Olivier Delalleau

> Non string keys are new in 2.1, so there are no backward compatibility concerns. The potential concern would be that if we implement it, then `${0}` may not work...

> I guess your concern is that implementing your proposal to use the grammar to differentiate will break such cases. My concern was simply to ensure that `{"0": 1, "x":...

Ah, that's a good point, I didn't realize that `np.float64` was a subclass of `float`. However I'm not sure if this is an implementation detail, and `np.float32` isn't (a subclass...

> I am okay with changing the check to `isinstance(value, (float, int, str))`. This only partially solves the issue, since there's no guarantee for numpy floats to inherit from float...

> I am not sure how hard it would be to add (@odelalleau, I am pretty sure you thought about it, what do you think?). IIRC at some point you...

> If anyone wants to try to enable it go ahead but I am learning toward revisiting in 2.2. Not going for it myself but @Jasha10 if you want to...

> I was thinking about introducing an eval resolver later to do arithmetic operations. Personally I would find it more intuitive to do it the way you initially wrote at...

Follow-up thoughts on this (**tl;dr**: I think it can be implemented at the level of resolver arguments, but this would make it a bit more cumbersome to use arithmetic operators...

> At this point I am not taking any new large feature requests for 2.1 Oh, no worries, I didn't intend to push for it in 2.1. What you said...

Ok thanks. I ended up doing exactly what you suggest :) Feel free to close this issue unless you want to keep it around for tracking purpose.