Hi everyone,
I decided to give my MacBookPro (13’’ 2019 Intel) a factory reset and now have a nice and clean set-up. Everything works fine, except importing PyPlot in Julia. Unfortunately, I don’t get any error message other than this
julia> import PyPlot
zsh: killed Julia
because Julia is just dead. Any ideas what could be wrong? I am using Julia 1.12.2 and PyPlot v2.11.6.
Thank you very much,
Georg
What is the output of
versioninfo()
at the Julia prompt?
It gives more details than just the version. That might be useful.
And what is your MacOS version?
The output is
ulia Version 1.12.2
Commit ca9b6662be4 (2025-11-20 16:25 UTC)
Build Info:
Official https://julialang.org release
Platform Info:
OS: macOS (x86_64-apple-darwin24.0.0)
CPU: 8 × Intel(R) Core(TM) i5-8279U CPU @ 2.40GHz
WORD_SIZE: 64
LLVM: libLLVM-18.1.7 (ORCJIT, skylake)
GC: Built with stock GC
Threads: 1 default, 1 interactive, 1 GC (on 8 virtual cores)
Mac OS is Sequoia 15.7.2
Can you try these instructions:
rm -rf ~/.julia/conda
mkdir MyProject
cd MyProject
julia --project=.
And on the Julia prompt:
using Pkg
ENV["PYTHON"]=""
pkg"add PyPlot"
This results in the following error:
julia> import PyPlot
ERROR: InitError: could not load library "/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib"
dlopen(/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib, 0x0009): tried: '/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib' (no such file), '/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib' (no such file)
Stacktrace:
[1] #dlopen#3
@ ./libdl.jl:120
[2] dlopen
@ ./libdl.jl:119 [inlined]
[3] __init__()
@ PyCall ~/.julia/packages/PyCall/1gn3u/src/pyinit.jl:149
[4] run_module_init(mod::Module, i::Int64)
@ Base ./loading.jl:1440
[5] register_restored_modules(sv::Core.SimpleVector, pkg::Base.PkgId, path::String)
@ Base ./loading.jl:1428
[6] _include_from_serialized(pkg::Base.PkgId, path::String, ocachepath::String, depmods::Vector{Any}; register::Bool)
@ Base ./loading.jl:1316
[7] _include_from_serialized
@ ./loading.jl:1271 [inlined]
[8] _require_search_from_serialized(pkg::Base.PkgId, sourcepath::String, build_id::UInt128, stalecheck::Bool; reasons::Dict{…}, DEPOT_PATH::Vector{…})
@ Base ./loading.jl:2099
[9] _require_search_from_serialized
@ ./loading.jl:2006 [inlined]
[10] __require_prelocked(pkg::Base.PkgId, env::String)
@ Base ./loading.jl:2624
[11] _require_prelocked(uuidkey::Base.PkgId, env::String)
@ Base ./loading.jl:2490
[12] macro expansion
@ ./loading.jl:2418 [inlined]
[13] macro expansion
@ ./lock.jl:376 [inlined]
[14] __require(into::Module, mod::Symbol)
@ Base ./loading.jl:2383
[15] require(into::Module, mod::Symbol)
@ Base ./loading.jl:2359
[16] top-level scope
@ REPL[4]:1
during initialization of module PyCall
Some type information was truncated. Use `show(err)` to see complete types.
I have had plenty of problems with PyPlot since Julia 1.12, probably related to threads. In the end I migrated to PythonPlot.
Unfortunately this did not help. Adding it resulted in
(@v1.12) pkg> add PythonPlot
Resolving package versions...
Updating `~/.julia/environments/v1.12/Project.toml`
[274fc56d] + PythonPlot v1.0.6
Updating `~/.julia/environments/v1.12/Manifest.toml`
[3da002f7] + ColorTypes v0.12.1
[5ae59095] + Colors v0.13.1
[992eb4ea] + CondaPkg v0.2.33
[53c48c17] + FixedPointNumbers v0.8.5
[0f8b85d8] + JSON3 v1.14.3
[1914dd2f] + MacroTools v0.5.16
[0b3b1443] + MicroMamba v0.1.14
[fa939f87] + Pidfile v1.3.0
[6099a3de] + PythonCall v0.9.30
[274fc56d] + PythonPlot v1.0.6
[6c6a2e73] + Scratch v1.3.0
[856f2bd8] + StructTypes v1.11.0
[e17b2a0c] + UnsafePointers v1.0.0
[81def892] + VersionParsing v1.3.0
⌅ [f8abcde7] + micromamba_jll v1.5.12+0
[4d7b5844] + pixi_jll v0.41.3+0
[0dad84c5] + ArgTools v1.1.2
[f43a241f] + Downloads v1.7.0
[7b1f6079] + FileWatching v1.11.0
[4af54fe1] + LazyArtifacts v1.11.0
[b27032c2] + LibCURL v0.6.4
[76f85450] + LibGit2 v1.11.0
[44cfe95a] + Pkg v1.12.0
[a4e569a6] + Tar v1.10.0
[deac9b47] + LibCURL_jll v8.15.0+0
[e37daf67] + LibGit2_jll v1.9.0+0
[29816b5a] + LibSSH2_jll v1.11.3+1
[8e850ede] + nghttp2_jll v1.64.0+1
[3f19e933] + p7zip_jll v17.7.0+0
Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`
Now importing the package kills the Julia again. Also removing the package, garbage collecting, and adding it again does not fix the problem. Although, there is no error message when adding it. However, it just crashes.
