request - running lag function

Sample function shown below for 1-dimension case.
General case would be more complex.
Useful for running differences.

using AxisKeys, Dates

data    = 1:10                             |> collect
date    = Date(2000,1,1):Date(2000,1,10)   |> collect 
ka      = KeyedArray( data; date,   )


function lagwise( ka::KeyedArray, dim::Symbol, n::Int, data_func::Function, dim_func::Function )

    data  = data_func.(   ka.data.data[1+n:end],      ka.data.data[1:end-n]       )

    dimV  = dim_func.(    axiskeys(ka,dim)[1:end-n],  axiskeys(ka,dim)[1+n:end]   )

    KeyedArray( data, dimV )

end


lagwise( ka, :date, 2, -, tuple )

Start by reading the existing KeyedArray, axiskeys, and data access APIs used in the sample lagwise function. Determine how a lag operation should handle arbitrary dimensions and construct the resulting keys; done means a tested running-difference operation that preserves the expected data and dimension keys.