PyObject -> Array fails on julia 0.4

[swt30]

I have the same problem as #99 with this FITS file using astropy.io.fits on julia 0.4. It works fine on julia 0.3.

julia> using PyCall
julia> @pyimport astropy.io.fits as pyf
julia> fitsfile = pyf.open("data/fits-image.fits")
    1-element Array{Any,1}:
     PyObject <astropy.io.fits.hdu.image.PrimaryHDU object at 0x92512e4c>
julia> image = fitsfile[1]
    PyObject <astropy.io.fits.hdu.image.PrimaryHDU object at 0x92512e4c>
julia> pyimdata = image[:data]

At this stage pyimdata is an Array{Float32, 2} on julia 0.3 (perfect!) but a PyObject array([[...]], dtype=float32) on julia 0.4. Calling convert(Matrix{Float32}, pyimdata) gives exactly the same error as in #99:

ERROR: PyError (:PyFloat_AsDouble) <type 'exceptions.TypeError'>
TypeError('only length-1 arrays can be converted to Python scalars',)
...

v0.4 info:

Julia Version 0.4.0-dev+3307
Commit a387920* (2015-02-11 21:42 UTC)
Platform Info:
  System: Linux (i686-linux-gnu)
  CPU: Intel(R) Core(TM) i7-2630QM CPU @ 2.00GHz
  WORD_SIZE: 32
  BLAS: libopenblas (DYNAMIC_ARCH NO_AFFINITY Nehalem)
  LAPACK: libopenblas
  LIBM: libopenlibm
  LLVM: libLLVM-3.3

[stevengj]

I can't reproduce. Can you try a Pkg.update() to see if this was fixed recently?

[swt30]

Yeah, I can no longer reproduce it either. This seems to have been fixed somewhere along the line - but thanks for looking into it!