breaking changes for Trixi.jl v0.5

NEWS.md

@@ -4,6 +4,31 @@ Trixi.jl follows the interpretation of [semantic versioning (semver)](https://ju
 used in the Julia ecosystem. Notable changes will be documented in this file
 for human readability.
 
+## Changes when updating to v0.5 from v0.4.x
+
+#### Added
+
+#### Changed
+
+- Compile-time boolean indicators have been changed from `Val{true}`/`Val{false}`
+  to `Trixi.True`/`Trixi.False`. This affects user code only if new equations
+  with nonconservative terms are created. Change
+  `Trixi.has_nonconservative_terms(::YourEquations) = Val{true}()` to
+  `Trixi.has_nonconservative_terms(::YourEquations) = Trixi.True()`.
+- The (non-exported) DGSEM function `split_form_kernel!` has been renamed to `flux_differencing_kernel!`
+
+#### Deprecated
+
+- The signature of the `DGMultiMesh` constructors has changed - the `dg::DGMulti`
+  argument now comes first.
+- The undocumented and unused
+  `DGMultiMesh(triangulateIO, rd::RefElemData{2, Tri}, boundary_dict::Dict{Symbol, Int})`
+  constructor was removed.
+
+#### Removed
+
+- Everything deprecated in Trixi.jl v0.4.x has been removed.
+
 
 ## Changes in the v0.4 lifecycle
 

Project.toml

@@ -1,7 +1,7 @@
 name = "Trixi"
 uuid = "a7f1ee26-1774-49b1-8366-f1abc58fbfcb"
 authors = ["Michael Schlottke-Lakemper <[email protected]>", "Gregor Gassner <[email protected]>", "Hendrik Ranocha <[email protected]>", "Andrew R. Winters <[email protected]>", "Jesse Chan <[email protected]>"]
-version = "0.4.60-pre"
+version = "0.5.0-pre"
 
 [deps]
 CodeTracking = "da1fd8a2-8d9e-5ec2-8556-3022fb5608a2"

docs/literate/src/files/adding_nonconservative_equation.jl

@@ -62,7 +62,7 @@ end
 
 
 ## We use nonconservative terms
-have_nonconservative_terms(::NonconservativeLinearAdvectionEquation) = Val(true)
+have_nonconservative_terms(::NonconservativeLinearAdvectionEquation) = Trixi.True()
 
 ## This "nonconservative numerical flux" implements the nonconservative terms.
 ## In general, nonconservative terms can be written in the form
@@ -219,7 +219,7 @@ end
 
 
 ## We use nonconservative terms
-have_nonconservative_terms(::NonconservativeLinearAdvectionEquation) = Val(true)
+have_nonconservative_terms(::NonconservativeLinearAdvectionEquation) = Trixi.True()
 
 ## This "nonconservative numerical flux" implements the nonconservative terms.
 ## In general, nonconservative terms can be written in the form

examples/dgmulti_2d/elixir_euler_bilinear.jl

@@ -26,7 +26,7 @@ for i in eachindex(vertex_coordinates[1])
   setindex!.(vertex_coordinates, mapping(vx, vy), i)
 end
 
-mesh = DGMultiMesh(vertex_coordinates, EToV, dg, is_on_boundary=is_on_boundary)
+mesh = DGMultiMesh(dg, vertex_coordinates, EToV, is_on_boundary=is_on_boundary)
 
 boundary_condition_convergence_test = BoundaryConditionDirichlet(initial_condition)
 boundary_conditions = (; :top => boundary_condition_convergence_test,

examples/dgmulti_2d/elixir_euler_triangulate_pkg_mesh.jl

@@ -13,7 +13,7 @@ meshIO = StartUpDG.triangulate_domain(StartUpDG.RectangularDomainWithHole())
 
