Check configuration value types strictly and warn on unknown configuration
Examples in the repository now use Uv
0.8.1
Fix problem trying to forward SIGCHLD to kernel, which was maybe specific to
Python 3.8 (and not-windows) (#24)
0.8.0
Change how hatch projects are run: detect the (default) virtual env path
and run python from there. User has to ensure virtualenv dependencies are
synchronized. (#23)
0.7.2
Published documentation to a new website using mkdocs
Copyedited documentation
0.7.1
Use Uv as first fallback if no explicit project manager can be identified
from the pyproject if uv is in the command path. Use Rye as the second
fallback in the same way.
0.7.0
Use uv run for uv projects. Requires uv 0.2.29 or later.
Uses uv run --with ipykernel which means it will run as
before if ipykernel is already installed, or run in an overlay
environment (ipykernel + base pyproject) if not.
Include kernel start failure messages in user visible text in more cases such
as “command not found”.
0.6.0
Support configuration use-venv for setting a name of a virtualenv to use
Support Uv (in a basic way, by assuming use-venv=".venv" for uv)
0.5.5
Fix ProjectDetection.path which should hold the path to the pyproject file
Added trusted publishing (for PyPI) workflow on github
0.5.4
Enable debugging for the Pyproject Local kernel (just like the regular
ipykernel)
More extensive tests, on linux and windows, including notebook execution
0.5.3
Add tests. Moved main logic to module to facilitate.
Update readme
0.5.2
Use tomllib for newer Python, and tomli as fallback
Catch all exceptions when starting the kernel, so that the fallback can be
started in most situations.
More detailed error messages in the fallback
0.5.1
Cleaned up unnecessary log output
Set repo URL in package
0.5.0
Changed name to pyproject-local-kernel (previously ryeish-kernel)
Now supports Rye, PDM, Poetry, Hatch and custom configuration