Loginom 6.5.1 release notes

Quick view interface was improved. The memory leak error was corrected in Python. Version 3.10 is now supported. Errors in “Coarse classes”, “Logical regression” and other handlers were corrected.

Fixes

Python: The memory leak error while Python using was corrected. It became possible to sidestep the problem that is typical of the Python interpreter implementation when it was connected in the dynamic library form. Memory consumption was optimized (#9054).

Firebird: Full roll back of transactions to the nonempty table in "Clear table and append data" mode was not possible when exporting to the database. The table stayed cleared after the error, and export was not executed. Transactions roll back was corrected (#8103).

Auto save packages: Previously, the conflict that sometimes made impossible to provide reliable saving while simultaneous package auto saving and closing via "Session Manager" could occur. The introduced fixes enable to prevent such conflict providing reliable auto saving of packages (#9025, #9021, #9038).

Auto save packages: The users that did not have rights to edit packages were allowed to configure packages auto saving. Now, "Auto save" button is not displayed for such users (#9043).

Administration: During the first "Loginom Desktop" start and transfer to the application parameters settings it was incorrectly displayed that the locale had been chaged. This error was corrected. The active application locale setup was added when creating the default application settings (#9073).

"Chart" visualizer: The error connected with parameters saving was corrected. It could sometimes occur when "Apply" button was quickly pressed right after parameters change in any modal chart window.

"Chart" visualizer: In Loginom 6.5.0 the chart export in "svg" format was not operable. Corrected (#9123).

"Chart" visualizer: Field references editors were always highlighted in red as the incorrect ones if the reference was specified. Corrected (#9053).

"Chart" visualizer: Horizontal chart display became possible when pressing on "Horizontal orientation" button (#9001, #9097).

"Data Quality" visualizer: the visualizer error was corrected. It occurred when the field with the logical type was added to the existing and repeated calculation (#9095, #9085).

"Data Quality" visualizer: If all indicators disabled by default were selected in the indicators (Minimum, Average, Unique, Maximum, Median), "Calculate statistics" was pressed and transfer to "Discrete" tab was performed, the cells with the added indicators were absent in the displayed table, whereas headers for them were available. Corrected (#9082).

"Data Quality" visualizer: When pressing on the splitter button of the detailing bar that was located in the lower part of the window in the collapsed state, the bar was not loaded. Corrected (#9078).

"Cube" visualizer: Copy and paste operations were available only in the multiple "Cube" mode. In this case, the "included/excluded values" mode had no effect on the operation. Copy and paste operations were improved in the different "Cube" dimension filter modes as follows:

  • multiple; modes of included and excluded values — all included/excluded values are copied, successfully captured values are added to the included/excluded values while pasting;
  • single and combined modes — the current selected value is copied, only the first row value is processed while pasting, the other ones are ignored, the successfully captured value becomes the selected one, the table is scrolled to reach it (#9107).

The following errors were corrected:

  • precision loss while copying and pasting dates and figures in the "Cube" dimension filter(#9106);
  • the "Cube" dimension filter bar was collapsed when an error occurred during copying or pasting (#9108);
  • when the focus was in the search field in the "Cube" dimension filter bar, the filter state was copied instead of the text field content (#9105).

"Cube" visualizer: the minimum sizes are set for all modal windows (#9098).

"Cube" visualizer: When pressing on the splitter button of the detailing bar that was located in the lower part of the window in the collapsed state, the bar was not loaded. Corrected (#8963).

Node Execution: The error that occurred when "Node Execution" referred to the derived node for which the source component was not found was corrected. Previously, it was informed about an error, the package loading was terminated (such package was not opened). Upon correction it will be possible to edit the reference of the package that could not be previously opened and restore its operation (#9096).

Replace: possibility to set "Usage type" is disabled at the input port "Replacement table". Previously, it was possible to set several "Usage types" equal to "Input" and "Output" that resulted in an error in the "Replace" node operation (#8574).

Import from text file: The last column was not added when defining the column list if a separator was the last character in the first row (#9071).

Import from text file: The "Undefined" data type was excluded from the field wizard when importing from the text and "Excel" file. In some cases there were problems during the further data processing (#5903).

Coarse сlasses: The error that occurred while processing the volatile filed, for example, the one received after "Calculator" when using "Random" function. The problem caused the incorrect component operation if new values (that were absent while statistics calculation) appeared in the discrete data while training. Processing was not started (#8885).

Logistic regression: Memory access error occurred for the new data sets when using the automatic setting and specifying the maximum precision priority. The operation of partitioning to the training and test sets of handllers and cross validation in regressions was corrected. It excludes this error (#9080).

Errors display: The long file names were cut with an error in the pop-up window if there was no enough space for them across the width. Break of the message lines was enabled in the modal windows. It excludes the incorrect display (#7081).

Reports: Sometimes an error occurred in the navigation tree when opening the file storage folder or file storage in the next tab. The error was corrected (#9035).

Variables: Application of the package variables values was corrected for the group editing of the variables related to different types. Pressing on "Apply" button was ineffective, variables values were not changed (#8987).

Derived components: The situation when there was a reference to the package with derived components in the current package but the nodes matching these derived components had not been created in the program yet was taken into account. In the result of the error in "Base component selection" window it was sometimes displayed that the list was empty. Corrections exclude this error (#9042, #9044).