 # the pre-defined Triangulate geometry in StartUpDG has integer boundary tags. this routine
 # assigns boundary faces based on these integer boundary tags.
-mesh = DGMultiMesh(meshIO, dg, Dict(:outer_boundary=>1, :inner_boundary=>2))
+mesh = DGMultiMesh(dg, meshIO, Dict(:outer_boundary=>1, :inner_boundary=>2))
 
 boundary_condition_convergence_test = BoundaryConditionDirichlet(initial_condition)
 boundary_conditions = (; :outer_boundary => boundary_condition_convergence_test,

src/Trixi.jl

@@ -52,7 +52,7 @@ using P4est
 using Setfield: @set
 using RecipesBase: RecipesBase
 using Requires: @require
-using Static: Static, One
+using Static: Static, One, True, False
 @reexport using StaticArrays: SVector
 using StaticArrays: StaticArrays, MVector, MArray, SMatrix, @SMatrix
 using StrideArrays: PtrArray, StrideArray, StaticInt
@@ -235,7 +235,6 @@ export ode_norm, ode_unstable_check
 export convergence_test, jacobian_fd, jacobian_ad_forward, linear_structure
 
 export DGMulti, estimate_dt, DGMultiMesh, GaussSBP
-export VertexMappedMesh # TODO: DGMulti, v0.5. Remove deprecated VertexMappedMesh in next release
 
 export ViscousFormulationBassiRebay1, ViscousFormulationLocalDG
 

src/auxiliary/mpi.jl

@@ -47,9 +47,15 @@ const MPI_IS_ROOT = Ref(true)
 
 # This is not type-stable but that's okay since we want to get rid of it anyway
 # and it's not used in performance-critical parts. The alternative we used before,
-# calling something like `eval(:(mpi_parallel() = Val(true)))` in `init_mpi()`,
+# calling something like `eval(:(mpi_parallel() = True()))` in `init_mpi()`,
 # causes invalidations and slows down the first call to Trixi.
-mpi_parallel()::Union{Val{true}, Val{false}} = Val(mpi_isparallel())
+function mpi_parallel()
+  if mpi_isparallel()
+    return True()
+  else
+    return False()
+  end
+end
 
 @inline mpi_isroot() = MPI_IS_ROOT[]
 

src/callbacks_step/stepsize_dg1d.jl

@@ -6,7 +6,7 @@
 
 
 function max_dt(u, t, mesh::TreeMesh{1},
-                constant_speed::Val{false}, equations, dg::DG, cache)
+                constant_speed::False, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -27,7 +27,7 @@ end
 
 
 function max_dt(u, t, mesh::TreeMesh{1},
-                constant_speed::Val{true}, equations, dg::DG, cache)
+                constant_speed::True, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -43,7 +43,7 @@ end
 
 
 function max_dt(u, t, mesh::StructuredMesh{1},
-                constant_speed::Val{false}, equations, dg::DG, cache)
+                constant_speed::False, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -68,7 +68,7 @@ end
 
 
 function max_dt(u, t, mesh::StructuredMesh{1},
-                constant_speed::Val{true}, equations, dg::DG, cache)
+                constant_speed::True, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))

src/callbacks_step/stepsize_dg2d.jl

@@ -6,7 +6,7 @@
 
 
 function max_dt(u, t, mesh::TreeMesh{2},
-                constant_speed::Val{false}, equations, dg::DG, cache)
+                constant_speed::False, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -28,7 +28,7 @@ end
 
 
 function max_dt(u, t, mesh::TreeMesh{2},
-                constant_speed::Val{true}, equations, dg::DG, cache)
+                constant_speed::True, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -44,7 +44,7 @@ end
 
 
 function max_dt(u, t, mesh::ParallelTreeMesh{2},
-                constant_speed::Val{false}, equations, dg::DG, cache)
+                constant_speed::False, equations, dg::DG, cache)
   # call the method accepting a general `mesh::TreeMesh{2}`
   # TODO: MPI, we should improve this; maybe we should dispatch on `u`
   #       and create some MPI array type, overloading broadcasting and mapreduce etc.
@@ -60,7 +60,7 @@ end
 
