Turn mutable structs into normal (immutable) structs about lazysets.jl HOT 7 CLOSED

I reviewed and found these places:

ExponentialMap.jl: mutable struct SparseMatrixExp
ExponentialMap.jl: mutable struct ExponentialMap <: LazySet
ExponentialMap.jl: mutable struct ProjectionSparseMatrixExp
ExponentialMap.jl: mutable struct ExponentialProjectionMap <: LazySet
CartesianProduct.jl: mutable struct CartesianProductArray <: LazySet
MinkowskiSum.jl: mutable struct MinkowskiSumArray <: LazySet
Polygon.jl: mutable struct HPolygon <: LazySet
Polygon.jl: mutable struct HPolygonOpt <: LazySet
Polygon.jl: mutable struct VPolygon <: LazySet
Polyhedron.jl: mutable struct Polyhedron <: LazySet

Most of the types just wrap an array.
If we remove the mutable, we may still modify the array; we are only not allowed to assign a new array anymore, which I think is fine (more docs here).

I expect that in Polyhedron.jl the dimension dim should not change.

The only place where I am not confident that the change is save is in HPolygonOpt.jl:
Do we want to be able to modify the field ind?

One more thing: We have one occurrence of type, which should probably also become a struct:

LinearMap.jl: type LinearMap <: LazySet

from lazysets.jl.

If we remove the mutable, we may still modify the array; we are only not allowed to assign a new array anymore, which I think is fine (more docs here).

good point, let's do that.

I expect that in Polyhedron.jl the dimension dim should not change.

yes. we wanted dim to be the ambient dimension, i.e. even this is square patch in 3d-space

The only place where I am not confident that the change is save is in HPolygonOpt.jl: Do we want to be able to modify the field ind?

yes, i think that ind should be modified because it is updated at each iteration. @viryfrederic do you agree?

One more thing: We have one occurrence of type, which should probably also become a struct:

yes we should update that for future versions

from lazysets.jl.

I pushed the changes on top of the branch of PR #21.
Let us see how this works 😼

from lazysets.jl.

Cool. Then your proposed changes for this issue are already on master branch, or some but not all of them?

from lazysets.jl.

Nope, they are in a separate branch which I based on the other branch.
I fear that the merging will be a mess, but I wanted to try.
I will add the PR later and merge it directly after the tests have passed.

from lazysets.jl.

i think you can just rebase t/schilllic/17 into the current master, see the network graph

from lazysets.jl.

You are right, I have just queried the PR.

from lazysets.jl.