Do you have similar errors with PyCall.jl and/or PythonCall.jl ?
Adding and importing PythonCall works. After adding PyCall, however, import PyCall results in
julia> import PyCall
ERROR: InitError: could not load library "/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib"
dlopen(/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib, 0x0009): tried: '/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib' (no such file), '/Users/georg/.julia/conda/3/x86_64/lib/libpython3.12.dylib' (no such file)
Stacktrace:
[1] #dlopen#3
@ ./libdl.jl:120
[2] dlopen
@ ./libdl.jl:119 [inlined]
[3] __init__()
@ PyCall ~/.julia/packages/PyCall/1gn3u/src/pyinit.jl:149
[4] run_module_init(mod::Module, i::Int64)
@ Base ./loading.jl:1440
[5] register_restored_modules(sv::Core.SimpleVector, pkg::Base.PkgId, path::String)
@ Base ./loading.jl:1428
[6] _include_from_serialized(pkg::Base.PkgId, path::String, ocachepath::String, depmods::Vector{Any}; register::Bool)
@ Base ./loading.jl:1316
[7] _include_from_serialized
@ ./loading.jl:1271 [inlined]
[8] _require_search_from_serialized(pkg::Base.PkgId, sourcepath::String, build_id::UInt128, stalecheck::Bool; reasons::Dict{…}, DEPOT_PATH::Vector{…})
@ Base ./loading.jl:2112
[9] _require_search_from_serialized
@ ./loading.jl:2006 [inlined]
[10] __require_prelocked(pkg::Base.PkgId, env::String)
@ Base ./loading.jl:2624
[11] _require_prelocked(uuidkey::Base.PkgId, env::String)
@ Base ./loading.jl:2490
[12] macro expansion
@ ./loading.jl:2418 [inlined]
[13] macro expansion
@ ./lock.jl:376 [inlined]
[14] __require(into::Module, mod::Symbol)
@ Base ./loading.jl:2383
[15] require(into::Module, mod::Symbol)
@ Base ./loading.jl:2359
[16] top-level scope
@ REPL[6]:1
during initialization of module PyCall
Some type information was truncated. Use `show(err)` to see complete types.
Did you consider setting the Python path manually and rebuilding PyCall.jl?
You may also need to install matplotlib in your python environment.
I didn’t.
I am not sure what to set it to. Currently there is no python installation on my machine. To my understanding, this why the Julia packages try building a minimal python installation using miniconda.
There is currently also no entry "Python" in my ENV.
julia> ENV
Base.EnvDict with 29 entries:
"__CFBundleIdentifier" => "com.apple.Terminal"
"TMPDIR" => "/var/folders/cd/gnnpy0j96vxc99wdm1p03h5m0000gn/T/"
"XPC_FLAGS" => "0x0"
"LaunchInstanceID" => "3A30F679-31A4-4ECD-A483-001718CBA49F"
"TERM" => "xterm-256color"
"SSH_AUTH_SOCK" => "/private/tmp/com.apple.launchd.P1NsbIKk3B/Listeners"
"SECURITYSESSIONID" => "186a1"
"XPC_SERVICE_NAME" => "0"
"TERM_PROGRAM" => "Apple_Terminal"
"TERM_PROGRAM_VERSION" => "455.1"
"TERM_SESSION_ID" => "38EAAAA0-1A63-4AF6-9FFA-5C6F84A62482"
"SHELL" => "/bin/zsh"
"HOME" => "/Users/georg"
"LOGNAME" => "georg"
"USER" => "georg"
"PATH" => "/Users/georg/.julia/environments/v1.12/.CondaPkg/.pixi/envs/default/bin:/Users/…
"SHLVL" => "1"
"PWD" => "/Users/georg"
"OLDPWD" => "/Users/georg/.julia/conda/3/x86_64/lib"
"LANG" => "de_DE.UTF-8"
"_" => "/Users/georg/.juliaup/bin/julia"
"__CF_USER_TEXT_ENCODING" => "0x1F5:0x0:0x3"
"OPENBLAS_MAIN_FREE" => "1"
"OPENBLAS_DEFAULT_NUM_THREADS" => "1"
"CONDA_PREFIX" => "/Users/georg/.julia/environments/v1.12/.CondaPkg/.pixi/envs/default"
"CONDA_DEFAULT_ENV" => "/Users/georg/.julia/environments/v1.12/.CondaPkg/.pixi/envs/default"
"CONDA_SHLVL" => "1"
"CONDA_PROMPT_MODIFIER" => "(/Users/georg/.julia/environments/v1.12/.CondaPkg/.pixi/envs/default) "
"PYTHON_JULIACALL_INIT" => "no"
Should I try installing some Python to my machine and then link it?
