This is pretty cool!

[asinghvi17]

I just saw this on the new packages feed - this looks pretty cool to me!

A couple of questions:

https://github.com/JuliaAPlavin/MakieExtra.jl/blob/000000006d421b318d28db435c70c78ce96659d9/src/scales.jl#L24 Why InverseFunctions instead of Base.inv?

https://github.com/JuliaAPlavin/MakieExtra.jl/blob/000000006d421b318d28db435c70c78ce96659d9/src/scales.jl#L2-L6 all of these should be Float64

https://github.com/JuliaAPlavin/MakieExtra.jl/blob/000000006d421b318d28db435c70c78ce96659d9/src/MakieExtra.jl#L6 maybe consider Makie.Format instead ;)

[aplavin]

Why InverseFunctions instead of Base.inv?

InverseFunctions appears to be the way to invert supported functions and to define new inverses. It's quite common in the Julia ecosystem.

In particular, here we don't just define new inverse methods, but reuse existing ones as well:

julia> using Accessors, InverseFunctions

julia> f = @o 10 * (5 + log(2, _ / 10))
Base.Fix1{typeof(*), Int64}(*, 10) ∘ Base.Fix1{typeof(+), Int64}(+, 5) ∘ Base.Fix1{typeof(log), Int64}(log, 2) ∘ Base.Fix2{typeof(/), Int64}(/, 10)

julia> inverse(f)
Base.Fix2{typeof(*), Int64}(*, 10) ∘ (Base.Fix1{typeof(InverseFunctions.invlog1), Int64}(InverseFunctions.invlog1, 2) ∘ (Base.Fix2{typeof(-), Int64}(-, 5) ∘ Base.Fix1{typeof(\), Int64}(\, 10)))

Everything fully automatic, with no potential for typos :)

I think in the future Makie should just use/define InverseFunction.inverse instead of its own inverse_transform function.

[aplavin]

all of these should be Float64

Oh, thanks, I kinda thought Makie is "Float32 all the way".

[aplavin]

maybe consider Makie.Format instead ;)

I just like how nice the Python string formatting syntax is (PyFormattedStrings copies the syntax directly), see https://github.com/JuliaAPlavin/MakieExtra.jl/blob/000000006d421b318d28db435c70c78ce96659d9/src/ticks.jl#L66 How that would look with Makie.Format?

Of course it's not crucial for such a small one-time use, just what different people are used to in terms of syntax.

[asinghvi17]

I think Format has FormatExpr which obeys the same spec as Python.