You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Dependencies for jupyter notebook support are by definition optional dependencies and should not be listed in required dependencies, but in optional dependencies as officially recommended by PyPA and described here.
Please consider removing nbconvert and ipython from the required dependencies and moving them to optional dependencies.
It makes absolutely no sense to:
install an entire python interactive suite for simple static code analysis
Add 45 packages, bringing from only 7 to 52 total dependencies, i.e. multiply the number of dependencies by 7, and also multiply the total package size by 16 (from 2.15Mib to 32Mib) for a single feature that is not even the default behavior but accessible via an option that needs to be explicitly specified, and used by a minority of users.
Tell users to install your package with --no-deps and then install manually the actual required dependencies by hand, in order to get what should be the default install process.
Make these users receive pip error messages because it then believes that the package has not been installed correctly as the environment is missing falsely "required" dependencies for pipreqs
Please simply use a optional dependencies block as is intended, recommended, the best practice used by all major python projects, and frankly the only logical solution to this dependency management, and all these issues will be resolved with 5 lines of code.
@alan-barzilay @mateuslatrova @Fernando-crz
Dependencies for jupyter notebook support are by definition optional dependencies and should not be listed in required dependencies, but in optional dependencies as officially recommended by PyPA and described here.
Please consider removing
nbconvert
andipython
from the required dependencies and moving them to optional dependencies.It makes absolutely no sense to:
--no-deps
and then install manually the actual required dependencies by hand, in order to get what should be the default install process.pip
error messages because it then believes that the package has not been installed correctly as the environment is missing falsely "required" dependencies forpipreqs
Please simply use a optional dependencies block as is intended, recommended, the best practice used by all major python projects, and frankly the only logical solution to this dependency management, and all these issues will be resolved with 5 lines of code.
PS: For reference ->
with jupyter support
without jupyter support
(need to remove
pip
andsetuptools
from the total)The text was updated successfully, but these errors were encountered: