Sebastian Berg

So I guess it actually is the nearest "larger" floating point number (by absolute value)... I agree that "distance" should really be an absolute value, I have never used this...

Well, right now we still have a chance to change Python. E.g. if we think that our definition for the largest representable float is more reasonable (`inf` rather than the...

Sorry, nvm. I see you already commented about the existence of `np.spacing`. So the only question would be if someone here disagrees with the choice there to say that `ulp(float_max)...

> Note in particular this means that the absolute value of the spacing depends on the sign of the input for powers of two in float16, Hmmm, I think we...

Actually, at least pandas doesn't see to notice really.

I am tempted to say that this is best to "just do", rather than try to deprecate (yes, its not quite finished). There are some issues though, i.e. `timedelta64(0) ==...

Fixing up the typing is proving more complicated and probably needs another approach (and is not finished yet). I added a release note, the two biggest changes are: * Returning...

Yeah, I don't really like the implication either. Generalizing that `0`/`0.0`/`inf`/`nan` special handling choice seems really tricky unfortunately. Some random thoughts: * Yes, we can certainly just make all time...

I think I may implement the warning there (unfortunately), and then revive it as a "2.0 PR" i.e. only active with opt-in... Maybe we could pull it off otherwise, but...

This mainly needs a decision of whether we can get away with this. Otherwise, I will try to do this soonish, but hide it behind "numpy 2.0"...