Unable to install PyPlot (ResolvePackageNotFound conda error)

[Octogonapus]

I'm trying to install the latest version of PyPlot on Julia v1.9.1 on Ubuntu 22.04 and getting this error:

julia> using PyPlot
[ Info: Installing matplotlib via the Conda matplotlib package...
[ Info: Running `conda install -y matplotlib` in root environment
Collecting package metadata (current_repodata.json): done
Solving environment: unsuccessful initial attempt using frozen solve. Retrying with flexible solve.
Solving environment: unsuccessful attempt using repodata from current_repodata.json, retrying with next repodata source.

ResolvePackageNotFound: 
  - conda==23.1.0

Here's a Dockerfile you can build to reproduce the error:

FROM ubuntu:22.04

RUN apt-get update -y && \
    apt-get upgrade -y && \
    apt-get install -y wget && \
    rm -rf /var/lib/apt/lists/*

RUN wget -nv https://julialang-s3.julialang.org/bin/linux/x64/1.9/julia-1.9.1-linux-x86_64.tar.gz && \
    tar xf julia-1.9.1-linux-x86_64.tar.gz && \
    rm julia-1.9.1-linux-x86_64.tar.gz && \
    ln -s /julia-1.9.1/bin/julia /usr/local/bin/julia

ENV PYTHON=""
RUN julia -e 'import Pkg; Pkg.add("PyPlot")'
RUN julia -e 'using PyPlot'

And here is the full error:

 > [5/5] RUN julia -e 'using PyPlot':                                                                                                                                                                               
#0 1.709 [ Info: Installing matplotlib via the Conda matplotlib package...                                                                                                                                          
#0 1.709 [ Info: Running `conda install -y matplotlib` in root environment                                                                                                                                          
#0 2.245 Collecting package metadata (current_repodata.json): ...working... done                                                                                                                                    
#0 21.68 Solving environment: ...working... unsuccessful initial attempt using frozen solve. Retrying with flexible solve.                                                                                          
#0 29.68 Solving environment: ...working... unsuccessful attempt using repodata from current_repodata.json, retrying with next repodata source.
#0 34.31 
#0 34.31 ResolvePackageNotFound: 
#0 34.31   - conda==23.1.0
#0 34.31 
#0 35.52 ERROR: InitError: failed process: Process(setenv(`/root/.julia/conda/3/x86_64/bin/conda install -y matplotlib`,["PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin", "CONDA_PREFIX=/root/.julia/conda/3/x86_64", "PWD=/", "HOME=/root", "CONDARC=/root/.julia/conda/3/x86_64/condarc-julia.yml", "OPENBLAS_MAIN_FREE=1", "PYTHONIOENCODING=UTF-8", "OPENBLAS_DEFAULT_NUM_THREADS=1"]), ProcessExited(1)) [1]
#0 37.01 
#0 37.01 Stacktrace:
#0 37.01   [1] pipeline_error
#0 37.01     @ ./process.jl:565 [inlined]
#0 37.01   [2] run(::Cmd; wait::Bool)
#0 37.17     @ Base ./process.jl:480
#0 37.17   [3] run
#0 37.17     @ ./process.jl:477 [inlined]
#0 37.17   [4] runconda(args::Cmd, env::String)
#0 37.17     @ Conda ~/.julia/packages/Conda/xMClC/src/Conda.jl:128
#0 37.17   [5] add(pkg::String, env::String; channel::String, satisfied_skip_solve::Bool, args::Cmd)
#0 37.17     @ Conda ~/.julia/packages/Conda/xMClC/src/Conda.jl:229
#0 37.17   [6] add (repeats 2 times)
#0 37.17     @ ~/.julia/packages/Conda/xMClC/src/Conda.jl:222 [inlined]
#0 37.17   [7] pyimport_conda
#0 37.17     @ ~/.julia/packages/PyCall/ilqDX/src/PyCall.jl:721
#0 37.17   [8] pyimport_conda
#0 37.17     @ ~/.julia/packages/PyCall/ilqDX/src/PyCall.jl:715 [inlined]
#0 37.17   [9] __init__()
#0 37.17     @ PyPlot ~/.julia/packages/PyPlot/H01LC/src/init.jl:173
#0 37.17  [10] register_restored_modules(sv::Core.SimpleVector, pkg::Base.PkgId, path::String)
#0 37.17     @ Base ./loading.jl:1115
#0 37.17  [11] _include_from_serialized(pkg::Base.PkgId, path::String, ocachepath::String, depmods::Vector{Any})
#0 37.17     @ Base ./loading.jl:1061
#0 37.17  [12] _require_search_from_serialized(pkg::Base.PkgId, sourcepath::String, build_id::UInt128)
#0 37.17     @ Base ./loading.jl:1506
#0 37.17  [13] _require(pkg::Base.PkgId, env::String)
#0 37.17     @ Base ./loading.jl:1783
#0 37.17  [14] _require_prelocked(uuidkey::Base.PkgId, env::String)
#0 37.17     @ Base ./loading.jl:1660
#0 37.17  [15] macro expansion
#0 37.17     @ ./loading.jl:1648 [inlined]
#0 37.17  [16] macro expansion
#0 37.17     @ ./lock.jl:267 [inlined]
#0 37.17  [17] require(into::Module, mod::Symbol)
#0 37.17     @ Base ./loading.jl:1611
#0 37.17 during initialization of module PyPlot
#0 37.18 
#0 37.18 caused by: PyError (PyImport_ImportModule
#0 37.26 
#0 37.26 The Python package matplotlib could not be imported by pyimport. Usually this means
#0 37.26 that you did not install matplotlib in the Python version being used by PyCall.
#0 37.26 
#0 37.26 PyCall is currently configured to use the Julia-specific Python distribution
#0 37.26 installed by the Conda.jl package.  To install the matplotlib module, you can
#0 37.26 use `pyimport_conda("matplotlib", PKG)`, where PKG is the Anaconda
#0 37.26 package that contains the module matplotlib, or alternatively you can use the
#0 37.26 Conda package directly (via `using Conda` followed by `Conda.add` etcetera).
#0 37.26 
#0 37.26 Alternatively, if you want to use a different Python distribution on your
#0 37.26 system, such as a system-wide Python (as opposed to the Julia-specific Python),
#0 37.26 you can re-configure PyCall with that Python.   As explained in the PyCall
#0 37.26 documentation, set ENV["PYTHON"] to the path/name of the python executable
#0 37.26 you want to use, run Pkg.build("PyCall"), and re-launch Julia.
#0 37.26 
#0 37.26 ) <class 'ModuleNotFoundError'>
#0 37.26 ModuleNotFoundError("No module named 'matplotlib'")
#0 37.26 
#0 37.26 Stacktrace:
#0 37.26   [1] pyimport(name::String)
#0 37.26     @ PyCall ~/.julia/packages/PyCall/ilqDX/src/PyCall.jl:558
#0 37.26   [2] pyimport_conda
#0 37.26     @ ~/.julia/packages/PyCall/ilqDX/src/PyCall.jl:716
#0 37.26   [3] pyimport_conda
#0 37.26     @ ~/.julia/packages/PyCall/ilqDX/src/PyCall.jl:715 [inlined]
#0 37.26   [4] __init__()
#0 37.26     @ PyPlot ~/.julia/packages/PyPlot/H01LC/src/init.jl:173
#0 37.26   [5] register_restored_modules(sv::Core.SimpleVector, pkg::Base.PkgId, path::String)
#0 37.26     @ Base ./loading.jl:1115
#0 37.26   [6] _include_from_serialized(pkg::Base.PkgId, path::String, ocachepath::String, depmods::Vector{Any})
#0 37.26     @ Base ./loading.jl:1061
#0 37.26   [7] _require_search_from_serialized(pkg::Base.PkgId, sourcepath::String, build_id::UInt128)
#0 37.26     @ Base ./loading.jl:1506
#0 37.26   [8] _require(pkg::Base.PkgId, env::String)
#0 37.26     @ Base ./loading.jl:1783
#0 37.26   [9] _require_prelocked(uuidkey::Base.PkgId, env::String)
#0 37.26     @ Base ./loading.jl:1660
#0 37.26  [10] macro expansion
#0 37.26     @ ./loading.jl:1648 [inlined]
#0 37.26  [11] macro expansion
#0 37.26     @ ./lock.jl:267 [inlined]
#0 37.26  [12] require(into::Module, mod::Symbol)
#0 37.26     @ Base ./loading.jl:1611
------
Dockerfile:15
--------------------
  13 |     ENV PYTHON=""
  14 |     RUN julia -e 'import Pkg; Pkg.add("PyPlot")'
  15 | >>> RUN julia -e 'using PyPlot'
  16 |     
--------------------
ERROR: failed to solve: process "/bin/sh -c julia -e 'using PyPlot'" did not complete successfully: exit code: 1

[stevengj]

If you don't want PyPlot to install its own Conda distro and use that to install matplotlib, then you shouldn't use ENV PYTHON="".

[Octogonapus]

That's fair, my mistake. Here is our actual use case:

FROM ubuntu:22.04

RUN apt-get update -y && \
    apt-get upgrade -y && \
    apt-get install -y wget python3-venv && \
    rm -rf /var/lib/apt/lists/*

RUN wget -nv https://julialang-s3.julialang.org/bin/linux/x64/1.9/julia-1.9.1-linux-x86_64.tar.gz && \
    tar xf julia-1.9.1-linux-x86_64.tar.gz && \
    rm julia-1.9.1-linux-x86_64.tar.gz && \
    ln -s /julia-1.9.1/bin/julia /usr/local/bin/julia

RUN python3 -m venv venv
RUN . venv/bin/activate && \
    pip install -U pip && \
    pip install -U matplotlib==3.7.0
ENV PYTHON="/root/venv/bin/python"
RUN julia -e 'import Pkg; Pkg.add("PyPlot")'
RUN julia -e 'using PyPlot'

The same error occurs:

 > [7/7] RUN julia -e 'using PyPlot':                                                                                                                                                                               
#0 1.434 [ Info: Installing matplotlib via the Conda matplotlib package...                                                                                                                                          
#0 1.435 [ Info: Running `conda install -y matplotlib` in root environment                                                                                                                                          
#0 1.892 Collecting package metadata (current_repodata.json): ...working... done                                                                                                                                    
#0 17.04 Solving environment: ...working... unsuccessful initial attempt using frozen solve. Retrying with flexible solve.                                                                                          
#0 23.96 Solving environment: ...working... unsuccessful attempt using repodata from current_repodata.json, retrying with next repodata source.
#0 27.86 
#0 27.86 ResolvePackageNotFound: 
#0 27.86   - conda==23.1.0