I have newly updated my ubuntu os to 21.04 (Xorg), I am struggling using c++ in the notebook. The kernel is failing repeatedly. I am attaching the errors.
You said you installed ROOT via conda, but it seems you are using a root installation in your home directory (/home/priyanka/root_install) and you are setting your environment with thisroot.sh?
Perhaps there are two installations? If you are using the conda one, you just need to activate the conda environment (no thisroot.sh needed).
Hi,
I have already installed following it, but the issue is “kernel failed”. This I only face when i use ROOT c++ instead of ipykernel. Python works fine but c++ doesn’t.
From the first image you sent it seems there is some interference between the conda ROOT and the other ROOT you installed - you activate the conda environment but the error trace shows that the other ROOT is kicking in (the one installed in your home).
Please make sure that no environment is set for the other ROOT when you activate the conda environment for the conda ROOT.
To make sure no other env is set or no interference, what i should recheck ? I already commented the source …/…/…thisroot.sh from the bashrc. Still it’s not working. The same error, kernel died and can’t load library error i am getting.
I’d check LD_LIBRARY_PATH and PATH to see if they point to the ROOT in your home dir, it must be pointed to from somewhere since it ends up being used.
As you can see both LD_LIBRARY_PATH and PATH contain your other ROOT installation (the one in /home/priyanka/root_install, that explains the interference. I’d start a fresh terminal session that doesn’t have these settings.
If it won’t change with a new terminal, it’s because you have that setting somewhere (e.g. bashrc), so I’d remove the setting from there.
If you can’t find how to remove the setting, just set those variables to a value that does not contain the ROOT home directory installation (export LD_LIBRARY_PATH=...). They have to contain what conda set, though, otherwise conda won’t work.
I have uninstalled root, which i build from source. Now only root with conda env remains. But i can’t see the ROOT c++ kernel in the notebook. Screenshot attached.
To make sure i am not doing anything wrong i have removed the root env from conda and reinstalled it, screenshot is attached. Only python package is installed and again there is not ROOT c++ kernel only ipykernel is there. Is there some other settings for ROOT c++, please let me know.
One last thing to try, is to start Jupyter with root --notebook instead of jupyter, and see if that shows the ROOT C++ kernel. Otherwise I’ll ask the maintainer of the ROOT conda package is he has any idea of what can be wrong.
yes root --notebook is working. I already tried two days back following this https://anaconda.org/conda-forge/root/ link. I don’t know why jupyter didn’t work. Anyway, thanks a lot for your help.