Purge all references to the int64_t type. about benchmark HOT 7 CLOSED

bytes and items processed should remain (u)int64_t. the rest can go, especially taking into account the overhead of 64-bit math on 32-bit systems.

from benchmark.

@EricWF I'd like to take this up. Could you give a little more info on how to go about it?

from benchmark.

Honestly last time I checked the status of this bug I used grep -R int64_t.

from benchmark.

So just swap out int for all occurrences of int64_t?

from benchmark.

Nope, some of these should be size_t. Also, int is typically 32 bits, so you must make sure that you are not introducing overflows.

This is a tricky change, it's certainly not a global substitute.

from benchmark.

Exactly what @pleroy said. The remainder of the changes are tricky, most long hanging fruit is gone.

from benchmark.

At some point we started replacing size_t with int64_t explicitly, and I think we're in a position where everything is correct, and we shouldn't monkey around with it further.

If you agree, @EricWF, please close this.

from benchmark.