Thanks for this extension! I have written a Python package which ships with an entrypoint to install the jupyter_app_launcher config into the respective Jupyter data directories. For conda-forge packaging, this will happen already during package build, so that installing my conda package will directly show me the icons in the launcher. I really like this approach for my use case, but I have one problem with it: If for some reason, some other package would do the exact same trick, one would overwrite the other's config.
A solution to this would be to not fix the name of the configuration file to config.yml, but only fix its location to PREFIX/jupyter_app_launcher. The code discovering the config file would then need to glob the path for potential configuration files and use all of them. What do you think?
Thanks for this extension! I have written a Python package which ships with an entrypoint to install the jupyter_app_launcher config into the respective Jupyter data directories. For conda-forge packaging, this will happen already during package build, so that installing my conda package will directly show me the icons in the launcher. I really like this approach for my use case, but I have one problem with it: If for some reason, some other package would do the exact same trick, one would overwrite the other's config.
A solution to this would be to not fix the name of the configuration file to
config.yml
, but only fix its location toPREFIX/jupyter_app_launcher
. The code discovering the config file would then need to glob the path for potential configuration files and use all of them. What do you think?