gcd() ... wrong answer, lcm() hangs

[vefatica]

This is correct.

v:\> echo %@eval[gcd(2546068882500000000 99999900000000000)]
357142500000000

If I multiply both numbers by 10, their GCD should be multiplied by 10. But this is not correct.

v:\> echo %@eval[gcd(25460688825000000000 999999000000000000)]
3584

And this puts TCC in an uninterruptible, 100% CPU usage state.

v:\> echo %@eval[lcm(25460688825000000000 999999000000000000)]

 

[vefatica]

And here's a peculiar one. The correct answer is 10**18, but I get 2**18.

v:\> echo %@eval[gcd(%@eval[10**18] %@eval[10**19])]
262144

 

[vefatica]

Note that @EVAL was necessary in my last example. The arguments to gcd() are not expanded, and, as mentioned in another thread, variable names without a leading '%' are not expanded.

 

[rconn]

WAD - you're overflowing the 64-bit integer arguments & result. I've added an overflow check to lcm in build 47 to prevent it from looping.

Those operators could (with substantial effort) be rewritten to use BCD instead, but I doubt you actually need it.

 

[vefatica]

WAD - you're overflowing the 64-bit integer arguments & result. I've added an overflow check to lcm in build 47 to prevent it from looping.

Those operators could (with substantial effort) be rewritten to use BCD instead, but I doubt you actually need it.

Actually, I needed it. How do you think I discovered it?

 

[vefatica]

LCM's overflow check apparently works. GCD can also hang TCC and would seem to need a similar check. THis will do it.

echo %@eval[gcd(1111111111118077966350999 1111111111118077966350888)]