Re: 64-bit division anomolies (Solved.)

[syphilis]

The latest I've tried this on is 5.18, so it might have been fixed in later builds

It hasn't been - and it goes back to at least 5.16, and it affects 32-bit perls in the same way for unsigned values greater than 2 ** 31.

I couldn't find a way to successfully perform integer division/modulation with these large unsigned values without resorting to XS (as you've done).
(Update: Obviously, I've ignored Math::BigInt.)
However, I'm sure the integer pragma is supposed to cater for this and I couldn't find anything in the docs to suggest otherwise.

So I've submitted a bug report against integer for this - in case there's someone who cares enough to fix it.
(I hit "Send" before thinking to reference the original PM post in that report ... sorry 'bout that.)

Cheers,
Rob

[syphilis]

So I've submitted a bug report against integer for this - in case there's someone who cares enough to fix it.

Turns out that the integer pragma only works with integer values in the range -(2**63) .. (2**63-1).
On 32-bit perls that range becomes -(2**31) .. (2**31-1).
Yes - it's that crappy, and I doubt that anyone cares.

It's all there in the integer docs if one reads that documentation properly - which I didn't do.

Wiping egg off face (... yet again),
Rob

[salva]

But still the unnecessary conversion to float when dividing two unsigned integers looks like a bug to me.

BTW, the multiplication is also affected.

[BrowserUk]

So I've submitted a bug report against integer for this - in case there's someone who cares enough to fix it. (I hit "Send" before thinking to reference the original PM post in that report ... sorry 'bout that.)

NP. It's more likely to get taken seriously with (only) your name attached.