For me, Jupyter(-lab) fails to work with the most recent version of Mayavi 4.8.3. Other Python interpreter software (like Spyder, ipython, ...) works well. For me, this can be boiled down to a bare-bones-test-environment, as described next. 1) Create fresh test environment ~> conda create --name tes (This runs on Python 3.13.5 for me, but the issue persists for other versions of Python.) Activate ~> conda activate tes. Then conda install jupyter conda install mayavi conda install ipyevents conda install itkwidgets conda install mesalib conda install puremagic will install a bunch of dependencies. 2) In test environment, open (tes)~> jupyter-lab. Then do [1]: from mayavi import mlab mlab.init_notebook() Output is: Notebook initialized with ipy backend. [2]: mlab.figure() The expected output is a Mayavi figure canvas window. Instead: The jupyter-lab kernel dies and restarts with no indication why. 3) For sanity, open another interpreter in the same environment, e.g., ipython by, e.g., (tes)~> terminator --command="ipython --pylab". Then do In [1]: from mayavi import mlab In [2]: mlab.figure() This generates a Mayavi canvas window as expected. Same for Spyder. Conclusion: Jupyter breaks the most recent Mayavi 4.8.3 Any suggestions what is wrong? Continue reading...
