April 2024 - Dashboard Release Notes

A new option has been added for the HTTP Method, to be used for sending requests from an Editable Table. This new method, named “POST (JSON)” as opposed to the pre-existing method that has been renamed “POST (Classic)” will send all modified rows in a single request formatted in the following JSON structure:

{
  dashboardUser: usercode,
  dashboardToken: token,
  modifiedRows: [
    {colValue: firstRowValue1, colValue2: firstRowValue2},
    {colValue: secondRowValue1, colValue2: secondRowValue2}
  ]
}

To accommodate this new option, the radio buttons for this setting have been changed to a drop down as shown below:

A new checkbox has been added to control whether the labels are shown for dimension subtotal breaks. By default, the checkbox will be unticked, hiding the labels next to subtotal break values.

The checkbox is only be available to Table chart type.

Once the checkbox is ticked, the labels will appear to all subtotal values, e.g.

Added a new Magic Variable [[LAST_MONTH_YEAR]] that resolves to the year (formatted yyyy) the previous month was in i.e. if the current month is January 2024 it will yield 2023.

Adding the ability to skip migrations by providing the following flag, under the ‘advanced’ section under ‘repository’ in the configuration tool:

It can also be applied by setting the following environment variable under the dashboard service in docker: PI_DB_SKIP_MIGRATIONS=”true”

The following new entries have been added to the messagesStatic.properties file, to allow the translations of Category Objects types in the dropdown list.

List of new key=value pairs:

category.object.type.droplist.label=Drop List
category.object.type.search.label=Multi-picker
category.object.type.checkbox.label=Checkboxes
category.object.type.images.label=Images
category.object.type.slider.label=Slider
category.object.type.num.picker.label=Number Picker
category.object.type.single.date.picker.label=Single Date Picker
category.object.type.date.picker.label=Date Range
category.object.type.free.text.label=Free Text

If you wish to use these new fields, you will need to update the existing language files (e.g. messagesStatic_en_GB.properties, messagesStatic_pt_PT.properties) with the above keys and add your translated values.

Example of the result in the UI:

New associations and entries have been added between the Dashboard UI and the messagesStatic.properties file, to allow the translations of both the text and the hover text of elements on the pop up window - which appears when you attempt to close a tab with unsaved changes. Additionally, the yellow ‘unsaved changes’ indicator’s hover text is now translatable.

List of new key=value pairs:

default.button.edited.indicator.title=Modified and not saved
default.button.save.title=Save changes
default.button.cancel.title=Cancel to close this dialog
default.button.discard.title=Discard changes and close the tab
default.status.bar.message=You have made some changes, do you want to:

Pre-existing Key=Value pairs used:

default.button.discard.label=Discard
default.button.cancel.label=Cancel
default.button.save.label=Save

If you wish to use these new fields, you will need to update the existing language files (e.g. messagesStatic_en_GB.properties, messagesStatic_pt_PT.properties) with the above keys and add your translated values.

Example of the result in the UI:

The JDBC URL field of Free Format Data Connections has been updated to allow the use of secure variables - it already supports insecure ones.

This also means that other data connections that have multiple fields that ultimately combine to form the JDBC URL, for example the ‘Host’ field, now support secure variables as well. Additionally, the Driver Class Path field that previously did not allow any variable replacement has been updated to support both secure and non-secure variable replacement.

The Username and Password fields still only allow for secure variable replacement however they now have a lock icon with a tooltip to indicate that this is the case

Previously, after having written in one of the text boxes within the data objects screen, followed by selecting another text box to write to, you would have to click twice to focus on the 2nd text box. Now, you can click once and it will focus immediately as normal.

Previously, upon duplicating a category, it would not copy the category layout. This means all of the charts would need to be added manually. Now, it will attempt to copy the layout, following this hierarchy:

1. If a default layout exists, it will copy that

2. If a user layout exists for the user attempting to duplicate the category, and no default layout exists, it will copy that

3. If no layout exists, it will copy nothing

User’s forename and surname will be used by default in the dropdown in ‘More Options’ menu and will also be added to user’s account screen.

When forename and/or surname are not provided, the option will default to using usercode.

More Options dropdown:

User Account screen:

Previously, upon detecting a disabled user or a user with a missing email, a schedule job would fail entirely. Now, it will simply skip over these users and continue scheduling to all other non-problematic users without fault. It will also skip over users under a role which are either disabled or have a missing email too.

We now include min_samples_leaf option to the decision tree logic in piAnalytics. This option imposes a constraint on the minimum number of samples required to be at a leaf node that prevents the tree from growing overly complex and capturing noise.

With this setting we can control the level of detail in the decision tree as larger leaf sizes result in simpler trees, which are easier to interpret. Each leaf node will contain at least 5% of the total samples. This setting for min_samples_leaf will be made configurable in the upcoming releases.

Fixed various display and scaling issues of card charts on both desktop and mobile view.

Changed the way the size of the card chart container is being calculated; now we will be using the width-height ratio that is applied to cards in the chart editor. As a result, when having multiple card charts in the same category, more content will be displayed in mobile view.

Example Video below; (Safari, iOS 17.2, iPhone 15):

Improvements have also been made around the scaling of card charts, specifically when displaying large amounts of content.

Fixed an issue where incorrect chart title was used in the excel spreadsheet. Excel export will mirror the chart title displayed for the chart in the UI that is consistent with the rest of export types e.g. if the chart title is defined in the ‘Attributes' section, this chart title will be used in the exported file.

Fixed an issue where the combined chart with multiple measures was breaking if the trend line was being used and the data for all measures was put on the second y axis.

PowerPoint & Excel exports were showing a hardcoded 'NULL' value when NULL values were used in the dimension data. Instead, we now convert any NULL values into empty strings to reflect thew view in the UI. Currently, NULL values in the dimensions are shown as empty cells in the table in the dashboard.

Removed the leading and trailing spaces from the domain values that are used as identifiers in export import functionality e.g. chart name, data connection name, category name, report name, schedule name. All leading and trailing spaces will now be trimmed before being saved to the database.

A high level SQL injection vulnerability (CVE-2024-27289) was reported in one of the go libraries used by the scheduler. Although we don’t believe it would be possible to exploit, we’ve upgraded the library to fully mitigate any risk.

Column added to mis_hierarchies table