The (long) error is attached below. It can be reproduced with:
Also wondering myself why this was not detected by GitHub actions ... perhaps it is because we did not run the tests of GridapDistributed after moving from Gridap 0.20.3 to 0.20.4?
ERROR: LoadError: MethodError: no method matching LinearAlgebra.Diagonal{Float64, Vector{Float64}}(::UndefInitializer, ::Tuple{Int64, Int64})
Closest candidates are:
LinearAlgebra.Diagonal{T, V}(::Any) where {T, V<:AbstractVector{T}}
@ LinearAlgebra ~/.julia/juliaup/julia-1.10.10+0.x64.linux.gnu/share/julia/stdlib/v1.10/LinearAlgebra/src/diagonal.jl:8
Stacktrace:
[1] similar(::Type{LinearAlgebra.Diagonal{Float64, Vector{Float64}}}, dims::Tuple{Int64, Int64})
@ Base ./abstractarray.jl:876
[2] similar(::Type{LinearAlgebra.Diagonal{Float64, Vector{Float64}}}, ::Int64, ::Int64)
@ Base ./abstractarray.jl:874
[3] testvalue(::Type{LinearAlgebra.Diagonal{Float64, Vector{Float64}}})
@ Gridap.Arrays ~/.julia/packages/Gridap/1FvFK/src/Arrays/Interface.jl:211
[4] testitem
@ ~/.julia/packages/Gridap/1FvFK/src/Arrays/LazyArrays.jl:299 [inlined]
[5] map
@ ./tuple.jl:292 [inlined]
[6] testitem(a::Gridap.Arrays.LazyArray{FillArrays.Fill{typeof(Gridap.Fields.linear_combination), 1, Tuple{Base.OneTo{Int64}}}, Gridap.Fields.LinearCombinationFieldVector{LinearAlgebra.Diagonal{Float64, Vector{Float64}}, Gridap.Fields.BroadcastOpFieldArray{Gridap.ReferenceFEs.ContraVariantPiolaMap, Gridap.Fields.OperationField{Gridap.ReferenceFEs.ContraVariantPiolaMap, Tuple{Gridap.Fields.LinearCombinationField{LinearAlgebra.Diagonal{Int64, FillArrays.Ones{Int64, 1, Tuple{Base.OneTo{Int64}}}}, Gridap.Fields.LinearCombinationFieldVector{LinearAlgebra.Diagonal{Int64, StaticArraysCore.MVector{12, Int64}}, Gridap.Polynomials.CompWiseTensorPolyBasis{2, Gridap.TensorValues.VectorValue{2, Float64}, Gridap.Polynomials.ModalC0}}}, Gridap.Fields.ConstantField{Gridap.TensorValues.TensorValue{2, 2, Float64, 4}}}}, 1, Tuple{Gridap.Fields.LinearCombinationFieldVector{LinearAlgebra.Diagonal{Int64, FillArrays.Ones{Int64, 1, Tuple{Base.OneTo{Int64}}}}, Gridap.Fields.LinearCombinationFieldVector{LinearAlgebra.Diagonal{Int64, StaticArraysCore.MVector{12, Int64}}, Gridap.Polynomials.CompWiseTensorPolyBasis{2, Gridap.TensorValues.VectorValue{2, Float64}, Gridap.Polynomials.ModalC0}}}, Gridap.Fields.ConstantField{Gridap.TensorValues.TensorValue{2, 2, Float64, 4}}}}}, 1, Tuple{Gridap.Arrays.LazyArray{FillArrays.Fill{Gridap.FESpaces.NormalSignMap{Gridap.Geometry.CartesianDiscreteModel{2, Float64, typeof(identity)}}, 1, Tuple{Base.OneTo{Int64}}}, LinearAlgebra.Diagonal{Float64, Vector{Float64}}, 1, Tuple{FillArrays.Fill{Gridap.ReferenceFEs.GenericRefFE{Gridap.ReferenceFEs.RaviartThomas, 2}, 1, Tuple{Base.OneTo{Int64}}}, FillArrays.Fill{SubArray{Vector{Int64}, 1, Vector{Vector{Int64}}, Tuple{UnitRange{Int64}}, true}, 