 
 function max_dt(u, t, mesh::ParallelTreeMesh{2},
-                constant_speed::Val{true}, equations, dg::DG, cache)
+                constant_speed::True, equations, dg::DG, cache)
   # call the method accepting a general `mesh::TreeMesh{2}`
   # TODO: MPI, we should improve this; maybe we should dispatch on `u`
   #       and create some MPI array type, overloading broadcasting and mapreduce etc.
@@ -76,7 +76,7 @@ end
 
 
 function max_dt(u, t, mesh::Union{StructuredMesh{2}, UnstructuredMesh2D, P4estMesh{2}},
-                constant_speed::Val{false}, equations, dg::DG, cache)
+                constant_speed::False, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -109,7 +109,7 @@ end
 
 
 function max_dt(u, t, mesh::Union{StructuredMesh{2}, UnstructuredMesh2D, P4estMesh{2}},
-                constant_speed::Val{true}, equations, dg::DG, cache)
+                constant_speed::True, equations, dg::DG, cache)
   @unpack contravariant_vectors, inverse_jacobian = cache.elements
 
   # to avoid a division by zero if the speed vanishes everywhere,
@@ -136,7 +136,7 @@ end
 
 
 function max_dt(u, t, mesh::ParallelP4estMesh{2},
-                constant_speed::Val{false}, equations, dg::DG, cache)
+                constant_speed::False, equations, dg::DG, cache)
   # call the method accepting a general `mesh::P4estMesh{2}`
   # TODO: MPI, we should improve this; maybe we should dispatch on `u`
   #       and create some MPI array type, overloading broadcasting and mapreduce etc.
@@ -152,7 +152,7 @@ end
 
 
 function max_dt(u, t, mesh::ParallelP4estMesh{2},
-                constant_speed::Val{true}, equations, dg::DG, cache)
+                constant_speed::True, equations, dg::DG, cache)
   # call the method accepting a general `mesh::P4estMesh{2}`
   # TODO: MPI, we should improve this; maybe we should dispatch on `u`
   #       and create some MPI array type, overloading broadcasting and mapreduce etc.

src/callbacks_step/stepsize_dg3d.jl

@@ -6,7 +6,7 @@
 
 
 function max_dt(u, t, mesh::TreeMesh{3},
-                constant_speed::Val{false}, equations, dg::DG, cache)
+                constant_speed::False, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -29,7 +29,7 @@ end
 
 
 function max_dt(u, t, mesh::TreeMesh{3},
-                constant_speed::Val{true}, equations, dg::DG, cache)
+                constant_speed::True, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -45,7 +45,7 @@ end
 
 
 function max_dt(u, t, mesh::Union{StructuredMesh{3}, P4estMesh{3}},
-                constant_speed::Val{false}, equations, dg::DG, cache)
+                constant_speed::False, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -80,7 +80,7 @@ end
 
 
 function max_dt(u, t, mesh::Union{StructuredMesh{3}, P4estMesh{3}},
-                constant_speed::Val{true}, equations, dg::DG, cache)
+                constant_speed::True, equations, dg::DG, cache)
   # to avoid a division by zero if the speed vanishes everywhere,
   # e.g. for steady-state linear advection
   max_scaled_speed = nextfloat(zero(t))
@@ -110,7 +110,7 @@ end
 
 
 function max_dt(u, t, mesh::ParallelP4estMesh{3},
-                constant_speed::Val{false}, equations, dg::DG, cache)
+                constant_speed::False, equations, dg::DG, cache)
   # call the method accepting a general `mesh::P4estMesh{3}`
   # TODO: MPI, we should improve this; maybe we should dispatch on `u`
   #       and create some MPI array type, overloading broadcasting and mapreduce etc.
@@ -126,7 +126,7 @@ end
 
 
 function max_dt(u, t, mesh::ParallelP4estMesh{3},
-                constant_speed::Val{true}, equations, dg::DG, cache)
+                constant_speed::True, equations, dg::DG, cache)
   # call the method accepting a general `mesh::P4estMesh{3}`
   # TODO: MPI, we should improve this; maybe we should dispatch on `u`
   #       and create some MPI array type, overloading broadcasting and mapreduce etc.

