Open EmmanuelJeanBriand opened 1 year ago
Thanks for reporting. Unfortunately, I have no machine on which to test this (and we're not adding 10x the builds just to get tests on the platform where this is ocurring), so will have to defer to a suitably equipped member of @conda-forge/jupyterlab to debug further.
Since april 30, 2023, with Mac OS X 10.14.6
FYI, according to https://endoflife.date/macos (not official, but informative), OS X 10.14 was end-of-life about 1.5 years ago (Oct 2021). I'm not sure what the conda-forge policy is towards end-of-life OS versions, but I wouldn't be surprised if support was dropped.
Hello I got the same issue with macOS Mojave 10.14.6. Since I still have some not updated working environments I was able to track this down to a change of libffi version (and other related differences).
You can recover a working environment with last jupyterlab version by pinning libffi to 3.3.
micromamba create -f https://prefix.dev/envs/marcbernot/mojave/conda-lock.yml -n mojave
or
conda create -n testmojave python=3.9 libffi=3.3 notebook jupyterlab
If someone wants to investigate further, these are two environments with python=3.9.7. The first one is ok (libffi=3.3), the second one gives the segmentation fault (libffi>3.3).
micromamba create -f https://prefix.dev/envs/marcbernot/mojave/conda-lock.yml -n mojave
micromamba create -f https://prefix.dev/envs/marcbernot/mojave_nok/conda-lock.yml -n mojave_nok
A conda compare
of the two environments yields
cffi found but mismatch. Specification pkg: cffi==1.15.1=py39h131948b_3, Running pkg: cffi==1.14.6=py39hb71fe58_0
libffi found but mismatch. Specification pkg: libffi==3.4.2=h0d85af4_5, Running pkg: libffi==3.3=h046ec9c_2
openssl found but mismatch. Specification pkg: openssl==3.1.1=h8a1eda9_1, Running pkg: openssl==1.1.1u=h8a1eda9_0
pyobjc-core found but mismatch. Specification pkg: pyobjc-core==9.1.1=py39h0504f80_0, Running pkg: pyobjc-core==7.3=py39h89e85a6_2
pyobjc-framework-cocoa found but mismatch. Specification pkg: pyobjc-framework-cocoa==9.1.1=py39h0504f80_0, Running pkg: pyobjc-framework-cocoa==7.3=py39h6e9494a_1
python found but mismatch. Specification pkg: python==3.9.7=h38b4d05_3_cpython, Running pkg: python==3.9.7=hd187cdc_0_cpython
@marcbernot thanks for sharing your workaround
@marcbernot a big thank you!
I investigated a little further. The pyobjc-core package from conda-forge seems to be the culprit.
Indeed if you pin the pyobjc-core version to 8.2 everything is ok.
conda create -n mojave_py310_pyobjc82 python=3.10 pyobjc-core=8.2 notebook
It is starting to fail at 8.4
conda create -n mojave_py310_pyobjc84 python=3.10 pyobjc-core=8.4 notebook
However you can fix a failing environment (like mojave_py310_pyobjc84
) with
pip install pyobjc-core pyobjc-framework-cocoa --force-reinstall
So the package from pypi works where the one from conda-forge does not. All this looks related to this issue in the feedstock of the conda-forge pyobjc-core
package.
I think this issue can be closed as it is not related to jupyterlab
feedstock.
I relabelled the issue as question instead of closing it. So it is easier to discover in case an user hit this.
I investigated a little further. The pyobjc-core package from conda-forge seems to be the culprit.
Indeed if you pin the pyobjc-core version to 8.2 everything is ok.
conda create -n mojave_py310_pyobjc82 python=3.10 pyobjc-core=8.2 notebook
It is starting to fail at 8.4
conda create -n mojave_py310_pyobjc84 python=3.10 pyobjc-core=8.4 notebook
However you can fix a failing environment (like
mojave_py310_pyobjc84
) withpip install pyobjc-core pyobjc-framework-cocoa --force-reinstall
So the package from pypi works where the one from conda-forge does not. All this looks related to this issue in the feedstock of the conda-forge
pyobjc-core
package.I think this issue can be closed as it is not related to
jupyterlab
feedstock.
