refactor: remove ParameterIndexingProxy, use new getp and setp from SII

src/integrator_interface.jl

@@ -445,8 +445,6 @@ function Base.getproperty(A::DEIntegrator, sym::Symbol)
     if sym === :destats && hasfield(typeof(A), :stats)
         @warn "destats has been deprecated for stats"
         getfield(A, :stats)
-    elseif sym == SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME
-        return ParameterIndexingProxy(A)
     else
         return getfield(A, sym)
     end
@@ -462,8 +460,8 @@ Base.@propagate_inbounds function Base.getindex(A::DEIntegrator, ::ScalarSymboli
     if is_variable(A, sym)
         return A[variable_index(A, sym)]
     elseif is_parameter(A, sym)
-        Base.depwarn("Indexing with parameters is deprecated. Use `$(nameof(typeof(A))).$(SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME)[$sym]` for parameter indexing.", :parameter_getindex)
-        return ParameterIndexingProxy(A)[sym]
+        Base.depwarn("Indexing with parameters is deprecated. Use `getp(sys, $sym)(integrator)` for parameter indexing.", :parameter_getindex)
+        return getp(A, sym)(A)
     elseif is_independent_variable(A, sym)
         return A.t
     elseif is_observed(A, sym)
@@ -502,7 +500,8 @@ function Base.setindex!(A::DEIntegrator, val, sym)
         if is_variable(A, sym)
             A.u[variable_index(A, sym)] = val
         elseif is_parameter(A, sym)
-            A.p[parameter_index(A, sym)] = val
+            Base.depwarn("Parameter indexing is deprecated. Use `setp(sys, $sym)(integrator, $val)` to set parameter value.", :parameter_setindex)
+            setp(A, sym)(A, val)
         else
             error("Invalid indexing of integrator: $sym is not a state or parameter, it may be an observed variable.")
         end

src/problems/problem_interface.jl

@@ -1,25 +1,13 @@
 SymbolicIndexingInterface.symbolic_container(prob::AbstractSciMLProblem) = prob.f
 SymbolicIndexingInterface.parameter_values(prob::AbstractSciMLProblem) = prob.p
 
-function Base.propertynames(prob::AbstractSciMLProblem)
-    (fieldnames(typeof(prob))..., :ps)
-end
-
-function Base.getproperty(prob::AbstractSciMLProblem, sym::Symbol)
-    if sym == SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME
-        return ParameterIndexingProxy(prob)
-    else
-        return getfield(prob, sym)
-    end
-end
-
 Base.@propagate_inbounds function Base.getindex(prob::AbstractSciMLProblem, sym)
     if symbolic_type(sym) == ScalarSymbolic()
         if is_variable(prob.f, sym)
             return prob.u0[variable_index(prob.f, sym)]
         elseif is_parameter(prob.f, sym)
-        Base.depwarn("Indexing with parameters is deprecated. Use `$(nameof(typeof(prob))).$(SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME)[$sym]` for parameter indexing.", :parameter_getindex)
-            return getproperty(prob, SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME)[sym]
+        Base.depwarn("Indexing with parameters is deprecated. Use `getp(prob, $sym)(prob)` for parameter indexing.", :parameter_getindex)
+            return getp(prob, sym)(prob)
         elseif is_independent_variable(prob.f, sym)
             return getindepsym(prob)
         elseif is_observed(prob.f, sym)
@@ -49,7 +37,8 @@ function ___internal_setindex!(prob::AbstractSciMLProblem, val, sym)
         if is_variable(prob.f, sym)
             prob.u0[variable_index(prob.f, sym)] = val
         elseif is_parameter(prob.f, sym)
-            prob.p[parameter_index(prob.f, sym)] = val
+            Base.depwarn("Indexing with parameters is deprecated. Use `setp(prob, $sym)(prob, $val)` to set parameter value.", :parameter_setindex)
+            setp(prob, sym)(prob, val)
         else
             error("Invalid indexing of problem: $sym is not a state or parameter, it may be an observed variable.")
         end