src/equations/acoustic_perturbation_2d.jl

@@ -330,7 +330,7 @@ end
 end
 
 
-@inline have_constant_speed(::AcousticPerturbationEquations2D) = Val(false)
+@inline have_constant_speed(::AcousticPerturbationEquations2D) = False()
 
 @inline function max_abs_speeds(u, equations::AcousticPerturbationEquations2D)
   v1_mean = u[4]

src/equations/equations.jl

@@ -204,10 +204,10 @@ with or without nonconservative terms. Classical conservation laws such as the
 [`CompressibleEulerEquations2D`](@ref) do not have nonconservative terms. The
 [`ShallowWaterEquations2D`](@ref) with non-constant bottom topography are an
 example of equations with nonconservative terms.
-The return value will be `Val(true)` or `Val(false)` to allow dispatching on the return type.
+The return value will be `True()` or `False()` to allow dispatching on the return type.
 """
-have_nonconservative_terms(::AbstractEquations) = Val(false)
-have_constant_speed(::AbstractEquations) = Val(false)
+have_nonconservative_terms(::AbstractEquations) = False()
+have_constant_speed(::AbstractEquations) = False()
 
 default_analysis_errors(::AbstractEquations)     = (:l2_error, :linf_error)
 """
@@ -225,7 +225,7 @@ Return the conserved variables `u`. While this function is as trivial as `identi
 it is also as useful.
 """
 @inline cons2cons(u, ::AbstractEquations) = u
-                                                                            
+
 @inline Base.first(u, ::AbstractEquations) = first(u)
 
 """

src/equations/hyperbolic_diffusion_1d.jl

@@ -166,7 +166,7 @@ end
 end
 
 
-@inline have_constant_speed(::HyperbolicDiffusionEquations1D) = Val(true)
+@inline have_constant_speed(::HyperbolicDiffusionEquations1D) = True()
 
 @inline function max_abs_speeds(eq::HyperbolicDiffusionEquations1D)
   return sqrt(eq.nu * eq.inv_Tr)

src/equations/hyperbolic_diffusion_2d.jl

@@ -206,7 +206,7 @@ end
 
 
 
-@inline have_constant_speed(::HyperbolicDiffusionEquations2D) = Val(true)
+@inline have_constant_speed(::HyperbolicDiffusionEquations2D) = True()
 
 @inline function max_abs_speeds(eq::HyperbolicDiffusionEquations2D)
   λ = sqrt(eq.nu * eq.inv_Tr)

src/equations/hyperbolic_diffusion_3d.jl

@@ -198,7 +198,7 @@ end
 
 
 
-@inline have_constant_speed(::HyperbolicDiffusionEquations3D) = Val(true)
+@inline have_constant_speed(::HyperbolicDiffusionEquations3D) = True()
 
 @inline function max_abs_speeds(eq::HyperbolicDiffusionEquations3D)
   λ = sqrt(eq.nu * eq.inv_Tr)

src/equations/ideal_glm_mhd_1d.jl

@@ -25,7 +25,7 @@ struct IdealGlmMhdEquations1D{RealT<:Real} <: AbstractIdealGlmMhdEquations{1, 8}
   end
 end
 
-have_nonconservative_terms(::IdealGlmMhdEquations1D) = Val(false)
+have_nonconservative_terms(::IdealGlmMhdEquations1D) = False()
 varnames(::typeof(cons2cons), ::IdealGlmMhdEquations1D) = ("rho", "rho_v1", "rho_v2", "rho_v3", "rho_e", "B1", "B2", "B3")
 varnames(::typeof(cons2prim), ::IdealGlmMhdEquations1D) = ("rho", "v1", "v2", "v3", "p", "B1", "B2", "B3")
 default_analysis_integrals(::IdealGlmMhdEquations1D)  = (entropy_timederivative, Val(:l2_divb), Val(:linf_divb))

src/equations/ideal_glm_mhd_2d.jl

@@ -28,7 +28,7 @@ function IdealGlmMhdEquations2D(gamma; initial_c_h=convert(typeof(gamma), NaN))
 end
 
