Sympy throws on returning a tuple out of diophantine call

[DanielFabian]

I respect that most bugs are in SymPyCore, but this one looks like a serialization / type error, so maybe it belongs here (happy to move if you think no).

When I tried following a sympy tutorial, I consistently get a type error. Here's the python snippet

from sympy import symbols
from sympy.solvers.diophantine import diophantine
x, y, z = symbols("x, y, z", integer=True)
print(diophantine(x**2 + y**2 - z**2))

and this is my translation to julia:

using SymPy

@syms a::integer, b::integer, c::integer
pythag_eq = Eq(a^2 + b^2, c^2)

typeof(pythag_eq)

sympy.solvers.diophantine(pythag_eq)

And I get this error:

I'm trying to learn how to translate python examples to julia. In particulae, I have a little bit of python code:

```python
from sympy import symbols
from sympy.solvers.diophantine import diophantine
x, y, z = symbols("x, y, z", integer=True)
print(diophantine(2*x + 3*y - 5))

and I'm trying to translate it to julia as this:

using SymPy

@syms a::integer, b::integer, c::integer
pythag_eq = Eq(a^2 + b^2, c^2)

typeof(pythag_eq)

sympy.solvers.diophantine(pythag_eq)

and I'm getting crashes at the last call:

ERROR: MethodError: Cannot `convert` an object of type 
  Tuple{PyCall.PyObject, PyCall.PyObject, PyCall.PyObject} to an object of type 
  Sym
The function `convert` exists, but no method is defined for this combination of argument types.

Closest candidates are:
  Sym(::Tuple)
   @ SymPyCore ~/.julia/packages/SymPyCore/Dg45i/src/types.jl:38
  Sym(::T) where T
   @ SymPyCore ~/.julia/packages/SymPyCore/Dg45i/src/types.jl:30
  convert(::Type{T}, ::T) where T
   @ Base Base.jl:126
  ...

Stacktrace:
 [1] push!(a::Vector{Sym}, item::Tuple{PyCall.PyObject, PyCall.PyObject, PyCall.PyObject})
   @ Base ./array.jl:1260
 [2] collect(::Type{Sym}, o::PyCall.PyObject)
   @ PyCall ~/.julia/packages/PyCall/1gn3u/src/pyiterator.jl:93
 [3] ↑(::Type{PyCall.PyObject}, x::PyCall.PyObject)
   @ SymPy ~/.julia/packages/SymPy/qBv8D/src/python_connection.jl:52
 [4] ↑(x::PyCall.PyObject)
   @ SymPyCore ~/.julia/packages/SymPyCore/Dg45i/src/gen_methods.jl:191
 [5] (::SymPyCore.SymbolicCallable{PyCall.PyObject})(::Sym{PyCall.PyObject}, ::Vararg{Sym{…}}; kwargs::@Kwargs{})

I have verified that `diophantine` is found as a python callable object. It looks more like an interop issue. But I have no idea what.

I have verified that diophantine is found as a python callable object. It looks more like an interop issue. But I have no idea what.

I even asked on the julia discord and someone else looked at the issue too:

https://discord.com/channels/762167454973296644/1341536826501890198

and when they added a type conversion like this:

Base.convert(::Type{Sym}, pos::NTuple{N, SymPy.PyCall.PyObject}) where N = Sym(Ref(pos));

we do indeed see a result, which is why they recommended to file an upstream bug. Apparently the result from the diophantine call throws every time (fairly independent of the input).

[jverzani]

Thanks. There are a few ways to handle this. This function returns a python object which is a set of tuples. The conversion of this type wasn't begin picked up properly as the tuple wasn't being converted. That should be fixed now with #562 Let me know if not.

[DanielFabian]

thx, can confirm it works.