PyCall installation problem?


#1

Hi,
I’m trying to update PyPlot and IJulia, but the update/addition stalls – it seems like the problem is PyCall:

julia> Pkg.add(“PyCall”)
INFO: Installing PyCall v1.15.0
INFO: Building Conda
INFO: Building PyCall
INFO: Using the Python distribution in the Conda package by default.
To use a different Python version, set ENV[“PYTHON”]=“pythoncommand” and re-run Pkg.build(“PyCall”).
INFO: Downloading miniconda installer …
INFO: Installing miniconda …
INFO: Downloading miniconda installer …
INFO: Installing miniconda …
INFO: Downloading miniconda installer …
INFO: Installing miniconda …
… and then ad infinitum…

Any clues? [I do have both Anaconda 2.7 and 3.x installed for doing some Python work, but that did not create problems in the past.]


#2

Does

using Conda
Conda.add("numpy")

work?


#3

Yes,

using Conda

Conda.add(“numpy”)

works.

-B


#4

When I did

using Conda
Conda.add(“numpy”)

the appearance in the REPL was similar to using conda in a Python shell

After that, I did:

Pkg.add(“PyPlot”)

and the REPL appearance was again similar to when using conda (growing bars indicating state of installation, etc.)

Next, installing IJulia, etc. also worked – in summary most things work again. Thanks! [I don’t know what the difference between Conda.add() and Pkg.add(), or why things work now.]


#5

Pkg.* functions are managing Julia packages, while Conda.* is for the packages space managed by conda. Eg python uses the conda, but it can work with other languages, eg R. Conda.jl is Julia’s conda interface for some non-Julia dependencies.


#6

Thanks for clarification. I assume that for packages managed by conda (presumably PyCall, numpy, PyPlot, Ijulia, etc., I can use either Conda.add() or Pkg.add()… (?)

-B


#7

No. Pkg.add() does not directly manage packages by conda. It may do so indirectly, if the build script of the relevant Julia package asks for it.


#8

The PyCall build script just calls Conda.add("numpy"), so I’m not sure why it would be different when you ran it manually. Perhaps it was a temporary network glitch?