 
-have_nonconservative_terms(::IdealGlmMhdEquations2D) = Val(true)
+have_nonconservative_terms(::IdealGlmMhdEquations2D) = True()
 varnames(::typeof(cons2cons), ::IdealGlmMhdEquations2D) = ("rho", "rho_v1", "rho_v2", "rho_v3", "rho_e", "B1", "B2", "B3", "psi")
 varnames(::typeof(cons2prim), ::IdealGlmMhdEquations2D) = ("rho", "v1", "v2", "v3", "p", "B1", "B2", "B3", "psi")
 default_analysis_integrals(::IdealGlmMhdEquations2D)  = (entropy_timederivative, Val(:l2_divb), Val(:linf_divb))

src/equations/ideal_glm_mhd_3d.jl

@@ -28,7 +28,7 @@ function IdealGlmMhdEquations3D(gamma; initial_c_h=convert(typeof(gamma), NaN))
 end
 
 
-have_nonconservative_terms(::IdealGlmMhdEquations3D) = Val(true)
+have_nonconservative_terms(::IdealGlmMhdEquations3D) = True()
 varnames(::typeof(cons2cons), ::IdealGlmMhdEquations3D) = ("rho", "rho_v1", "rho_v2", "rho_v3", "rho_e", "B1", "B2", "B3", "psi")
 varnames(::typeof(cons2prim), ::IdealGlmMhdEquations3D) = ("rho", "v1", "v2", "v3", "p", "B1", "B2", "B3", "psi")
 default_analysis_integrals(::IdealGlmMhdEquations3D)  = (entropy_timederivative, Val(:l2_divb), Val(:linf_divb))

src/equations/ideal_glm_mhd_multicomponent_1d.jl

@@ -45,7 +45,7 @@ end
 
 @inline Base.real(::IdealGlmMhdMulticomponentEquations1D{NVARS, NCOMP, RealT}) where {NVARS, NCOMP, RealT} = RealT
 
-have_nonconservative_terms(::IdealGlmMhdMulticomponentEquations1D) = Val(false)
+have_nonconservative_terms(::IdealGlmMhdMulticomponentEquations1D) = False()
 
 function varnames(::typeof(cons2cons), equations::IdealGlmMhdMulticomponentEquations1D)
 

src/equations/ideal_glm_mhd_multicomponent_2d.jl

@@ -47,7 +47,7 @@ end
 
 @inline Base.real(::IdealGlmMhdMulticomponentEquations2D{NVARS, NCOMP, RealT}) where {NVARS, NCOMP, RealT} = RealT
 
-have_nonconservative_terms(::IdealGlmMhdMulticomponentEquations2D) = Val(true)
+have_nonconservative_terms(::IdealGlmMhdMulticomponentEquations2D) = True()
 
 function varnames(::typeof(cons2cons), equations::IdealGlmMhdMulticomponentEquations2D)
 

src/equations/lattice_boltzmann_2d.jl

@@ -360,7 +360,7 @@ end
 
 
 
-@inline have_constant_speed(::LatticeBoltzmannEquations2D) = Val(true)
+@inline have_constant_speed(::LatticeBoltzmannEquations2D) = True()
 
 @inline function max_abs_speeds(equations::LatticeBoltzmannEquations2D)
   @unpack c = equations

src/equations/lattice_boltzmann_3d.jl

@@ -368,7 +368,7 @@ end
 
 
 
-@inline have_constant_speed(::LatticeBoltzmannEquations3D) = Val(true)
+@inline have_constant_speed(::LatticeBoltzmannEquations3D) = True()
 
 @inline function max_abs_speeds(equations::LatticeBoltzmannEquations3D)
   @unpack c = equations