src/solutions/dae_solutions.jl

@@ -47,7 +47,6 @@ Base.@propagate_inbounds function Base.getproperty(x::AbstractDAESolution, s::Sy
     if s === :destats
         Base.depwarn("`sol.destats` is deprecated. Use `sol.stats` instead.", "sol.destats")
         return getfield(x, :stats)
-    elseif s === SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME
     end
     return getfield(x, s)
 end

src/solutions/ode_solutions.jl

@@ -48,8 +48,6 @@ Base.@propagate_inbounds function Base.getproperty(x::AbstractODESolution, s::Sy
     if s === :destats
         Base.depwarn("`sol.destats` is deprecated. Use `sol.stats` instead.", "sol.destats")
         return getfield(x, :stats)
-    elseif s === SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME
-        return ParameterIndexingProxy(x)
     end
     return getfield(x, s)
 end

src/solutions/optimization_solutions.jl

@@ -119,8 +119,6 @@ Base.@propagate_inbounds function Base.getproperty(x::AbstractOptimizationSoluti
         Base.depwarn("`sol.prob` is deprecated. Use getters like `get_p` or `get_syms` on `sol` instead.",
             "sol.prob")
         return getfield(x, :cache)
-    elseif s === SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME
-        return ParameterIndexingProxy(x)
     end
     return getfield(x, s)
 end

src/solutions/rode_solutions.jl

@@ -55,8 +55,6 @@ Base.@propagate_inbounds function Base.getproperty(x::AbstractRODESolution, s::S
     if s === :destats
         Base.depwarn("`sol.destats` is deprecated. Use `sol.stats` instead.", "sol.destats")
         return getfield(x, :stats)
-    elseif s === SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME
-        return ParameterIndexingProxy(x)
     end
     return getfield(x, s)
 end

src/solutions/solution_interface.jl

@@ -58,21 +58,13 @@ SymbolicIndexingInterface.is_time_dependent(::AbstractNoTimeSolution) = false
 # TODO make this nontrivial once dynamic state selection works
 SymbolicIndexingInterface.constant_structure(::AbstractSolution) = true
 
-function Base.getproperty(A::AbstractSolution, sym::Symbol)
-    if sym == SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME
-        return ParameterIndexingProxy(A)
-    else
-        return getfield(A, sym)
-    end
-end
-
 Base.@propagate_inbounds function Base.getindex(A::AbstractNoTimeSolution, sym)
     if symbolic_type(sym) == ScalarSymbolic()
         if is_variable(A, sym)
             return A[variable_index(A, sym)]
         elseif is_parameter(A, sym)
-            Base.depwarn("Indexing with parameters is deprecated. Use `$(nameof(typeof(A))).$(SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME)[$sym]` for parameter indexing.", :parameter_getindex)
-            return getproperty(A, SymbolicIndexingInterface.PARAMETER_INDEXING_PROXY_PROPERTY_NAME)[sym]
+            Base.depwarn("Indexing with parameters is deprecated. Use `getp(sys, $sym)(sol)` for parameter indexing.", :parameter_getindex)
+            return getp(A, sym)(A)
         elseif is_observed(A, sym)
             return SymbolicIndexingInterface.observed(A, sym)(A.u, A.prob.p)
         else

test/downstream/integrator_indexing.jl

@@ -1,4 +1,4 @@
-using ModelingToolkit, OrdinaryDiffEq, RecursiveArrayTools, StochasticDiffEq, Test
+using ModelingToolkit, OrdinaryDiffEq, RecursiveArrayTools, StochasticDiffEq, SymbolicIndexingInterface, Test
 
