reach those overflow/underflow limits.

ETA: Also, you mentioned rounding, but there is also exponent overflow and underflow.

The problem here is that unsigned arithmetic, though much easier to hit underflow, is fully defined in C where signed {under,over}flow is UB.

That way I'd get precise rationals (including int64s), immunity from underflow and overflow and a smoother error curve than floats.

Signed integers have undefined overflow, but unsigned integers underflow really easily.

The first two issues (JIT code compilation and integer overflow/underflow) add up to a significant fraction of the crashing bugs in Gecko that have security exposure.

Fascinating; you can see the original "nuke-happy Gandhi" bug underflow first-hand, and the world maps were really just planar bitmaps, so when you'd figured out how they were stored, it wasn't hard to knock up an editor.

Don't pull the number of buffers you have out of thin air ("Oh, six is enough, maybe twelve, don't want to run out in my layer of the system, after all...") Don't assume your local underflow or overflow recovery strategy actually works in the whole system.