Jupytext Versions Save

Changed

• We have updated the GitHub workflows - thanks to Matthew Feickert and to Cristian Le for their help on this subject (#1037)
• We have removed the upper bound on markdown-it-py<3. Now we test Jupytext with markdown-it-py in versions 2.x and 3.x on the CI (#1075)

Fixed

• Notebooks with an empty YAML header work (#1070)
• Double quote strings in R Markdown options can contain single quotes (#1079)
• The necessary directories are created when paired notebooks are moved to a sub-folder (#1059)
• Commented quotes are recognized as such (#1060)
• Jupytext can use either pkg_resources or packaging to parse version numbers (#1085)

Changed

• This version comes with a build requirement jupyterlab>=3,<4, as the Jupyterlab extension for Jupytext is not compatible with JupyterLab 4 yet (#1054)
• The JupyterLab extension was released to npm in version 1.3.9.

Added

• Added Stata as a supported language (#1027) - thanks to Raffaele Mancuso for this contribution
• Added SAS as a supported language (#1047) - thanks to lawrencet149 for this contribution
• We have added a series of test to make sure that the main formats support cell tags (#1024)

Fixed

• When a metadata key is not a valid identifier, a warning is emitted and the metadata is not saved to the text representation (#1042)
• The CI was fixed by Matthew Feickert (#1035)
• We now use concurrency to cancel previous runs on the same branch/PR (#1037)
• We use both codecov.notify.after_n_builds and comment.after_n_builds to get only the final codecov comment

Changed

• Empty tags are not exported to the text notebook anymore (#960)
• We updated the yarn.lock file for the jupyter lab extension to address security vulnerabilities (#1030, #1036)
• In the pre-commit tests we now use main for the main branch

Added

• Added Wolfram Language as a supported language (#1014) - thanks to Etienne Dechamps for this contribution

Fixed

• When the default contents manager is async (i.e. jupyter_server>=2.0.0), the Jupyter server extension for Jupytext derives a contents manager from LargeFileManager instead, as async contents managers are not supported by Jupytext at the moment (#1020)
• We have made adjustments on the CI as flake8 was moved to GitHub, and Python 3.6 is not available anymore on ubuntu-latest

Fixed

• The sample notebooks have been normalized with nbformat.validator.normalize (#1002).
• The warnings in the test suite that we cannot fix are filtered using a new pytest.ini file
• We updated the yarn.lock file for the jupyter lab extension to address security vulnerabilities (#984, #1005, #1011)

Changed

• The CI uses Python 3.9 rather than 3.7 when testing conda environments

Added

• Gnuplot is now supported (#998) - thanks to razimantv for this contribution
• We now test Jupytext against Python 3.6 to 3.11 on the CI
• We have added a test to document how to use the folder and prefix matching when pairing notebooks (#974)

Fixed

• The timestamp of a paired notebook is the timestamp of the most recent paired file. This fixes the warning "File Changed" after reloading the notebook in Jupyter (#978).

Changed

• The Jupytext configuration file has a new option cm_config_log_level that defaults to info_if_changed. With that value, the contents manager will log a line regarding the configuration file used only when the config file is not the same as the one previously used (#959) - many thanks to R.C. Thomas for suggesting this and thoughtfully testing the patch.
• Hidden configuration files like .jupytext.toml or .jupytext.py are now ignored by Jupytext's contents manager when allow_hidden=False (that option was introduced in jupyter_server==2.0.0a1) (#964).
• We have changed jupytext --set-formats to make it more similar to jupytext --sync. Now --set-formats will not override existing paired files anymore (#969).

Added

• We have added a test test_pre_commit_hook_sync_with_no_config that documents how to use the pre-commit hook without a configuration file (#967)

Fixed

• Text-only notebooks are always trusted (as they don't include any output cells) (#941)
• We made sure that our tests also work in absence of a Python kernel (#906)
• The coverage of the tests folder has been restored at 100%
• Bash commands like !{cmd} are now correctly escaped in the py:percent format (#938)

Added

• Added Tcl as a supported language (#930) - thanks to shishitao for this contribution
• Added Maxima as a supported language (#927) - thanks to Alberto Lusiani for contributing a sample Maxima notebook.

Changed

• The Jupytext contents manager is derived from the LargeFileManager imported from jupyter_server rathen than notebook (#933)
• Allow for markdown-it-py v2 (#924)
• We have updated the hooks used in the test pre-commits, to fix an issue on the CI (#940, #942)
• We updated the yarn.lock file for the jupyter lab extension (#904, #925, #935, #939)

Fixed

• The Jupytext CLI only suggest --update when the target is an .ipynb file (#905) - thanks to st-- for this contribution
• We made sure that commands like cat notebook.md | jupytext --execute work (#908)

Added

• Added Haskell as supported language (#909) - thanks to codeweber for this contribution

Changed

• We have updated the pre-commit hooks and in particular we switched to the first stable version of black==22.1.0.
• We require pandoc==2.16.2 for testing. The representation for code cells changed from ``` {.python} to ``` python in that version of Pandoc (#906). We don't use pandoc>=2.17 in tests at the moment because of the introduction of cell ids that cannot be filtered.
• Jupytext will not add anymore a UTF-8 encoding on Python scripts when the notebook contains non-ascii characters (#907)
• We have added pyupgrade to the pre-commit hooks used for developing Jupytext (#907)