There have been issues in which the model providing packages that wrap python using PythonCall cannot successfully resolve a mutually compatible python environment. (It seems that PythonCall/CondaPkg do not provide for separate python environments for julia-python packages that are to be imported simultaneously. ) In that case calling update() will likely fail. The workaround is to skip those packages with update(skip=...) and then update each one separately, as in update("MLJScikitlearnInterface", manifest=false).
At time of writing, the effected packages are:
- MLJSckitlearnInterface
- OutlierDetectionPython
- CatBoost
There have been issues in which the model providing packages that wrap python using PythonCall cannot successfully resolve a mutually compatible python environment. (It seems that PythonCall/CondaPkg do not provide for separate python environments for julia-python packages that are to be imported simultaneously. ) In that case calling
update()will likely fail. The workaround is to skip those packages withupdate(skip=...)and then update each one separately, as inupdate("MLJScikitlearnInterface", manifest=false).At time of writing, the effected packages are: