Page Comparison

You can now add a ‘languages.properties’ file into your locale directory to dictate what languages appear in the Language Selectors and under what name they are displayed.

The presence of this file disables the auto-generation of languages from the ‘messages*.properties’ filenames and will not provide 'English' as a default. Languages in this file are defined by simple key/value pairs of the ‘language code’ and the display name separated by an equals sign e.g. to define the language associated with files with the ‘en_GB’ language code in their filename and have it appear as “British” in the Language Selectors I would have the following on its own line within the file: en_GB=British.

Previously, translation files uploaded through the UI had to include both a language and locale as part of their ‘language code’ e.g. ‘messagesStatic_en_GB.properties' was accepted but the language only version of ‘messagesStatic_en.properties’ would have failed validation.

If not using the 'languages.properties’ file detailed in the release note above, languages defined without the locale will appear in the Language Selectors without the parenthetical locale i.e. “English” rather than “English (United Kingdom)”

The following magic variables relating to Date have been added to the system:

START_OF_NEXT_WEEK
END_OF_NEXT_WEEK
START_OF_NEXT_MONTH
END_OF_NEXT_MONTH
START_OF_LAST_YEAR
END_OF_LAST_YEAR
START_OF_NEXT_YEAR
END_OF_NEXT_YEAR
START_OF_QUARTER
END_OF_QUARTER
START_OF_LAST_QUARTER
END_OF_LAST_QUARTER
START_OF_NEXT_QUARTER
END_OF_NEXT_QUARTER
CURRENT_DAY_SHORT_NUMBER
CURRENT_DAY_WITH_SUFFIX
CURRENT_MONTH_SHORT_NUMBER
CURRENT_WEEKDAY_SHORT_NAME
LAST_MONTH_SHORT_NUMBER
NEXT_YEAR
NEXT_MONTH_YEAR
NEXT_MONTH
NEXT_MONTH_NAME
NEXT_MONTH_SHORT_NAME
NEXT_MONTH_SHORT_NUMBER

The full description of these environment variables can be found here: /wiki/spaces/DEV/pages/2151153665

A footer detailing the ‘generation date’ and ‘pages’ has been added to Browser PDF exports, to maintain consistency with the previous Renderer export.

The formatting of the date can be adjusted with the Global Variable PDF_DATE_FORMAT, whilst the following CSS styles can be used to customise the three footer related classes defined in the Themes > PDF tab (.pi-style__pdf-footer, .pi-style__pdf-footer-date, .pi-style__pdf-footer-pages):

• background-color: The alpha channel of RGBA values will be ignored

• color

• text-align

• font-size: Only supports ‘px’ values or numeric values with no unit specified

• font-weight: Only “bold” has any effect

• font-style: Only “italic” has any effect

• margin: Only supports ‘px’ values or numeric values with no unit specified. Both specific margins (e.g. margin-right) and ‘margin’ value/array are supported

• display: Only supports ‘none’ to remove specific parts of the footer (.pi-style__pdf-footer-date, .pi-style__pdf-footer-pages) or the entirety of it if applied to `.pi-style__pdf-footer`

Below is an example of the generated footer with some applied styling as mentioned above:

Embed actions (used to handle post message front-end actions when embedding) now supports a wildcard '*' in the trusted hosts setting, and will allow a post message from any subdomain if the wildcard is provided.

Previously, you would be unable to edit a chart in full screen view. Now, you can freely edit the chart, and it will return you to full screen view once you have made your changes.

All users will now be moved across to the new version of the Excel export functionality, this version is significantly improved over the older version.

Previously, you would have to close the snackbar that appears when exporting a table to Excel, when one or more columns was marked as being excluded from export. It now disappears automatically after a few seconds.

When a Rotatable Table included three columns and three rows of data, when attempting to export this to excel, this would lead to the export failing.

Fixed an issue that prevented certain chart types (gauge, flow, treemap) from being successfully exported to PDF format. The issue has been resolved, ensuring all chart types are now exported accurately, matching the visualisations in the UI.

When entering full-screen mode, the custom map will now be completely redrawn to adjust its dimensions, ensuring that tooltips are positioned correctly near the data labels.

Previously, data labels and tooltips on stacked bar charts would omit negative values from the total value for each bar, which would also affect the percentages displayed for each bar segment. Now, the positive values minus the negative values is used for the total, and percentages for each segment are taken as a percentage of this new total.

A handful of additional fixes have been made off the back of a series of smaller bugs revealed due to this implementation;

1. Ensure that ‘NaN%’ does not show when you click on Legend items on Bar Charts and/or Stacked Bar Charts

2. The percentage labels on the bars within 100% Stacked Bar Charts, are treated similarly to individual pie charts i.e. the percentage of a bar always adds up to 100% - regardless of if the chart consists of 1 dimension and 1 measure, multiple dimensions, or multiple measure,

3. Negative values are suppressed from data labels and tooltips on 100% Stacked Bar Charts - including regular and percentage data labels

4. Any of the following charts that have negative values within the measures display a ‘suppressed negatives’ warning indicator on the top left of the chart:

   1. Bar, Stacked Bar, 100% Stacked Bar, Pie, Doughnut, Variable Pie, Variable Doughnut, Sunburst, Doughnut Sunburst, Treemap, Funnel & Pyramid

5. If only negative values appear on a bar within a regular Bar Chart or Stacked Bar Chart, the percentage values are positive (i.e. the grand total is -30, and a segment is -15, this is 50% - NOT -50%)

Previously, certain chart sizes would result in multi-line x-axis labels that no longer properly matched the position of their data point. They would appear left-aligned with respect to the axis space reserved for that data point rather than central to it. This positional discrepancy looked particularly bad after zooming. The labels are now properly centrally aligned.