For rebuilding PyCall you could try, before adding PyPlot:
julia -e 'ENV["PYTHON"]=""; using Pkg; Pkg.add("PyCall"); Pkg.build("PyCall")'
This installs Python in a subfolder of .julia.
In case that threading should cause a problem you can start Julia with:
julia -t 1
which disables the interactive thread introduced in Julia 1.12.
Unfortunately, this didn’t help. Adding and building PyCall works. starting then using Julia -t 1and adding PyPlot also works. But importing PyPlots results in the following error:
julia> import PyPlot
[ Info: Installing matplotlib via the Conda matplotlib package...
[ Info: Running `conda install -y matplotlib` in root environment
Channels:
- conda-forge
Platform: osx-64
Collecting package metadata (repodata.json): done
Solving environment: done
==> WARNING: A newer version of conda exists. <==
current version: 25.9.1
latest version: 25.11.0
Please update conda by running
$ conda update -n base -c conda-forge conda
## Package Plan ##
environment location: /Users/georg/.julia/conda/3/x86_64
added / updated specs:
- matplotlib
The following packages will be downloaded:
package | build
---------------------------|-----------------
brotli-1.1.0 | h1c43f85_4 20 KB conda-forge
brotli-bin-1.1.0 | h1c43f85_4 17 KB conda-forge
contourpy-1.3.3 | py312hd099df3_3 263 KB conda-forge
cycler-0.12.1 | pyhd8ed1ab_1 13 KB conda-forge
fonttools-4.60.1 | py312hacf3034_0 2.7 MB conda-forge
freetype-2.14.1 | h694c41f_0 170 KB conda-forge
kiwisolver-1.4.9 | py312h90e26e8_2 67 KB conda-forge
lcms2-2.17 | h72f5680_0 221 KB conda-forge
lerc-4.0.0 | hcca01a6_1 243 KB conda-forge
libbrotlicommon-1.1.0 | h1c43f85_4 66 KB conda-forge
libbrotlidec-1.1.0 | h1c43f85_4 30 KB conda-forge
libbrotlienc-1.1.0 | h1c43f85_4 288 KB conda-forge
libdeflate-1.25 | h517ebb2_0 69 KB conda-forge
libfreetype-2.14.1 | h694c41f_0 8 KB conda-forge
libfreetype6-2.14.1 | h6912278_0 366 KB conda-forge
libjpeg-turbo-3.1.2 | h8616949_0 572 KB conda-forge
libpng-1.6.51 | h380d223_0 292 KB conda-forge
libtiff-4.7.1 | ha0a348c_1 395 KB conda-forge
libwebp-base-1.6.0 | hb807250_0 357 KB conda-forge
libxcb-1.17.0 | hf1f96e2_0 316 KB conda-forge
matplotlib-3.10.8 | py312hb401068_0 17 KB conda-forge
matplotlib-base-3.10.8 | py312h7894933_0 7.9 MB conda-forge
munkres-1.1.4 | pyhd8ed1ab_1 15 KB conda-forge
openjpeg-2.5.4 | h87e8dc5_0 327 KB conda-forge
pillow-12.0.0 | py312h4148a4b_1 939 KB conda-forge
pthread-stubs-0.4 | h00291cd_1002 8 KB conda-forge
pyparsing-3.2.5 | pyhcf101f3_0 102 KB conda-forge
python-dateutil-2.9.0.post0| pyhe01879c_2 228 KB conda-forge
qhull-2020.2 | h3c5361c_5 516 KB conda-forge
six-1.17.0 | pyhe01879c_1 18 KB conda-forge
tornado-6.5.2 | py312h80b0991_2 829 KB conda-forge
unicodedata2-17.0.0 | py312h80b0991_1 394 KB conda-forge
xorg-libxau-1.0.12 | h8616949_1 13 KB conda-forge
xorg-libxdmcp-1.1.5 | h8616949_1 19 KB conda-forge
zlib-ng-2.2.5 | h55e386d_0 107 KB conda-forge
------------------------------------------------------------
Total: 17.8 MB
The following NEW packages will be INSTALLED:
brotli conda-forge/osx-64::brotli-1.1.0-h1c43f85_4
brotli-bin conda-forge/osx-64::brotli-bin-1.1.0-h1c43f85_4
