Release Notes

Bug fixes:

• Fixed an additional potential issue where it was possible for scopes to not be sent properly when refreshing an OAuth token if the token URL did not echo back the scopes originally in the response.
• The form interface to resume a process instance in error could incorrectly return an error that the instance was not found.
• Optimized working with large boxed contexts.
• When importing an XSD from the data types dialog, the Gregorian XSD types were not imported (gDay, gMonth and gYear).
• When requesting to select a shape in the modeler, if that shape was on another page, it was not always selected properly.

Bug fixes:

• The custom FEEL functions were not always packaged in Digital Distributed Containers.
• The simple FEEL data types type names are not properly rendered in reports.
• Copy of and semantic links could be marked as missing when references came from different graphs.

Bug fixes:

• A fix in version 11.3.2 to the XSD import introduced a non-desired side effect of renaming the virtual attribute to hold the element text when facing a mixed of text and attributes. That side effect was reverted.
• Testing results of decision tables inside boxed context are now properly highlighted.
• The testing result of a decision table is now displayed with the information icon.
• It was possible that a model would refuse to load if the requirement panel was open.
• Sources URI should be clickable in the report.
• Switching to/from FEEL unparsed could cause a DMN 1.2 export to use the wrong FEEL XML namespace.
• In rare cases, the debug raw outputs event for ServiceNodeFinished of a boxed context containing a decision .
• The number of output headers that will be automatically created when creating a new decision table has been increased from 5 to 15.

Bug fixes:

• A concurrency problem that could cause dependent asynchronous events to be executed in parallel could cause a process instance serialization corruption using the filesystem persistence.
• Importing test cases that did not define the TCK name attribute would cause a rendering issue in the test case panel and the test case selection for testing.
• The ability to use rich text when doing the data mapping was not always offered and prevented the data mapping tab from properly expanding.

Bug fixes:

• For digital distributed containers mounting a persistent volume on /data/des/pinstances, there was a possibility that it prevented active triggers from being registered properly.
• When synchronizing a large number of elements from the Digital Enterprise Graph, the modeler user interface could become unresponsive for a period of time.
• Improved the documentation of the any() and all() FEEL functions as well as their non-null equivalents.

Bug fixes:

• Importing the same XSD for the second time could break local data types referencing the types in the XSD.
• Optimized the performance when deleting a large number of data types.
• The FEEL Quick Guide panel should be restored after testing a decision table page.
• Data Mapping for workflow and cases could incorrectly be made read-only by a preference of the Decision Modeler to view the decision logic in standard mode.
• An issue affecting a small number of services could cause unexpected 500 HTTP responses on a REST service call under heavy load.
• Fixed a few typos in the BPMN Quick Guide.
• The include dialog for CQL, PMML, DMN and Custom FEEL functions incorrectly had up/down mouse arrows to reorder them.

Bug fixes:

• The preferences were not properly transfered to Digital Distributed Containers resulting in some containers not being licensed properly.

New and improved features:

• The Digital Distributed Containers family was expanded to also now allow to build containers with multiple services.
  • The existing Digital Distributed Container was migrated into what is now called Single Service Container.
  • Introduced Multi Service Container that allows to push containers that contain multiple services to a container registry from the Service Library.
  • The Single Service Container was enhanced to allow to push containers to an external container registry.
  • The admnistrator can now control if Single Service Containers are built for each execution environment and their destination (local or external registry).
  • Note that separate subscriptions are required for Single Service Container and Multi Service Container. Clients that purchased the Digital Distributed Container will be migrated to Single Service Container on renewal.
• A new automation service storage for long-running services (workflows and cases) based on MongoDB persistence can now be used in the Digital Automation Suite and in Digital Distributed Containers.
• An administrator can now create execution environments that do not register active triggers (Kafka, RabbitMQ, …) when deploying a service. These services can still be built into containers and deployed with active triggers.
• In the Form interface to services, BPMN signals accessible in the context of the current task are now presented as user actions that can be triggered from the user interface.
• When testing a decision model, it is now possible to scope the test to a given decision service, decision or business knowledge model.
• Flags are now displayed in the terms view.
• The test button is not disabled anymore when the workflow process is not marked as executable. It is now possible to make it executable from a new dialog.
• When validating an executable process with the validate button in the BPMN ribbon, it will also be validated for attriubtes required by the Trisotech Automation Suite.
• A data type that has an attributed named instance will not generate a warning anymore.
• It is now possible to iterate over a range of dates in FEEL. This is an upcoming feature of DMN 1.5.
• The Trisotech Help now contains the documentation of the Administrator inferface.