 ### Tests on non-layered model (everything should work). ###
 
@@ -21,30 +21,30 @@ integrator = init(oprob, Rodas4())
 @test_deprecated integrator[a]
 @test_deprecated integrator[population_model.a]
 @test_deprecated integrator[:a]
-@test integrator.ps[a] == integrator.ps[population_model.a] == integrator.ps[:a] == 2.0
-@test integrator.ps[b] == integrator.ps[population_model.b] == integrator.ps[:b] == 1.0
-@test integrator.ps[c] == integrator.ps[population_model.c] == integrator.ps[:c] == 1.0
-@test integrator.ps[d] == integrator.ps[population_model.d] == integrator.ps[:d] == 1.0
+@test getp(oprob, a)(integrator) == getp(oprob, population_model.a)(integrator) == getp(oprob, :a)(integrator) == 2.0
+@test getp(oprob, b)(integrator) == getp(oprob, population_model.b)(integrator) == getp(oprob, :b)(integrator) == 1.0
+@test getp(oprob, c)(integrator) == getp(oprob, population_model.c)(integrator) == getp(oprob, :c)(integrator) == 1.0
+@test getp(oprob, d)(integrator) == getp(oprob, population_model.d)(integrator) == getp(oprob, :d)(integrator) == 1.0
 
 @test integrator[s1] == integrator[population_model.s1] == integrator[:s1] == 2.0
 @test integrator[s2] == integrator[population_model.s2] == integrator[:s2] == 1.0
 
 step!(integrator, 100.0, true)
 
-@test integrator.ps[a] == integrator.ps[population_model.a] == integrator.ps[:a] == 2.0
-@test integrator.ps[b] == integrator.ps[population_model.b] == integrator.ps[:b] == 1.0
-@test integrator.ps[c] == integrator.ps[population_model.c] == integrator.ps[:c] == 1.0
-@test integrator.ps[d] == integrator.ps[population_model.d] == integrator.ps[:d] == 1.0
+@test getp(population_model, a)(integrator) == getp(population_model, population_model.a)(integrator) == getp(population_model, :a)(integrator) == 2.0
+@test getp(population_model, b)(integrator) == getp(population_model, population_model.b)(integrator) == getp(population_model, :b)(integrator) == 1.0
+@test getp(population_model, c)(integrator) == getp(population_model, population_model.c)(integrator) == getp(population_model, :c)(integrator) == 1.0
+@test getp(population_model, d)(integrator) == getp(population_model, population_model.d)(integrator) == getp(population_model, :d)(integrator) == 1.0
 
 @test integrator[s1] == integrator[population_model.s1] == integrator[:s1] != 2.0
 @test integrator[s2] == integrator[population_model.s2] == integrator[:s2] != 1.0
 
-integrator[a] = 10.0
-@test integrator.ps[a] == integrator.ps[population_model.a] == integrator.ps[:a] == 10.0
-integrator[population_model.b] = 20.0
-@test integrator.ps[b] == integrator.ps[population_model.b] == integrator.ps[:b] == 20.0
-integrator[c] = 30.0
-@test integrator.ps[c] == integrator.ps[population_model.c] == integrator.ps[:c] == 30.0
+setp(oprob, a)(integrator, 10.0)
+@test getp(integrator, a)(integrator) == getp(integrator, population_model.a)(integrator) == getp(integrator, :a)(integrator) == 10.0
+setp(population_model, population_model.b)(integrator, 20.0)
+@test getp(integrator, b)(integrator) == getp(integrator, population_model.b)(integrator) == getp(integrator, :b)(integrator) == 20.0
+setp(integrator, c)(integrator, 30.0)
+@test getp(integrator, c)(integrator) == getp(integrator, population_model.c)(integrator) == getp(integrator, :c)(integrator) == 30.0
 
 integrator[s1] = 10.0
 @test integrator[s1] == integrator[population_model.s1] == integrator[:s1] == 10.0
@@ -62,19 +62,19 @@ integrator = init(sprob, ImplicitEM())
 
 step!(integrator, 100.0, true)
 
