Pure-data: [int] resolution bug

[porres]

seems [int] can't handle values over 2ˆ24, but shouldn't it?

[Spacechild1]

the [int] object stores the number as a t_float but casts to an int and back to get rid of the fractional part. I think it should cast to a long long (the largest integer type) instead, because floats can be much larger than 2^31. actually, they can also be larger than 2^63, so the "correct" way to allow arbitrarily large floats to pass through would be using the trunc math function instead of an integer cast. on the other hand, 2^63 is already a pretty large number...

note that floating point numbers beyond 2^24 become imprecise anyway.

[Spacechild1]

BTW, did you mean to test with 2^24 (instead of 3^24)? that should work.

[porres]

BTW, did you mean to test with 2^24 (instead of 3^24)? that should work.

no, I was just showing how numbers above 2^24 get screwed up

[Spacechild1]

How does 3^24 show anything about 2^24? The former is orders of magnitude larger than the latter ;-) as I said, the current limit is really 2^31.

[porres]

as I said, the current limit is really 2^31

oh, I just thought and assumed it was 2^24 :)

[umlaeute]

closed in a0816dc2ab196f544d257db56df35a055288264e, included in update/0.50 (#508)

[Spacechild1]

@umlaeute please close :-)

[porres]

I guess is best to close when it's officially merged?

[Spacechild1]

you're probably right.

[porres]

well, it's been merged :)