Fixed Issues in Rulex 4.0.8
One of the functions in the OrderedDict third-party library used by Rulex (dict.items()) did not support parallelism and consequently randomly created deadlocks.
The source code has been changed in order to avoid using this function, and subsequently prevent deadlocks from occurring.
The Decision Tree and K-Nearest Neighbor classification tasks caused Rulex to crash due to errors in their internal algorithms.
These errors have been corrected and the tasks now work correctly.
Certain queries performed in Linux caused Rulex to crash due to a memory error.
The memory error has now been resolved.
There was an excessive use of memory (over 16GB) when joining two data tables with over 1 billion rows.
This error has been resolved and memory usage is within accepted parameters.
The catNames function caused Rulex to crash if there was only one column to concatenate.
The single column concatenation situation is now correctly handled.
Errors were not correctly managed in the case of memory overflow.
Memory overflow errors are now correctly managed.
If the characteristics of an attribute, which was not present in the dataset, were modified via a copy and paste operation in the History tab, problems occurred in the Rulex interface.
Modifications made on attributes not present in the dataset are now ignored.