-@test integrator.ps[a] == integrator.ps[noisy_population_model.a] == integrator.ps[:a] == 2.0
-@test integrator.ps[b] == integrator.ps[noisy_population_model.b] == integrator.ps[:b] == 1.0
-@test integrator.ps[c] == integrator.ps[noisy_population_model.c] == integrator.ps[:c] == 1.0
-@test integrator.ps[d] == integrator.ps[noisy_population_model.d] == integrator.ps[:d] == 1.0
+@test getp(sprob, a)(integrator) == getp(sprob, noisy_population_model.a)(integrator) == getp(sprob, :a)(integrator) == 2.0
+@test getp(sprob, b)(integrator) == getp(sprob, noisy_population_model.b)(integrator) == getp(sprob, :b)(integrator) == 1.0
+@test getp(sprob, c)(integrator) == getp(sprob, noisy_population_model.c)(integrator) == getp(sprob, :c)(integrator) == 1.0
+@test getp(sprob, d)(integrator) == getp(sprob, noisy_population_model.d)(integrator) == getp(sprob, :d)(integrator) == 1.0
 @test integrator[s1] == integrator[noisy_population_model.s1] == integrator[:s1] != 2.0
 @test integrator[s2] == integrator[noisy_population_model.s2] == integrator[:s2] != 1.0
 
-integrator[a] = 10.0
-@test integrator.ps[a] == integrator.ps[noisy_population_model.a] == integrator.ps[:a] == 10.0
-integrator[noisy_population_model.b] = 20.0
-@test integrator.ps[b] == integrator.ps[noisy_population_model.b] == integrator.ps[:b] == 20.0
-integrator[c] = 30.0
-@test integrator.ps[c] == integrator.ps[noisy_population_model.c] == integrator.ps[:c] == 30.0
+setp(integrator, a)(integrator, 10.0)
+@test getp(noisy_population_model, a)(integrator) == getp(noisy_population_model, noisy_population_model.a)(integrator) == getp(noisy_population_model, :a)(integrator) == 10.0
+setp(sprob, noisy_population_model.b)(integrator, 20.0)
+@test getp(noisy_population_model, b)(integrator) == getp(noisy_population_model, noisy_population_model.b)(integrator) == getp(noisy_population_model, :b)(integrator) == 20.0
+setp(noisy_population_model, c)(integrator, 30.0)
+@test getp(noisy_population_model, c)(integrator) == getp(noisy_population_model, noisy_population_model.c)(integrator) == getp(noisy_population_model, :c)(integrator) == 30.0
 
 integrator[s1] = 10.0
 @test integrator[s1] == integrator[noisy_population_model.s1] == integrator[:s1] == 10.0
@@ -137,13 +137,13 @@ step!(integrator, 100.0, true)
 @test integrator[lorenz1.x] isa Real
 @test integrator[t] isa Real
 @test integrator[α] isa Real
-@test integrator.ps[γ] isa Real
-@test integrator.ps[γ] == 2.0
-@test integrator.ps[(lorenz1.σ, lorenz1.ρ)] isa Tuple
+@test getp(prob, γ)(integrator) isa Real
+@test getp(prob, γ)(integrator) == 2.0
+@test getp(prob, (lorenz1.σ, lorenz1.ρ))(integrator) isa Tuple
 
 @test length(integrator[[lorenz1.x, lorenz2.x]]) == 2
-@test integrator.ps[[γ, lorenz1.σ]] isa Vector{Float64}
-@test length(integrator.ps[[γ, lorenz1.σ]]) == 2
+@test getp(integrator, [γ, lorenz1.σ])(integrator) isa Vector{Float64}
+@test length(getp(integrator, [γ, lorenz1.σ])(integrator)) == 2
 
 @variables q(t)[1:2] = [1.0, 2.0]
 eqs = [D(q[1]) ~ 2q[1]
@@ -178,23 +178,23 @@ integrator2 = init(prob2, Tsit5())
 end
 
 # Tests various interface methods:
-@test_throws Any integrator.ps[σ]
-@test in(integrator.ps[lorenz1.σ], integrator.p)
-@test in(integrator.ps[lorenz2.σ], integrator.p)
-@test_throws Any sol.ps[:σ]
+@test_throws Any getp(sys, σ)(integrator)
+@test in(getp(sys, lorenz1.σ)(integrator), integrator.p)
+@test in(getp(sys, lorenz2.σ)(integrator), integrator.p)
+@test_throws Any getp(sol, :σ)(sol)
 
