Fixed Issues in Rulex 4.7
| Issue | Solution |
|---|---|
Missing values in temporal attributes were not imputed correctly in the Fill and Clean task when a fixed value was chosen. | Now the task works as intended for all temporal attributes. |
| Python and R Bridges forwarded double to external code with only float precision digits. This could result in rounding issues in conversion. | Now double precision is maintained in external Python and R code. |
| The output dataset of a Python or R Bridge task could contain duplicated values on nominal attributes, leading to unexpected results in downstream tasks. | Now nominal values are simplified, and duplicate values are removed when exiting from a Python or R Bridge task. |
| Binary columns in EBCDIC format were not managed correctly in Import from DB tasks. | Now binary columns in EBCDIC format are correctly imported, and converted to UTF-8 code. |
| The textExtract, head and tail function in the Data Manager did not consider the dollar ($) sign in arguments after the first occurrence, even if permitted by the function. | The textExtract, head and tail function in the Data Manager now correctly accepts attributes in column, start and end parameters, thus permitting a different truncation, row by row. |
| Graphics management in versioning led to a Right Data Truncated warning on SQL Server. Such warning was annoying since it was repeated at every versioning operation. | Column maximum selection length for strings has been increased to deal with more graphics than before, without reaching SQL Server ODBC driver limit. Moreover, warnings are now returned only during Add/Clone/Merge versioning operations. |
| The error 08S01, Network TCP connection forcibly closed, was ignored when the ODBC driver sent it as a warning instead of an error. | The error 08S01 is now automatically converted into an error whenever it is returned, even if it is sent as a warning from an external source. |
| The cumSum and cumProd functions in Data Manager were not working correctly for currency and percentage types. | Now these functions work as intended on currency and percentage types. |
| SQL Server Database did not permit cursor fetching other than SQL_FETCH_NEXT. Consequently, in the case of a Right Data Truncated warning, recovery was not allowed. | Automatic String Length retrieval has been disabled on the SQL Server Database, to prevent incorrect fetching. Internal SQL Server string length limit has been increased from 2^16 bytes to 2^20 bytes to prevent Right Data Truncated warnings on internal Rulex Graphics on versioning. |
| Blob upload on Azure Blob Storage did not work with SAS key authorization. | Blob upload with SAS key has been fixed, and it now works as intended. |
| Data in warning status were not exported in an exported PRCX. This prevented the use of the old data computation option after Import/Export PRCX procedures. | Now data are exported if the task is in COMPUTED, DIRTY or WARNING status. |
| Vault Variables could not be renamed, due to an unhandled exception. Moreover, they could not deleted if another process variable with the same name was defined. | The renaming issue has been fixed, and now works as intended. The function, which checked if Vault Variables were in use, has been erased, consequently enabling their deletion, and allowing the use of another process variable with the same name. |
| Long files in UTF16 code (e.g. tsv files from OMP) were truncated, possibly leading to incorrect values for NOMINAL attributes. | Long files in UTF-16 code are now correctly imported. |
| The presence on incorrect (huge) values in datetime cells of an Excel file could produce an unknown internal error when that file was imported in Rulex. | Now, incorrect (huge) values in datetime cells are correctly managed. |
| In rare situations, the algorithm used for implementing the fair share option in the Network Optimizer task lead to a rounding issue, which reduced the quantity to be sent by 1 element. | The rounding issue has now been solved, and the correct quantity is always produced. |
| The formulation of the linear programming problem considered in the Network Optimizer task did not correctly handle some situations, thus leading to unexpected results. | The formulation now correctly manages all situations, ensuring a correct optimal solution. |