1, Tuple{Base.OneTo{Int64}}}, Gridap.Arrays.LazyArray{FillArrays.Fill{Gridap.Arrays.Reindex{Gridap.Arrays.IdentityVector{Int32}}, 1, Tuple{Base.OneTo{Int64}}}, Int32, 1, Tuple{Gridap.Arrays.LazyArray{FillArrays.Fill{Gridap.Arrays.Reindex{Vector{Int32}}, 1, Tuple{Base.OneTo{Int64}}}, Int32, 1, Tuple{Vector{Int64}}}}}}}, FillArrays.Fill{Gridap.Fields.BroadcastOpFieldArray{Gridap.ReferenceFEs.ContraVariantPiolaMap, Gridap.Fields.OperationField{Gridap.ReferenceFEs.ContraVariantPiolaMap, Tuple{Gridap.Fields.LinearCombinationField{LinearAlgebra.Diagonal{Int64, FillArrays.Ones{Int64, 1, Tuple{Base.OneTo{Int64}}}}, Gridap.Fields.LinearCombinationFieldVector{LinearAlgebra.Diagonal{Int64, StaticArraysCore.MVector{12, Int64}}, Gridap.Polynomials.CompWiseTensorPolyBasis{2, Gridap.TensorValues.VectorValue{2, Float64}, Gridap.Polynomials.ModalC0}}}, Gridap.Fields.ConstantField{Gridap.TensorValues.TensorValue{2, 2, Float64, 4}}}}, 1, Tuple{Gridap.Fields.LinearCombinationFieldVector{LinearAlgebra.Diagonal{Int64, FillArrays.Ones{Int64, 1, Tuple{Base.OneTo{Int64}}}}, Gridap.Fields.LinearCombinationFieldVector{LinearAlgebra.Diagonal{Int64, StaticArraysCore.MVector{12, Int64}}, Gridap.Polynomials.CompWiseTensorPolyBasis{2, Gridap.TensorValues.VectorValue{2, Float64}, Gridap.Polynomials.ModalC0}}}, Gridap.Fields.ConstantField{Gridap.TensorValues.TensorValue{2, 2, Float64, 4}}}}, 1, Tuple{Base.OneTo{Int64}}}}})
@ Gridap.Arrays ~/.julia/packages/Gridap/1FvFK/src/Arrays/LazyArrays.jl:289
....
[34] main(distribute::Type{PartitionedArrays.DebugArray}, parts::Tuple{Int64, Int64})
@ Main.StokesHdivDGTestsSeq.StokesHdivDGTests ~/git-repos/GridapDistributed.jl/test/StokesHdivDGTests.jl:106
[35] #1
@ ~/git-repos/GridapDistributed.jl/test/sequential/StokesHdivDGTests.jl:5 [inlined]
[36] with_debug(f::Main.StokesHdivDGTestsSeq.var"#1#2")
@ PartitionedArrays ~/.julia/packages/PartitionedArrays/MVmxR/src/debug_array.jl:8
[37] top-level scope
@ ~/git-repos/GridapDistributed.jl/test/sequential/StokesHdivDGTests.jl:4
in expression starting at /home/u1134396/git-repos/GridapDistributed.jl/test/sequential/StokesHdivDGTests.jl:1
StokesHdivDGTests.jlfails with GridapDistributed 0.14.5, Gridap >= 0.20.4 (I tried with Gridap 0.20.3 and it does not fail), and julia 1.10.10.The (long) error is attached below. It can be reproduced with:
@JordiManyer ... does it ring a bell what might have been introduced in Gridap 0.20.4 versus 0.20.3 which may cause this behaviour?
Also wondering myself why this was not detected by GitHub actions ... perhaps it is because we did not run the tests of GridapDistributed after moving from Gridap 0.20.3 to 0.20.4?
I have manually triggered Github actions here: https://github.com/gridap/GridapDistributed.jl/actions/runs/25112393338/job/73589905062 to see if I can reproduce this error there. UPDATE: The error can be reproduced in Github actions.
####################