Yep that was the issue for me. Thanks!
I'm a complete noob, but I've found that the version of Anaconda3-2021.05-MacOSX-x86_64.pkg that I found on my TimeMachine backup doesn't cause problems on OS X 10.14. I found an older version for download here: https://docs.anaconda.com/free/anaconda/reference/packages/oldpkglists/ Hopefully this information will help someone.
I investigated a little further. The pyobjc-core package from conda-forge seems to be the culprit.
Indeed if you pin the pyobjc-core version to 8.2 everything is ok.
conda create -n mojave_py310_pyobjc82 python=3.10 pyobjc-core=8.2 notebook
It is starting to fail at 8.4
conda create -n mojave_py310_pyobjc84 python=3.10 pyobjc-core=8.4 notebook
However you can fix a failing environment (like
mojave_py310_pyobjc84
) withpip install pyobjc-core pyobjc-framework-cocoa --force-reinstall
So the package from pypi works where the one from conda-forge does not. All this looks related to this issue in the feedstock of the conda-forge
pyobjc-core
package.I think this issue can be closed as it is not related to
jupyterlab
feedstock.
This worked, thanks
I investigated a little further. The pyobjc-core package from conda-forge seems to be the culprit.
Indeed if you pin the pyobjc-core version to 8.2 everything is ok.
conda create -n mojave_py310_pyobjc82 python=3.10 pyobjc-core=8.2 notebook
It is starting to fail at 8.4
conda create -n mojave_py310_pyobjc84 python=3.10 pyobjc-core=8.4 notebook
However you can fix a failing environment (like
mojave_py310_pyobjc84
) withpip install pyobjc-core pyobjc-framework-cocoa --force-reinstall
So the package from pypi works where the one from conda-forge does not. All this looks related to this issue in the feedstock of the conda-forge
pyobjc-core
package.I think this issue can be closed as it is not related to
jupyterlab
feedstock.
Is there a way to apply this workaround if my jupyter setup was created via IJulia ?
I investigated a little further. The pyobjc-core package from conda-forge seems to be the culprit.
Indeed if you pin the pyobjc-core version to 8.2 everything is ok.
conda create -n mojave_py310_pyobjc82 python=3.10 pyobjc-core=8.2 notebook
It is starting to fail at 8.4
conda create -n mojave_py310_pyobjc84 python=3.10 pyobjc-core=8.4 notebook
However you can fix a failing environment (like
mojave_py310_pyobjc84
) withpip install pyobjc-core pyobjc-framework-cocoa --force-reinstall
So the package from pypi works where the one from conda-forge does not. All this looks related to this issue in the feedstock of the conda-forge
pyobjc-core
package.I think this issue can be closed as it is not related to
jupyterlab
feedstock.
This is the fix for me. thank you so much.
I'm a complete noob, but I've found that the version of Anaconda3-2021.05-MacOSX-x86_64.pkg that I found on my TimeMachine backup doesn't cause problems on OS X 10.14. I found an older version for download here: https://docs.anaconda.com/free/anaconda/reference/packages/oldpkglists/ Hopefully this information will help someone.
Thank you. This helped me. Downloading an older version is what worked after trying lots of other solutions.
Yes this worked for me, I noticed the problem seemed to start with the April 2023 update so I installed the October 2022 update. Works perfectly.
Solution to issue cannot be found in the documentation.
Issue
Since april 30, 2023, with Mac OS X 10.14.6 and jupyter installed from conda-forge, I cannot start anymore
jupyter lab
norjupyter notebook
; I getSegmentation fault: 11
.Other users report the same issue at:
Installed packages
Environment info