Python Scripting Extension - Release Notes
Release date: August 11, 2021
- Execute Python operator now uses the directory of the specified script file as a working directory, allowing the script to use other scripts, modules, models or other files in that directory
- Execute Python with script parameter, Python Transformer and Python Learner operators in saved processes now use the location of the process as a working directory
- Connection objects can now be read in Execute Python if connected to one of its input ports
- Show Preview... action and tag filter suggestions in Execute Python now work with the script file input port as well
- When using connections on the input ports of Execute Python in Python Scripting Extension 9.10, the
cryptographyPython package has to be installed in your Python environment
Release date: May 31, 2021
- New operator Remote Python Context allows to use a Remote Python Engine for running Python code
Release date: March 23, 2021
- New operator Python Learner to create custom Python based operator compatible with RapidMiner's model interfaces
- New operator Python Transformer to create custom Python based operators for generic transformations
- Fixed an issue where the rm_main structure was not detected correctly in notebook files
- Python Scripting Extension 9.9 requires RapidMiner Studio and RapidMiner AI Hub version 9.9 or later.
- Python Scripting Extension 9.9 requires Pandas 1.0 or later.
Release date: Oct 13, 2020
- Stopping a process with Execute Python now kills the already started Python process as well
- Newly installed environments now become immediately available in Job Agent or Web Service execution without restart
Release date: Jun 24, 2020
- Empty data set on the Execute Python input no longer throws EmptyDataError
- When a .py or .ipynb file is dropped onto the canvas from outside of Studio, the path parameter is now correctly set
- More descriptive error message is thrown if a date value outside of the allowed range of datetime is used in Execute Python instead of "Python int too large to convert to C long"
- Fixed potential false warning regarding missing rm_main method when a script file is used in
Release date: Jun 3, 2020
When upgrading RapidMiner Server to version 9.7 or later, you must also update the Python Scripting Extension to version 9.7.
- As the new repositories support storing arbitrary files, support for dragging and dropping .py and .ipynb files to the canvas has been introduced. RapidMiner Studio will take care of creating the necessary operators.
- Note: no adjustments are made to the code when dragging and dropping, so users still need to take care of creating the rm_main function so the code will get executed properly. A run-time error will be displayed if this is not done.
- Added support for projects, so the rapidminer Python library can function correctly.
- Optional hidden parameter (
rapidminer.python_scripting.shell_command) added to avoid potential freeze during extension initialization on Linux and macOS(default value:
bash -l -i -c)
Release date: Feb 25, 2020
- Macros are now accessible in Execute Python in an optional dict parameter. Their value can be modified in the script.
- Pandas 1.0+ is now supported
- Array-like values in a DataFrame cell no longer cause error when the DataFrame is returned on the output
- Irrelevant potential DtypeWarning while loading an input is suppressed now
Release date: Dec 11, 2019
- None at this time.
- Improved robustness of Python execution on Mac and Linux systems
- Improved data handover between RapidMiner and Python (expected to be faster, but slightly increased memory footprint)
- Fixed NoSuchMethodError when Execute Python is used in a web service on Server
- Macros are now also substituted if you are using a .py or .ipynb file in Execute Python
- False "Cannot retrieve file name" warning message is no longer logged
- (Bugfixes listed below apply if you use an Execute Python operator with Compatibility level above 9.3.1)
- ExampleSets containing multiline text are now handled correctly
- Special characters in attribute names, roles and nominal values are now handled correctly
- Time zone in date types is now handled correctly
- Boolean nominal values are no longer converted to "True" and "False"
- Explicit errors are now thrown when rm_metadata settings are in conflict with the values in the dataframe
- Empty nominal value remains an empty string and no longer becomes missing value