Bug fixes:

• When submitting an XML file to the automation engine, there was a condition that could cause an element defined with a collection data type to be parsed as an element instead of a single element in a collection.
• It was possible for scopes to not be sent properly when refreshing an OAuth token if the token URL did not echo back the scopes originally in the response.
• Links inserted into the rich text editor (description/notes) could fail to open in the Kommunicator because of iframe domain restrictions.
• Adding a second source to a term could result in an error.
• When editing the type of a case file item with the data type overlay on, the data type Any would appear in error on the overlay until a refresh.
• Empty example and notes are not displayed anymore in the term view.
• Labels were not always displayed on connectors.
• Some elements were not properly alphabetically sorted in the graph scope.
• An error could be incorrectly displayed when creating a new model and saving a previous one.
• Optimized the loading type of the Data Type dialog when hundreds of data types were present in the model.
• Optimized the number of graph queries required to sync data types.
• When importing Open API in the Operation library of a modeler, item definitions name were created with their title instead of their name which causes referencing problems.
• When importing Open API in the Operation library of a modeler, simple types definition could incorrectly create circular data types.
• The description field of connectors and annotations were not properly exported to CMMN XML.
• Some decision logic pages were not properly made read-only and returned to editable when testing a decision model.
• Links in comments were rendered smaller then intended.
• Message flows were incorrectly highlighted in blue in the animator.
• Suggestion did not show up in decision table, list and relation columns with a boolean or enumeration data type.
• In rare cases, the DMN XML export of a decision model could contain an import twice. This caused an issue with automating these models.
• Some errors were not properly reported to the user on a failed deployment to the automation.
• When using the dot (.) operator after a filter in a literal expression, null values were incorrectly filtered out of the result.
• When testing or automating an invocation that was calling a function with an undefined prefix, it could cause a non descript exception. The error is now clear.
• When reloading the Discovery Accelerator with an opened model sometimes all graph links are marked as missing.

Non-Backward Compatible Changes:

• The monitoring and health checks REST operations were removed from the main HTTP port (8181) and moved to a dedicated port (9080 inside the container). These operations are no longer protected by a basic authorization on the new port and port 9080 should not be exposed publicly. Clients hosting the Digital Enterprise Suite that previously used port 8181 with a basic authorization to access the health check functionality will need to update to port 9080 without authorization.

Bug fixes:

• Fixed a rare condition that could cause a timer trigger to be unregistered before being registered causing other triggers of the process to be removed as well.
• In the workflow modeler, when saving your models under certain conditions, relative digital enterprise graphs to process, decision or case could be made absolute.

Bug fixes:

• For clients not subscribing to the Trisotech Automation Suite, models will not be upgraded to use custom FEEL functions anymore.
• The DMN XML export had the wrong value serialized in the name attribute of a custom FEEL function library import element.
• The non-null functions (ex: nn count(), nn sum(), …) are still recognized properly by the FEEL highlighter.
• A digital enterprise graph semantic link was incorrectly added on top of a reuse link for BPMN call activities.
• The authorization header of the Digital Distributed Container base image can now be properly resetted from the admin interface when left empty.

Bug fixes:

• An initialization issue could rarelly prevent a decision model from properly loading in the Decision Modeler.

New and improved features:

• Custom FEEL function libraries are now available for modeling and automation.
  • The Trisotech extended (non standard) FEEL functions were removed by default from the Learn ribbon.
  • Custom FEEL libraries can be added to a model through the Learn ribbon (custom drop-down) or the standard ribbon (BPMN/CMMN/DMN) in the include section.
  • The initial offering for the custom FEEL libraries includes:
    • Automation: access the automation runtime. Those functions were previously always available in the default context.
    • Financial: domain-specific functions for the financial domain. Those functions require the optional Financial Feature Set subscription.
    • Healthcare: domain-specific functions for the healthcare domain. Those functions require the optional Healthcare Feature Set subscription.
    • Non-Null: specialized list functions that better handle null in lists. Those functions were previously always available in the default context.
    • Utility: conversion utility methods. These initially include converting from unix/epoch time to FEEL date and time, encoding/decoding JSON/Base64.
  • Some of the functions (Automation, Non Null, Healthcare) that previously existed in the default namespace had their names slightly changed to be namespaced in their new library (ex: healthcare term codes() becomes healthcare.term codes()). That change is backward compatible with previous version of the automation and the models will automatically get updated with the new function names. Note that the automation backward compatibility will be guaranteed only for the 11.X serie and models will need to be published before the release of the 12.X serie.
  • A few functions in the default FEEL namespace are not yet standard and were documented accordingly.
• The delivery platform for most of the modeling suite was updated to provide much faster load times.
• The data type interfaces now allows to import data types coming from a JSON schema.
• The OpenAPI endpoint for receiving messages on workflow services will now explicitly have the name of the message its receiving in its description.
• The Trisotech Help was moved to a new platform.

