This comes from a more general problem, but for an MWE, consider a function that returns the indices of a vector where one element is not equal to the previous one. I want it to work for general indexes, so I wrote it as

```
function jumpindexes(v::AbstractVector)
xprev = first(v)
ι = LinearIndices(v)
j = similar(ι, 0)
for i in ι[2:end] # PROBLEM HERE
x = v[i]
x == xprev || push!(j, i)
xprev = x
end
j
end
```

but

```
julia> using OffsetArrays
julia> v = [1, 1, 2, 2, 2, 3, 3, 4]
8-element Array{Int64,1}:
1
1
2
2
2
3
3
4
julia> o = OffsetArray(v, -4:3)
OffsetArray(::Array{Int64,1}, -4:3) with eltype Int64 with indices -4:3:
1
1
2
2
2
3
3
4
julia> jumpindexes(v)
3-element Array{Int64,1}:
3
6
8
julia> jumpindexes(o) # wrong, drops negative indexes
2-element Array{Int64,1}:
2
3
```

The problem basically boils down to

```
LinearIndices(o)[2:end]
```

How can I implement this algorithm with general indexing?

In general, how can I take subranges for general indexes?

(edit: on v0.7)