CartesianIndex hard to use

[trafficfan]

# Orginal in Julia 0.6.x
(Tv[i,:],dr[i,:]) = findmax(u_temp+beta*Ev,1)
# in Julia 1.11.2
(Tv[i, :], inds) = findmax(u_temp + beta * Ev, dims=1)
dr[i, :]=[inds[1][2] inds[2][2]] 

using CartesianIndex makes such work hard to understand

[arnavk23]

Thanks for raising this — many users transitioning from older Julia versions hit the same friction.

In newer Julia versions (≥ 0.7), findmax(..., dims=...) returns a tuple:

• First element: the max values

• Second: an array of CartesianIndex objects for the corresponding positions.

So the modern equivalent of:

(Tv[i,:], dr[i,:]) = findmax(u_temp + beta * Ev, 1)

would be something like:

vals, inds = findmax(u_temp + beta * Ev, dims=1)
Tv[i, :] = vals
dr[i, :] = getindex.(inds, 1)  # Get row indices from CartesianIndex

This change supports N-dimensional arrays more robustly, but yes, it's more verbose for simple 2D use-cases.
Suggestions:

• Maybe we could have a helper function like:

plain_indices(inds::Array{CartesianIndex}) = getindex.(inds, 1)

Or improved documentation/examples for common 2D array use.