Bug fixes:

• Textual search in the modelers was not finding text in BPMN annotations, certain types of links and some other minor elements.
• The learn tab was not displayed in the Landscaping Modeler.
• Using the copy and paste modeling functionality on an element that is reused from the graph could not paste the graph link properly.
• Deleting the last row of a boxed context inside a Boxed Function would not render properly.
• Some models were generating duplicated identifies when exporting to CMMN XML.
• When binding a data store with a definition containing a non-typed column, it would cause a typing issue.
• The public API functionality to download a model with all of its dependencies (/publicapi/repositorycontent/package) will now consider modeling dependencies. It previously only considered execution dependencies.
• Deployment errors would prevent the ability to test a decision model.
• SMART on FHIR application relying on patient inputs could only receive their identifier at runtime.
• Some case file item data types could cause the automation form to think that a case file item is not available to a user incorrectly.
• Notes added to a new term could be lost when reloading the browser.
• Semantic link information were not always properly updated when the target of the link was renamed.
• The Microsoft Visio import was not working properly for the Case Modeler.
• The Save As action could fail in rare cases when workflow models were referenced from a different place.
• The change data type button was still usable when displaying the interface in read-only mode.
• Fixed an issue where sometimes multiple dialog opened in the relationship view.
• Sometimes copy of links were incorrectly reported as missing after the referred model was saved.
• Keyboard shortcut could not be used anymore after a Visio import.

Bug fixes:

• When selecting to not gather end events or milestones in a workflow or case model, it is now possible to use the end/milestones variables for other purposes.
• Under certain conditions, the OpenAPI documentation of a message was not aligned with its definition.
• When publishing a decision model, validation errors and warnings were not displayed in the user interface.
• In some rare cases, a missing data definition was assumed to be a String instead of an Any for workflow automation. This could cause typing warnings.
• When validating a decision model that had missing dependencies, the validation errors were not properly displayed.
• A workflow process using an event based gateway with instantiate set to true was incorrectly reporting a validation error about a missing start event.
• The order of the user tasks inner data inputs were not respected when displaying the input form at automation.
• Corrected an event serialization issue for FEEL functions that could cause performance issues in while testing a decision model.
• It was not possible to delete a disabled user.
• The RedHat connector was not using the right version of DMN when publishing.
• Dragging a BKM from the Digital Enterprise Graph could fail.
• Data stores were sometimes incorrectly synching with the Digital Enterprise Graph.

Bug fixes:

• The description of process, user tasks and decision models were not always displayed when starting or resuming a service.

New and improved features:

• Improvements to the test and form functionality of workflows and decisions automation:
  • The test input tab was renamed to in and the output tab to out. They are now always visible across the test session.
  • The test input mode selector was redesigned to use buttons instead of a drop-down to make it easier to find.
  • A button was added to lists to allow to upload and download the list data to/from Excel.
  • The test trace detail dialog now display a human readable rendering of the inputs and outputs.
  • A new input mode is now available to test the service from Test Data (requires the automation subscription).
  • The chevron next to collections and structured data type was replaced with a caret to expand or collapse the input.
  • Collapsing a collection or structure does not make its value null anymore, a separate null indicator was added that is either black (value is null) or gray (value is not null but collapsed).
  • Collections inputs and outputs now display borders to improve readability.
• The service OpenAPI v3 generated documentation now includes the nullable attribute to true to better reflect the actual behavior of the service.

Bug fixes:

• For some types of workflow nodes, the NodeStarted and NodeCompleted events had incorrect inputs/outputs/rawInputs/rawOutputs values.
  • inputs: the values actually passed to the task after the input mapping
  • outputs: the values returned by the task before the output mapping
  • rawInputs: the process variables used as input before the data mapping was applied
  • rawOutputs: the process variables returned to the process as output after the data mapping was applied
• NodeStarted events for decisions will not have inputs in the debug profile.
• Data objects/inputs/outputs with newline characters in their name would not display their data content in the test panel when focussed.
• When testing a decision or workflow, in some cases it was possible for the information marker to incorrectly display [Object object] instead of the result of the evaluation.
• Refreshing an OAuth 2 token with a token provider that does not return a scope should not cause a refresh failure anymore.
• Most Smart on FHIR application were failing to start with an error.
• Script tasks using data stores without a script defined were not able to access the datastore properly.
• The Message Event dialog could fail to close in some cases.
• New item definitions could not be created in the data object data dialog.
• Custom accelerators that does not define a known type for a node could cause an error preventing the semantic link panel from opening.
• It was not possible to successfully assign a performer on a workflow process.
• When displaying a null result for a structured data type, the test and form output were incorrectly rendering all its attributes as null instead of marking the element as null.

Bug fixes:

• In some client configurations, the Kommunicator was not able to load workflow models anymore.
• For rare cases of data definition, the OpenAPI documentation of a collection was incorrectly represented as a collection of collection.
• The order of the results of a service execution (test in modelers or through form in service library) could be represented in the wrong order.
• Loading a test case in the input form when a collection was changed to not be a collection anymore could cause the import to stop.
• Rendering the service results of a list of a simple type that was defined as an Any field could cause an issue that prevented the rendering of that field.