Open cmcclellen opened 4 years ago
It is unclear whether this is an issue with jedi itself or not - it seems the code in jedi does not take into account the fact that project could be None
in this case, or perhaps it does not do that becuase that's not how its clients are supposed to behave.
Any way to downgrade things to a working version? Pretty annoying to work with the constant exceptions.
Any way to downgrade things to a working version? Pretty annoying to work with the constant exceptions.
yes! just run:
sudo pip install --upgrade jedi==0.16.0```
Encountered the same problem. The command above works, sudo
and --upgrade
may not be required. The following worked for me:
pip install jedi==0.16.0
It is unclear whether this is an issue with jedi itself or not - it seems the code in jedi does not take into account the fact that project could be
None
in this case, or perhaps it does not do that becuase that's not how its clients are supposed to behave.
Jedi's behaviour in 0.17 changed, the sys_path
argument to Script
is deprecated, and they now expect you to supply a project
https://github.com/davidhalter/jedi/blob/master/jedi/api/__init__.py#L132
Would be nice to continue supporting Jedi, MPLS doesn't work on installed packages, which is annoying.
I almost cried trying to fix this error, uninstalled anaconda, reinstalled python, rollbacked to python 3.7 instead of 3.8. Thank you, and it would be nice to fix this or add this to doc
It is a bug of Jedi 0.17.0 and the patch has been merged.
davidhalter/jedi/pull/1549
coc-python 1.2.12 has switched to the new api introduced by Jedi 0.17.0. Upgrading coc-python to 1.2.12 works well for me.
Looks like Jedi 0.17.0 was released yesterday and code completion in coc.nvim/coc-python is broken. Whenever it tries to complete something I get:
The above file path in the exception was edited for brevity.
Moving back to jedi 0.16.0 makes things work again.