contourpy conda-forge/osx-64::contourpy-1.3.3-py312hd099df3_3
cycler conda-forge/noarch::cycler-0.12.1-pyhd8ed1ab_1
fonttools conda-forge/osx-64::fonttools-4.60.1-py312hacf3034_0
freetype conda-forge/osx-64::freetype-2.14.1-h694c41f_0
kiwisolver conda-forge/osx-64::kiwisolver-1.4.9-py312h90e26e8_2
lcms2 conda-forge/osx-64::lcms2-2.17-h72f5680_0
lerc conda-forge/osx-64::lerc-4.0.0-hcca01a6_1
libbrotlicommon conda-forge/osx-64::libbrotlicommon-1.1.0-h1c43f85_4
libbrotlidec conda-forge/osx-64::libbrotlidec-1.1.0-h1c43f85_4
libbrotlienc conda-forge/osx-64::libbrotlienc-1.1.0-h1c43f85_4
libdeflate conda-forge/osx-64::libdeflate-1.25-h517ebb2_0
libfreetype conda-forge/osx-64::libfreetype-2.14.1-h694c41f_0
libfreetype6 conda-forge/osx-64::libfreetype6-2.14.1-h6912278_0
libjpeg-turbo conda-forge/osx-64::libjpeg-turbo-3.1.2-h8616949_0
libpng conda-forge/osx-64::libpng-1.6.51-h380d223_0
libtiff conda-forge/osx-64::libtiff-4.7.1-ha0a348c_1
libwebp-base conda-forge/osx-64::libwebp-base-1.6.0-hb807250_0
libxcb conda-forge/osx-64::libxcb-1.17.0-hf1f96e2_0
matplotlib conda-forge/osx-64::matplotlib-3.10.8-py312hb401068_0
matplotlib-base conda-forge/osx-64::matplotlib-base-3.10.8-py312h7894933_0
munkres conda-forge/noarch::munkres-1.1.4-pyhd8ed1ab_1
openjpeg conda-forge/osx-64::openjpeg-2.5.4-h87e8dc5_0
pillow conda-forge/osx-64::pillow-12.0.0-py312h4148a4b_1
pthread-stubs conda-forge/osx-64::pthread-stubs-0.4-h00291cd_1002
pyparsing conda-forge/noarch::pyparsing-3.2.5-pyhcf101f3_0
python-dateutil conda-forge/noarch::python-dateutil-2.9.0.post0-pyhe01879c_2
qhull conda-forge/osx-64::qhull-2020.2-h3c5361c_5
six conda-forge/noarch::six-1.17.0-pyhe01879c_1
tornado conda-forge/osx-64::tornado-6.5.2-py312h80b0991_2
unicodedata2 conda-forge/osx-64::unicodedata2-17.0.0-py312h80b0991_1
xorg-libxau conda-forge/osx-64::xorg-libxau-1.0.12-h8616949_1
xorg-libxdmcp conda-forge/osx-64::xorg-libxdmcp-1.1.5-h8616949_1
zlib-ng conda-forge/osx-64::zlib-ng-2.2.5-h55e386d_0
Downloading and Extracting Packages:
Preparing transaction: done
Verifying transaction: done
Executing transaction: done
zsh: killed julia -t 1
Have a look at: using PyPlot hangs/kills julia 1.8.5 or 1.9.0 on MacOS · Issue #567 · JuliaPy/PyPlot.jl · GitHub
Try the last suggestion there:
If you are using Python via Conda.jl (which is the default on MacOS), you could try removing conda (rm -rf ~/.julia/conda) and rebuilding PyCall and PyPlot.
If that doesn’t help, create a new issue (you are probably using a newer Julia version than 1.8.5 or 1.9.0).
I know that PyPlot on macOS works for many people, but sometimes a Conda or Matplotlib update might temporarily break it. I don’t have a Mac for testing.
Thanks every one for the fruitful suggestions. I managed to solve the issue. As mentioned by several of you, the main issue is the handling of Python on Mac OS. What helped was
ENV[PYTHON]to point to the prepared Python installationAs I just noticed this is one of the possibilities pointed out in the installation instructions on the PyPlot GitHub page. (Though, they have several suggestions, and only this one worked for me).
Thanks again to everyone for the support.