August 2025 Releases

The null option within the oneOf schema could be auto-transformed to an undefined value. This transformation caused the schema validation to fail. Subsequently, it blocked the selected null option from saving on the content page.

When handling changes under the component edit form, the fix adds a transformation layer to ensure the unexpected undefined value is normalised to null (for oneOf composition only).

The null option within the oneOf schema can now be saved successfully on content pages.

The Preview option in Page Builder content page assets does not work with a custom /_nocache suffix.

All customizable suffixes on the Content Management System Configuration screen are no longer editable from this release onwards.

Effective date: August 1, 2025

Due to the PHP 8.1 upgrade in function signatures, htmlspecialchars was escaping single quotes in the nested content area. The system did not properly identify the GET parameters in nested content, and they were not saved.

The fix adjusts the htmlspecialchars behaviour to prevent escaping single quote characters, which allows GET parameters to be saved.

Squiz Content Management was generating warnings for keywords that accessed attributes that did not exist.

Depending on the configuration, warnings would appear on the frontend. More generally, the error log would contain many warnings about missing attributes that were ignored because the implementation worked fine with the missing data.

The fix makes adjustments to the asset_attribute keywords so they do not produce any warnings when something is wrong with the keyword. This change brings this keyword behavior inline with most other keywords. Implementations (such as designs) that generically reference attributes that can sometimes be missing will log fewer warnings.

The CDP search database would only scan 100 records (the first page of results) instead of scanning an unlimited number of records when searching for a customer. When users searched for a particular customer, not all records came back in the result.

The fix removes the database limit scan of 100 records which allows all records can be scanned when searching for a particular customer in CDP search. Users can now view all records when searching for a customer.

The Human-readable document type plugin supports defining a default filetype label, which is set for documents that do not match more specific rules.

Users can configure the label that will be set when there is no mapping for the detected file type.

Pre-defined plugin labels for different file types are now optional and can be turned on/off, giving users more control over how to customize content.

The removal of inline editable fields did not clear the field completely.

The removed field still had the empty <p> tags retained and rendered on the frontend.

The fix applied ensures that if an inline editable field is empty, the value is set to undefined to clear the primitive value when inline editing.

The inline editable field can now be fully cleared without any empty <p> tag remaining.

To ensure consistency with the behavior of standard page assets, cloned content pages assets now use the live page version instead of the draft page version.

The Suggested Tags feature for thesaurus terms under the Tagging screen was identified as broken in a bug report. Upon investigation, the feature was broken in multiple ways, and rebuilding the screen using modern UI practices offered a better resolution pathway.

The Suggested Tags feature has been rebuilt inside the Tagging screen. The rebuild keeps the core functionality unchanged while improving the user interface and the experience of selecting tags.

Due to a PHP upgrade to the underlying platform, a TypeError was returned when a date/time form field was set to use 12-hour time. The form would not load on the frontend, producing a fatal PHP error.

The code has been adjusted to ensure that empty values for the time components are passed back as strings. Users can now successfully set and save 12-hour time entries in date/time form questions without error.

Passing an array to addslashes was triggering a fatal error, due to stricter parameter types. The fix ensures array values are properly handled and parsed, so that addslashes always gets a string passed. Due to the proper handling of array parameters, addslashes always gets a string and does not throw a fatal error.

Passing null to rawurlencode triggered a deprecation warning, due to stricter parameter types.

The fix ensures null values are cast to empty strings, so that rawurlencode always gets a string passed.

When converting from null to empty strings, rawurlencode always gets a string and no longer throws a deprecated warning.

A value was being used in multiplication without validating the data type first, triggering warnings on the Search Page asset when that value was incorrect.

The fix ensures that the value is validated before any calculations. The non-numeric warning no longer appears on the Search Page asset.

Updates to the Squiz DXP platform prevented some previously valid automatic type conversions. Mathematical keyword modifiers (add, subtract, multiply, divide, modulo, and square root) produced a fatal error when working on non-numeric values.

The fix ensures non-numeric arguments to mathematical keyword modifiers are now treated as 0 (zero) Fatal errors caused by invalid type conversion no longer occur for mathematical keyword modifiers.

When an error occurred with a request from a REST Resource asset, the resulting message was erroneously passed as a format string for the error function.

A syntax error resulted in a 500 Server Error response if the message contained percent (%) characters, losing the error message in the process.

The fix ensures the format string is specified correctly, and the message is rendered within that string. When an error occurs with a REST Resource request, the resulting error message is now available for diagnosis.

Personas and variations have typically been a very rarely used feature that impacts the performance of Squiz Content Management systems.

New Content Management instances will not have access to personas and variations from the effective date.

Existing instances will see a deprecation warning when interacting with these asset types. Existing personas and variations will continue to function as normal.

Effective date: August 21, 2025

When personalization with personas was deactivated on the Feature configuration screen, the system still checked for available and applied personas on standard page assets' Content screen.

The fix ensures that the system no longer checks for available or applied personas if turned off, delivering faster loading times on the Content screen of standard page assets.

Previously, the Content screen would perform multiple slow queries to determine what component templates were available for each component on the screen.

The improvement now uses a single, relatively fast query to determine the available component templates regardless of the number of components on the contents screen.

There are two side effects:

The event system previously looked up some system asset IDs from the database already in memory.

The system now skips database queries for any system asset IDS looked up from the event system. This change results in a performance improvement in the event manager.

Previously, the system calculated asset type descendants through database queries, which added latency to the dependency resolution process. The system now calculates asset type descendants purely in code, which means workloads that depended heavily on that calculation (for example, Metadata-related asset fields) now run faster.

Metadata field-related asset root nodes were fetched separately from the database on all backend screen loads. This retrieval method resulted in extra queries that slowed down screens, particularly those with heavy use of metadata like standard page Content screens with many container templates.

Metadata field-related asset root nodes are retained with the asset object in DeJa Vu, meaning no extra time is required to load these fields compared with other field types.

Previously, the template applied to a component was looked up from the database for every screen load in the backend.

The applied template is now cached to DeJa Vu, improving the performance of Content screens.

The system was querying the database for metadata schemas, sections, or fields that belonged to each other when one was already in memory, resulting in unnecessary queries.

The system now keeps associated metadata schema, section, and field links cached within DeJa Vu, resulting in fewer DB queries.