"Dmitry Zamotkin" <
[hidden email]> wrote in message
news:
[hidden email]...
> And does not crash with an intermediate #asFloat message:
>
> r := RandomCRT new.
> 100000 timesRepeat: [ (r next negated asFraction/10) asFloat asInteger
> ]
>
This appears to be a bug in the VM large integer division primitive. You can
reproduce it more simply as follows:
u := -2**63-1.
v := 2**96-1.
u quo: v. "Bang!"
We'll fix this in the forthcoming release.
Regards
Blair