 @test_throws Any integrator[x]
 @test in(integrator[lorenz1.x], integrator.u)
 @test in(integrator[lorenz2.x], integrator.u)
-@test_throws Any sol[:x]
-
-@test_throws Any integrator.ps[σ]=2.0
-integrator[lorenz1.σ] = 2.0
-@test integrator.ps[lorenz1.σ] == 2.0
-@test integrator.ps[lorenz2.σ] != 2.0
-integrator[lorenz2.σ] = 2.0
-@test integrator.ps[lorenz2.σ] == 2.0
-@test_throws Any sol.ps[:σ]
+@test_throws Any getp(sol, :x)(sol)
+
+@test_throws Any setp(integrator, σ)(integrator, 2.0)
+setp(integrator, lorenz1.σ)(integrator, 2.0)
+@test getp(integrator, lorenz1.σ)(integrator) == 2.0
+@test getp(integrator, lorenz2.σ)(integrator) != 2.0
+setp(integrator, lorenz2.σ)(integrator, 2.0)
+@test getp(integrator, lorenz2.σ)(integrator) == 2.0
+@test_throws Any getp(sol, :σ)(sol)
 
 @test_throws Any integrator[x]=2.0
 integrator[lorenz1.x] = 2.0

test/downstream/symbol_indexing.jl

@@ -1,4 +1,4 @@
-using ModelingToolkit, OrdinaryDiffEq, RecursiveArrayTools, Test
+using ModelingToolkit, OrdinaryDiffEq, RecursiveArrayTools, SymbolicIndexingInterface, Test
 using Optimization, OptimizationOptimJL
 
 @parameters t σ ρ β
@@ -39,10 +39,10 @@ tspan = (0.0, 100.0)
 prob = ODEProblem(sys_simplified, u0, tspan, p)
 sol = solve(prob, Rodas4())
 
-@test_throws Any sol.ps[b]
-@test_throws Any sol.ps[b, 1]
-@test_throws Any sol.ps[b, 1:5]
-@test_throws Any sol.ps[b, [1, 2, 3]]
+@test_throws Any sol[b]
+@test_throws Any sol[b, 1]
+@test_throws Any sol[b, 1:5]
+@test_throws Any sol[b, [1, 2, 3]]
 @test_throws Any sol['a']
 @test_throws Any sol['a', 1]
 @test_throws Any sol['a', 1:5]
@@ -67,9 +67,9 @@ sol = solve(prob, Rodas4())
 @test sol[α] isa Vector
 @test sol[α, 3] isa Float64
 @test length(sol[α, 5:10]) == 6
-@test sol.ps[γ] isa Real
-@test sol.ps[γ] == 2.0
-@test sol.ps[(lorenz1.σ, lorenz1.ρ)] isa Tuple
+@test getp(prob, γ)(sol) isa Real
+@test getp(prob, γ)(sol) == 2.0
+@test getp(prob, (lorenz1.σ, lorenz1.ρ))(sol) isa Tuple
 
 @test sol[[lorenz1.x, lorenz2.x]] isa Vector{Vector{Float64}}
 @test length(sol[[lorenz1.x, lorenz2.x]]) == length(sol)
@@ -201,7 +201,7 @@ D = Differential(t)
 @named fol = ODESystem([D(x) ~ (1 - x) / tau])
 prob = ODEProblem(fol, [x => 0.0], (0.0, 10.0), [tau => 3.0])
 sol = solve(prob, Tsit5())
-@test sol.ps[tau] == 3
+@test getp(fol, tau)(sol) == 3
 
 @testset "OptimizationSolution" begin
     @variables begin
@@ -219,6 +219,6 @@ sol = solve(prob, Tsit5())
     sol = solve(prob, GradientDescent())
     @test sol[x]≈1 atol=1e-3
     @test sol[y]≈1 atol=1e-3
-    @test sol.ps[a] ≈ 1
-    @test sol.ps[b] ≈ 100
+    @test getp(sys, a)(sol) ≈ 1
+    @test getp(sys, b)(sol) ≈ 100
 end