Type stability of adjoint(::Sym)

[ranocha]

Currently, Julia v1.5.3 with SymPy v1.0.35 is not able to infer adjoint(::Sym) correctly as ::Sym. As a result, the return type of multiplying adjoint matrices of Sym by matrices of Sym are matrices with eltype Any. I recognized this issue since Matrix{Any} is really printed worse than Matrix{Sym} in Jupyter. Example:

julia> using SymPy

julia> A = ones(Sym, 1, 1)
1×1 Array{Sym,2}:
 1

julia> A * A # is inferred correctly
1×1 Array{Sym,2}:
 1

julia> A * A' # is inferred with eltype Any
1×1 Array{Any,2}:
 1

julia> A' * A # is inferred with eltype Any
1×1 Array{Any,2}:
 1

julia> adjoint(one(Sym)) |> typeof
Sym

julia> @code_warntype adjoint(one(Sym))
Variables
  #self#::Core.Compiler.Const(adjoint, false)
  ex::Sym
  args::Tuple{}

Body::Any
1 ─ %1 = Core.NamedTuple()::Core.Compiler.Const(NamedTuple(), false)
│   %2 = Base.pairs(%1)::Core.Compiler.Const(Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}(), false)
│   %3 = Core.tuple(%2, #self#, ex)::Core.Compiler.PartialStruct(Tuple{Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}},typeof(adjoint),Sym}, Any[Core.Compiler.Const(Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}(), false), Core.Compiler.Const(adjoint, false), Sym])
│   %4 = Core._apply_iterate(Base.iterate, SymPy.:(var"#adjoint#151"), %3, args)::Any
└──      return %4

Could an additional definition with type hint for the method adjoint(::Sym)::Sym help here?

[jverzani]

Yes, that seems like the right fix. Thanks for the report.

[ranocha]

Thanks for this fast fix!