Before:

After:

To ensure the app launches successfully across all Windows environments, we modified the build configuration to output the configuration-tool-GUI as an unpackaged directory containing all its files.

A new directory named configuration-tool has been created to house both the configuration-tool-GUI and configuration-tool command-line executables, along with all relevant files. The GUI will now launch directly from the configuration-tool directory.

Previously, changes to this configuration setting through the GUI would not be applied - this has now been resolved.

The scheduler service will now utilise the new browser export functionality to generate charts that support it. This relates to PDF, Word and PowerPoint single chart exports.

7-Zip is used to encrypt Schedules that you have added a password to but was missing from the Scheduler Docker image leading to the inability to use this feature. This has now been resolved and any previous workarounds you have applied to volume these files yourself can be deleted.

It was possible for a import that updates an existing Schedule that has Roles associated with it to fail with a message related to a unique constraint - this has been resolved

Previously, when you would try to zoom in on a merged chart with Y axis zooming enabled, then reset the zoom state, the tick intervals on the Y axis would be different from what they were when the chart was in its initial state. Now, the original tick intervals are retained once you have reset the zoom state.

Previously, in certain scenarios, ‘Reset Zoom’ button was overlapping chart content when zooming in. This has been resolved by adding some additional space to the top of the chart container when zooming in, allowing the zoom button to appear outside the chart body. When zooming out, the space reverts to its original state.

Previously, when setting the decimal places of a measure within the chart attributes, on a table, which contains 0 values, the decimal places would not apply and the 0 values would fall back to using the decimal places defined in the object definition. Now, the decimal places defined in the chart attributes will always take precedence, and apply to 0 values successfully.

Previously, measures in Crosstab tables (tables with a Header defined) would have their Data Object/Attribute formatting applied (e.g. a prefix) to not only their data values but also their respective headers. Now, this formatting only applies to the Header defined in the Chart Configuration as those headers are generated from the data values of that object.

Previously, the healthcheck script that runs on Docker would ping both ‘/version’ and ‘/pi’ to ascertain the health of the container. The latter ping was causing a repeated WARN log message to be produced and has now been removed. The ‘/version’ ping has been changed to include a --max-wait of 5s and hit the ‘/health’ endpoint instead.

When performing migration processing, should SQL failures be encountered, these are now added to the end to aid triage of the issue.

Previously, when starting up the Dashboard the catalina{date}.log file would show some warnings related to attempting to set ‘maximumSpareThreads’ and ‘proxyName’. These log messages should no longer appear.

Logging around the editable tables functionality has been improved, you will now always see the full outbound request detail for any failures and always if using debug level logging.

Previously, maps with null values would display errors in the console when zooming in and out. Additionally, maps with legends with null values would not render at all.

Previously, subtotal labels were not being translated.

When selecting a new language in the GUI, not all translations were being performed, specifically around where [[VARIABLES]] were being used in queries performed. When the language is now changed through the GUI, the category/page will be refreshed to ensure all aspects of the charts are translated using the newly selected language too.

During the process of creating a new Category, User, Role, Data Connection, Theme, API, Organisation, Report or Schedule, the new item would be described as 'No Title' (see the image below) until it was saved - this now respects the language in use without the need to save first.

When both the Category List Panel and Category Object Panel were pinned, it was possible for the Category Object Selection popup to be partially obscured by the Category List Panel

Resolved an issue where the drop down data in Category Object filters would become out of sync if a user removes a preceding temporary category object filter value

Resolved an issue where the start of the financial year was incorrectly calculated when the currentMonth was earlier than the START_OF_FINANCIAL_YEAR_MONTH. The logic has been updated to ensure that in such cases, the start of the financial year is correctly determined as belonging to the previous calendar year. This change applies to both the financial environment variables and the values in the ‘Financial’ tab in the category objects date range panel.

Due to changes in November related to how we initialise JDBC drivers in a more isolated way, the wrong tomcat base directory was being used for jdbc_drivers when using systemd. Similar issues were discovered due to this bug related to the dynamic drivers functionality and the location of the ‘out.{date}.log’ file (rotating Dashboard log file). These issues have now been resolved.

The API request for exporting a data connection now includes the associated Categories & Charts.

Info still needs to be added

Info still needs to be added

Updated Netty framework from version 4.1.100 to 4.1.115

Minor Tomcat upgraded from 9.0.97 to 9.0.98 and the useCanonCaches Tomcat property explicitly set to false - resolves CVE-2024-56337 & CVE-2024-50379.

Updated the Lettuce library from version 6.1.10 to 6.5.1.

Updated the Crypto and Net packages to the most recent versions. This addressed vulnerabilities around the Scheduler and Excel Reader modules.

Updated the Snowflake JDBC driver from version 3.14.1 to 3.20.0 (latest).

Updated the 7 zip library from version 23.01 to 24.09. This library is used for encryption within the Scheduler.

Index Added

mis_audits.audit_date_time and mis_audits.user_code

migrations#62

31663

Foreign Key Added

between pi_schedule_user_links.schedule_id and pi_schedules.id

migrations#62

31663

Unique Constraint Added

on pi_schedule_user_link between user_id and schedule_id

migrations#62

31663