update to TrixiTest.jl v2 (#2683)

* update to TrixiTest.jl v2

* up

* up

* up

* Update test/test_tree_2d_mhd.jl

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

* Update test/test_p4est_2d.jl

* Update test/test_tree_2d_mhd.jl

* change

* Apply suggestions from code review

* fix

* comments

* Update test/Project.toml

* Apply suggestions from code review

* Update test/test_structured_2d.jl

* Update test/test_p4est_3d.jl

* Update test/test_structured_2d.jl

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

* Update test/test_structured_2d.jl

---------

Co-authored-by: Daniel Doehring <daniel.doehring@rwth-aachen.de>
Co-authored-by: Joshua Lampert <51029046+JoshuaLampert@users.noreply.github.com>

Author: 3 people

test/Project.toml

@@ -74,4 +74,4 @@ SparseConnectivityTracer = "1.0.1"
 SparseMatrixColorings = "0.4.21"
 StableRNGs = "1.0.2"
 Test = "1"
-TrixiTest = "0.1.4"
+TrixiTest = "0.2.1"

test/test_cuda.jl

@@ -22,7 +22,7 @@ EXAMPLES_DIR = joinpath(examples_dir(), "p4est_2d_dgsem")
     @test real(ode.p.solver) == Float64
     @test real(ode.p.solver.basis) == Float64
     @test real(ode.p.solver.mortar) == Float64
-    # TODO: remake ignores the mesh itself as well
+    # TODO: `mesh` is currently not `adapt`ed correctly
     @test real(ode.p.mesh) == Float64
 
     @test ode.u0 isa Array
@@ -41,7 +41,7 @@ end
                         # Expected errors are exactly the same as with TreeMesh!
                         l2=nothing,   # TODO: GPU. [Float32(8.311947673061856e-6)],
                         linf=nothing, # TODO: GPU. [Float32(6.627000273229378e-5)],
-                        RealT=Float32,
+                        RealT_for_test_tolerances=Float32,
                         real_type=Float32,
                         storage_type=CuArray,
                         sol=nothing,) # TODO: GPU. Remove this once we can run the simulation on the GPU
@@ -51,7 +51,7 @@ end
     @test real(ode.p.solver) == Float32
     @test real(ode.p.solver.basis) == Float32
     @test real(ode.p.solver.mortar) == Float32
-    # TODO: remake ignores the mesh itself as well
+    # TODO: `mesh` is currently not `adapt`ed correctly
     @test real(ode.p.mesh) == Float64
 
     @test ode.u0 isa CuArray

test/test_p4est_2d.jl

@@ -26,7 +26,7 @@ isdir(outdir) && rm(outdir, recursive = true)
     @test real(semi32.solver) == Float32
     @test real(semi32.solver.basis) == Float32
     @test real(semi32.solver.mortar) == Float32
-    # TODO: remake ignores the mesh itself as well
+    # TODO: `mesh` is currently not `adapt`ed correctly
     @test real(semi32.mesh) == Float64
 end
 
@@ -35,7 +35,7 @@ end
                         # Expected errors are exactly the same as with TreeMesh!
                         l2=[Float32(8.311947673061856e-6)],
                         linf=[Float32(6.627000273229378e-5)],
-                        RealT=Float32,
+                        RealT_for_test_tolerances=Float32,
                         real_type=Float32)
     # Ensure that we do not have excessive memory allocations
     # (e.g., from type instabilities)
@@ -48,7 +48,7 @@ end
     @test real(ode.p.solver) == Float32
     @test real(ode.p.solver.basis) == Float32
     @test real(ode.p.solver.mortar) == Float32
-    # TODO: remake ignores the mesh itself as well
+    # TODO: `mesh` is currently not `adapt`ed correctly
     @test real(ode.p.mesh) == Float64
 end
 
@@ -278,7 +278,7 @@ end
                         ],
                         tspan=(0.0f0, 1.0f0),
                         rtol=10 * sqrt(eps(Float32)), # to make CI pass
-                        RealT=Float32)
+                        RealT_for_test_tolerances=Float32)
     # Ensure that we do not have excessive memory allocations
     # (e.g., from type instabilities)
     @test_allocations(Trixi.rhs!, semi, sol, 1000)

test/test_p4est_3d.jl

@@ -228,7 +228,7 @@ end
                             Float32(2.298161660974074e-14),
                             Float32(6.039613253960852e-14)
                         ],
-                        RealT=Float32)
+                        RealT_for_test_tolerances=Float32)
     # Ensure that we do not have excessive memory allocations
     # (e.g., from type instabilities)
     @test_allocations(Trixi.rhs!, semi, sol, 1000)

test/test_structured_2d.jl

@@ -30,7 +30,7 @@ end
                         # Expected errors are taken from elixir_advection_basic.jl
                         l2=[Float32(8.311947673061856e-6)],
                         linf=[Float32(6.627000273229378e-5)],
-                        RealT=Float32)
+                        RealT_for_test_tolerances=Float32)
     # Ensure that we do not have excessive memory allocations
     # (e.g., from type instabilities)
     @test_allocations(Trixi.rhs!, semi, sol, 1000)

test/test_tree_2d_mhd.jl

@@ -177,7 +177,7 @@ end
                                      0.09208131,
                                      0.14795369,
                                      2.2078211f-6],
-                        RealT=Float32)
+                        RealT_for_test_tolerances=Float32)
     # Ensure that we do not have excessive memory allocations
     # (e.g., from type instabilities)
     @test_allocations(Trixi.rhs!, semi, sol, 1000)
@@ -402,7 +402,7 @@ end
                         # In the `StepsizeCallback`, though, the less diffusive `max_abs_speeds` is employed which is consistent with `max_abs_speed`.
                         # Thus, we exchanged in PR#2458 the default wave speed used in the LLF flux to `max_abs_speed`.
                         # To ensure that every example still runs we specify explicitly `FluxLaxFriedrichs(max_abs_speed_naive)`.
-                        # We remark, however, that the now default `max_abs_speed` is in general recommended due to compliance with the 
+                        # We remark, however, that the now default `max_abs_speed` is in general recommended due to compliance with the
                         # `StepsizeCallback` (CFL-Condition) and less diffusion.
                         surface_flux=(FluxLaxFriedrichs(max_abs_speed_naive),
                                       flux_nonconservative_powell_local_jump),