contour with (x,y) being ranges

[PaulSoderlind]

crashes. (In contrast, several other plot types work with ranges, for instance, surf and plot.)

Example:

x = -10:10
y = range(-5,5,length=11)
z = x.^2 .+ y'.^2

figure()
  contour(x, y, z')                            #crashes
  #contour(collect(x), collect(y), z')         #works

[stevengj]

It throws an exception, which is not the same as "crashing".

[stevengj]

The problem seems to be y. If you replace y with collect(y) it works. PythonCall translates these as:

julia> Py(y)
Python VectorValue: <jl -5.0:1.0:5.0>

julia> Py(collect(y))
Python VectorValue: <jl [-5.0, -4.0, -3.0, -2.0, -1.0, 0.0, 1.0, 2.0, 3.0, 4.0, 5.0]>

so Matplotlib doesn't mind a PythonCall VectorValue built from a Vector, but it doesn't like a VectorValue built from a StepRangeLen for some reason. Maybe there is some missing method that PythonCall should have implemented in the latter case?

[stevengj]

In particular, the problem seems to be that np.asarray doesn't like StepRangeLen:

julia> np = pyimport("numpy")
Python module: <module 'numpy' from '/Users/stevenj/.julia/environments/v1.7/.CondaPkg/env/lib/python3.10/site-packages/numpy/__init__.py'>

julia> np.asarray(y, dtype=np.float64)
ERROR: Python: TypeError: ArrayValue.__array__() takes 1 positional argument but 2 were given
Python stacktrace: none
Stacktrace:
 [1] pythrow()
   @ PythonCall ~/.julia/packages/PythonCall/XgP8G/src/err.jl:94
 [2] errcheck
   @ ~/.julia/packages/PythonCall/XgP8G/src/err.jl:10 [inlined]
 [3] pycallargs
   @ ~/.julia/packages/PythonCall/XgP8G/src/abstract/object.jl:154 [inlined]
 [4] pycall(f::Py, args::StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}; kwargs::Base.Pairs{Symbol, Py, Tuple{Symbol}, NamedTuple{(:dtype,), Tuple{Py}}})
   @ PythonCall ~/.julia/packages/PythonCall/XgP8G/src/abstract/object.jl:165
 [5] #_#11
   @ ~/.julia/packages/PythonCall/XgP8G/src/Py.jl:360 [inlined]
 [6] top-level scope
   @ REPL[15]:1

julia> np.asarray(collect(y), dtype=np.float64)
Python ndarray: array([-5., -4., -3., -2., -1.,  0.,  1.,  2.,  3.,  4.,  5.])

[stevengj]

This works fine:

julia> np.asarray(y)
Python ndarray: array([-5., -4., -3., -2., -1.,  0.,  1.,  2.,  3.,  4.,  5.])

So the problem is the dtype argument.

[stevengj]

It looks like the problem is that this __array__ method in PythonCall doesn't accept the optional dtype argument.

[stevengj]

Closed by JuliaPy/PythonCall.jl@be521aa