Derived components: The following error was corrected: "Transferred derived component is not available for the node" when collapsing to the Supernode (#8931).

Derived components: The error that previously occurred while cloning the Supernode that contained the base and derived Supernode with the configured visualizers was corrected (#9116).

Derived components: The error occurred while loading the package that contained the Supernode with the node created on the basis of the faulty component was corrected. Previously, when trying to open such package, it was not opened, and an error was shown. Now, such package will be opened (#9099).

References to packages: The reference was damaged when the package was resaved. That's why it was not possible to find the nodes used from the connected file. The logics of restoring reference to the package was corrected. The autocorrect of references to packages was deleted if it was not possible to find the package while reference loading but the package with corresponding file name was found in the current folder (near the package that included the reference). It is still not possible to find the package in the current folder, and if it is found, the corresponding information is added to the reference restore error message. It also became possible to change the path to the package in "Properties of reference to package" window if the reference had been broken. When the broken reference to package was restored, restoration of the "destroyed" derived nodes, "Execution", "Loop" and Reference nodes" nodes was called in the current package (#8981).

Table to tree: When the handler wizard was used for the configuration purposes but the field marked as "Usage type" in the port was not used, "Used" field value was reset to "Unspecified". To solve this problem, possibility to set "Usage type" in the input port of "Table to tree" handler was removed (#8571).

Table to variables: When the "Table to variables" handler wizard was used for the configuration purposes but the field marked as "Usage type" in the port was not used, "Used" field value was reset to "Unspecified". To solve this problem, possibility to set "Usage type" in the input port of the handler was removed (#8570).

Table to variables: When quick view of "Table to variables" node was opened, the first cell was not selected that could sometimes cause an error. Corrected (#8904).

Filter: The displayed tip text was corrected when the filtering criteria were set for the filter by the time span. Previously, the following text was displayed: "last period to...". Now the following text is displayed: "last perfiod before...". Thus, peculiarities of the Russian language were taken into account (#8848).

Loop The base package inside the parallel Loop was incorrectly defined that caused incorrect calculations. Correction of this error provides the same field order at the node output for all Loop iterations that excludes errors while data processing (#9103).

Improvements

Python: Python 3.10 support was added. More detailed information on all changes of the new version is available on the official site. Python 3.10 support enables to use the recent versions of the popular libraries in the most efficient way (#9047).

Import from text file: Algorithm of the automatic definition of the table structure in the text file was improved. Now, not comma is used as the default column delimiter for csv files. "List item delimiter" parameter is taken from locale (it is ";" by default). To provide the automatic definition of parameters, all rows selected by a user for analysis are viewed (25 rows are set by default). When the column delimiter is defined, it is also important to take into account that a comma between 2 digits can be a delimiter of the integer and decimal parts. The date format and separator definition algorithm became "stricter". Date and time specification in the row is taken into account. The partial format is also taken into account. Namely, only date or only time must be displayed in the row (#9110, #3836).

Import from text file: "By default" value of "Text qualifier" parameter was renamed to "No" to simplify the perception. Previously, users often confused "By default" value for "Text qualifier" with the same value for other parameters (#5351).

The problem was that "By default" for "Text qaulifier" was not logically used in that context. It was confused with "By default" for other parameters.

Connection to Databases: ODBC connection was restored after break down while preparing and executing request during import from database, opening transaction during export (#8619).

Derived components: Apart from the node name, GUID with the package name is also displayed in the cases when it is not possible to find it in the source or opened packages. In the result, it becomes easier to search for the problem nodes. (#8952).

Derived components: The message shown while attempting to delete the derived component became more informative. The number of nodes in the opened packages that refer to this component (Execution and Loop nodes, the derived nodes created by this component and the components derived from it are taken into account) is currently specified in the message. If no dependent nodes are found but the derived component has "Public" visibility, this information is also provided (#9004).

Linear regression: The message is currently shown when training the "Linear regression" when the count of inputs exceeds the count of rows. The error will be shown for the methods that initially require availability of all coefficients (Enter, Backward, Ridge, LASSO и ElasticNet) up to the moment of model training if the row count is less than the count of coefficients (including the intercept). It will prevent occurrence of more serious errors at the next stages. Maximum possible count of coefficients will be included for the step-by-step methods with forward pass (Forward or Stepwise). Then the algorithm will stop or return according to its type. When the maximum number of coefficients is reached, it will be recorded in the log message (#7968).

Filter: When the filter is set by the time span, it is forbidden to set the period number value less than 1. Improvement excludes the logical error (#9052).

Quick View

Quick view interface was improved (#9065).

"View" button was removed from the workflow toolbar. 3 display options that can be switched are available:

  • modal window (default value);
  • modal window is maximized;
  • bar is fixed on the bottom.

The modal window is closed on click beyond its limits.

Possibility of bar locking in the right area is excluded.

#release notes

See also

Loginom 6.5.6 release note
The errors in Tree to table, Replace, Coarse сlasses, Binning, Loop handlers and in Quick view and Coarse сlasses visualizers were corrected.
Loginom 6.5.5 release note
Errors in Import from Excel file and Export to Database wizard were corrected. Chart Visualizer was corrected. The Loop Handler error was corrected.
Loginom 6.5.4 release notes
The Python operation with numexpr module was corrected in the code execution mode inside the process, errors occurring while executing requests to databases were corrected, regression bugs in...