Sympathy for Data version 1.1 is a significant minor release for Sympathy for Data. Since version 1.0 was released almost one year ago, the application has been rewritten entirely in Python. It is however not redesigned and with a few exceptions, all existing nodes and flows should work out of the box.
This maintenance release 1.1.3 improves on 1.1.2 with a minor changes to plugins, nodes and improved stability.
The bundled python installation has been upgraded with new versions of almost every package. Added to the packages is scikit-learn, used for machine learning. Our investigations suggest that the new package versions are reasonably compatible with old nodes and cause no significant differences for the standard library.
Download Sympathy for Data built for Windows: https://bitbucket.org/sysess/sympathy-for-data-public/downloads
Changes Compared To v1.1.2
- Importer plugins:
- Added CarMaker type 2 ERG ADAF importer plugin called “CM-ERG”
- Exporter plugins:
- Plugins can now export to non-ascii filenames
- Fixed MDF export of boolean signals
- Convert column nodes can convert to datetime
- Calculator node can produce compact output for length matched output
- Lookup nodes handles both event column and other columns with datetimes
- Time Sync nodes “SynchronizeLSF” strategy should work as expected again. The Vjoin index option is now only used for the “Sync parts” strategy
- API changes:
- Added has_column method in sytable and added corresponding method in table.File
- Accessing an ADAF basis which does not exist will raise a KeyError
- Warn about corner cases where the upcoming deprecation of has_parameter_view will make nodes behave differently
- Avoid hanging on Windows when too much output is produced during startup
- Avoid infinite wait during node validation
- New command line options:
- Made it possible to set the number of workers using –num_worker_processes n