Hm, that all looks fine. From your original post, it looks like something is wrong with the source code of the Compat.jl package–did you edit that by hand at some point? You may be able to resolve the issue by deleting ~/.julia/packages/Compat and then re-running:
I’ve deleted and then readded Compat. The adding has worked right, and I relaunch julia test_plot.jl which is the program whose code is in the link I’ve given above, and … I get an error (after at least one minute) telling me that I have to rebuild Conda. I’ve done, and I do same for PyPlot and Plots.
And now, that takes really long time to run, but I get this warning message :
┌ Warning: getindex(o::PyObject, s::AbstractString) is deprecated in favor of dot overloading (getproperty) so elements should now be accessed as e.g. o."s" instead of o["s"].
│ caller = top-level scope at none:0
└ @ Core none:0
but that seems working right. But now, when I run this :
julia> using Plots julia> plotly() Plots.PlotlyBackend() julia> plot(rand(5,5),linewidth=2,title="MyPlot")
That plots nothing, what did I do wrong ?
And, I’ve a question : using Plots takes long time (between two and ten seconds), is it normal ?
There was a recent update in the PyCall API — a bunch of Julia packages that use Python via PyCall will see deprecation warnings for a little while until they are updated. Plots.jl was just fixed, but SciKitLearn has not been updated yet:
Well, you could run the python executable through a pipe. For example, the IJulia package can launch the Python jupyter notebook program, and used to require jupyter command-line program to build in order to get the Jupyter version and install the kernel specification file, but it never used PyCall.