[feature-request]: evaluate should support apply()

[tmtron]

Describe the bug
Currently (v11.6.0) apply not work in evaluation mode. It would be good if that is supported.

see also comment in #2908 by @josdejong

That is an unfortunate side effect: the property apply is considered an unsafe property because it exists as a prototype method of Function: https://github.com/josdejong/mathjs/blob/develop/src/utils/customs.js#L52-L84. If we want to see if we can improve on that, let's discuss in a separate topic.

To Reproduce

apply([[1, 2], [3, 4]], 1, sum)
Error: No access to property "apply"

[josdejong]

Anyone able to help think this though?

[josdejong]

Thanks for looking into this David.

It makes sense to me to rename the function, or create an alias for it to keep things backward compatible. I don't have strong feelings about the best name for the function, I like mapSlice and applyOnDim for example. Do you have a preference?

Good idea to see if we can unify the code of apply and reduce!

[dvd101x]

• mapSlice for me it has more meaning, except that the mapping function that is being applied to every slice of the array should reduce one dimension of each slice and I don't think that's implied by the name.
• applyOnDim I like that it maintains the reference to the current name apply, and the change form apply to applyOnDim sounds similar to what happen between eval and evaluate. Also is not implied in the name apply what is the intention of the callback (reduce one dimension). Also I'm not sure about about dim as in the documentation I see in some functions the full word dimension is used.
• aggregate is also a good fit and found a nice explanation in pandas. The negative part is that it might be more used in database management.

[josdejong]

Shall we go for mapSlice?

[dvd101x]

Yes, I think it's a good name. Maybe it should be mapSlices regardles if it's a single dimension or multiple dimensions because even if it's a single dimension you can still have many slices of the matrix.

[josdejong]

Sounds good to me :)

[gwhitney]

Since this is more of an open-ended architecture discussion than a specific presently-actionable issue, moving to Discussions.