diag(::DataFrame)?

[anandijain]

The diag function from LinearAlgebra is pretty handy. It may just be extra clutter to have in DataFrames, but it isn't hard to write and I can think of a few cases where it'd be nice to have (heatmaps, etc).

I think that diag(::DataFrame) can be written st you can @assert diag(df, n) == diag(Matrix(df), n) so it wouldn't pollute the namespace as much as if it returned a different type.

Thoughts?

[bkamins]

I am not sure it would be useful. DataFrame is not a matrix and I think it is easy enough to write e.g. what you have written in practice.

A different pattern would be getindex.(Ref(df), axes(df)...) which should be much more efficient for large data frames.

[anandijain]

getindex. and axes is a nice pattern, although it doesn't work on non-square dfs.

this is what i wrote to solve that

min_axes = minimum(axes(df))
getindex.(Ref(df), (min_axes, min_axes...)

also what would be the nice way to write diag using the shift argument:

  julia> A = [1 2 3; 4 5 6; 7 8 9]
  3×3 Array{Int64,2}:
   1  2  3
   4  5  6
   7  8  9

  julia> diag(A,1)
  2-element Array{Int64,1}:
   2
   6

Overall, I think I agree that a separate diag() doesn't need to be written, and was mainly looking for the efficient pattern.