​Hi all,

We've been looking at a seemingly simple bug in [nbx], but we can't quite figure it out. This bug affects our [number~] abstraction in cyclone.

Prior to pd 0.47, NaNs did not display in either the number atom box or [nbx]; this was fixed for atom boxes in 0.47, but not [nbx]. [nbx] also does not pass nan from its output while the atom box does. See https://sourceforge.net/p/pure-data/bugs/1251/

We thought that this behavior was probably introduced here:

https://github.com/pure-data/pure-data/commit/cd737749b51e03ba1bdc6d8b66b890d89823d158

The comparison in this code would filter nans.

We tried a simple revert but it didn't seem to work, so something else must be going on that we couldn't find.

While we're at it, a related bug also affects [cyclone/number~]:
https://sourceforge.net/p/pure-data/bugs/1292/

Any thoughts?

Thanks,

Matt & Alex