* Refactor AdditionalConvertersConfigurator constructor
Simplify AdditionalConvertersConfigurator by using constructor injection directly in the class declaration. Also, annotate the class with [UsedImplicitly] to improve code clarity and maintainability.
* Update workflow serialization to be asynchronous
Refactor the workflow definition serialization to use asynchronous operations, improving efficiency and handling cancellation tokens. This change ensures the schema is correctly included in the serialized output.
Introduce JsonNodeBsonConverter for BSON serialization of JsonNode objects. Update MongoDbFeature to register the new serializer, enhancing JSON handling within MongoDB serialization context.
Switched from a hardcoded version to a property variable `$(SystemTextJsonVersion)` for the `System.Text.Json` package reference in multiple project files. This change centralizes version management and ensures consistency across all projects.
* Add endpoints for activity execution records and summaries
Created new classes and endpoints to handle activity execution records and summaries retrieval. Added related models and updated relevant interfaces.
* Rename endpoint classes for clarity
Updated endpoint classes in `ActivityExecutions` and `ActivityExecutionSummaries` to use the name `Endpoint` instead of `List`. This improves readability and consistency in the codebase.
Replaced private fields with constructor parameters in IdentityGraphService and introduced a logger. Added logging to handle cases where activity descriptors are not found, improving debugging and maintainability.
Enhanced exception messages to include activity type and version for better debugging. This provides clearer context when activity descriptors or inputs are not found.
* Switch to Azure Service Bus for message brokering
Updated the message broker from in-memory to Azure Service Bus to enhance scalability. Added new configuration settings and dependencies to support Azure Service Bus integration, including scoped service initialization and queue/topic management.
* Refactor StartWorkers for dependency injection
Updated StartWorkers class to use constructor parameter injection and service scope factory for retrieving services. This refactor eliminates the need for field-level dependency storage, enhancing testability and adherence to the dependency injection principles. Additionally, annotated the class with [UsedImplicitly] to eliminate unused code warnings.
* Refactor Worker to use IServiceScopeFactory for dependency resolution
Refactored the Worker class to use IServiceScopeFactory for DI instead of directly injecting IWorkflowInbox. This enhances the flexibility and lifecycle management of the dependencies.
* Add Azure Service Bus integration toggle
Introduced a new configuration constant `useAzureServiceBus` to control the optional integration with Azure Service Bus. Implemented conditional logic to bind Azure Service Bus options from configuration if the flag is enabled.
* Add IComposite interface and implement Setup method
Introduced a new IComposite interface and added a default Setup method to the Composite class. Updated the ActivityFactory to call the Setup method when an activity is an instance of IComposite. The CompositeExample class demonstrates the use of the new interface and method.
* Remove AzureServiceBus configuration setup
The AzureServiceBus configuration setup was removed from the Program.cs file. This change aims to streamline configuration and remove unused or unnecessary setup, ensuring the code remains clean and maintainable.
* Switch MassTransit broker to in-memory
Changed the MassTransit broker from AzureServiceBus to Memory in the application configuration. This adjustment aims to simplify deployment and reduce dependencies in the current environment.
* Improve dispatched workflow input handling
Addressed input handling in dispatch messages by adding `SerializedInput` property. Also removed initialization logic and moved input deserialization to a helper method, ensuring compatibility with both new and deprecated input property formats.
* Update workflow Docker images and version tags
Changed Docker image tags from v3-2-0-rc3 to v3-2-1-preview across multiple GitHub workflows. Updated the VERSION environment variable in packages.yml to reflect the new versioning scheme. These changes ensure consistency with the new preview release.
* Update versioning to include 'preview' in package workflow
Modified the workflow to append 'preview' to the version number for non-tagged builds. This ensures clearer differentiation between stable and non-stable versions in the CI pipeline.
* Add WorkflowInstanceStorageDriver for workflow variable storage
Introduced a new storage driver, WorkflowInstanceStorageDriver, to store workflow variables directly in the workflow state. Updated relevant classes and methods to incorporate this new storage driver, ensuring seamless read/write/delete operations and extending support for it throughout the codebase.
* Refactor object conversion and update variable retrieval.
Switched from JsonObject to JsonNode for object conversion and corrected a typo in the summary comment. Changed the return type of GetVariablesDictionary method and updated its implementation to use VariablesDictionary.
* Rename 'input' to 'serializedInput' in DispatchWorkflowDefinition.
This change clarifies that the input provided to the workflow should be serialized. It enhances the readability and accuracy of the code documentation, ensuring that developers understand the expected format of the input parameter.
* Add priority and deprecation attributes to storage drivers
Introduced a priority attribute to the `IStorageDriver` interface and implemented it in various storage drivers. Additionally, marked `WorkflowStorageDriver` as deprecated and reordered storage driver listing based on priority.
* Switch MassTransit broker to in-memory and refactor converter
Changed MassTransit broker from AzureServiceBus to in-memory for improved performance in development environment. Simplified PolymorphicObjectConverterFactory by removing redundant constructor and dependencies. Removed unused folder from the project file.
Refactor DataProcessingException to remove unique key logic and create a dedicated UniqueKeyConstraintViolationException. Update DbExceptionTransformer to use the new exception for unique key violations, improving code clarity and separation of concerns.
Removed the `ConverterHelper` and `ExpandoObjectConverter` classes, along with their integration in `JintJavaScriptEvaluator`. Updated unit tests to skip unsupported functionality and added references to related issues for context.
Introduce tests for resuming specific HTTP workflow instances. Enhanced `HttpWorkflowsMiddleware` to handle workflow instance IDs and correlation IDs for more precise control over workflow activation and resumption.
Replaced IWorkflowDispatcher in TaskReporter with IWorkflowInbox to streamline task reporting. This fixes race conditions when the workflow is attempted to be resumed before it committed its bookmarks.
* Add handlers and improve JavaScript engine configuration
Added several new handlers for configuring the JavaScript engine with common types, functions, variable accessors, and input/output accessors. Refactored the JintJavaScriptEvaluator for better clarity and modularity by breaking down configuration steps into separate methods. Also replaced `IsInsideCompositeActivity` with `IsContainedWithinCompositeActivity` for better semantic consistency.
* Refactor to use primary constructor syntax for TypeDefinitionProviders
Updated CommonTypeDefinitionProvider, VariableTypeDefinitionProvider, and ActivityOutputFunctionsDefinitionProvider to use primary constructors for dependency injection. This change reduces redundancy and simplifies the code structure for better readability and maintainability.
* Add byte conversion functions to JavaScript engine
Introduced functions for converting bytes to/from strings and Base64. Updated CommonFunctionsDefinitionProvider and ConfigureEngineWithCommonFunctions to incorporate these new functions. This enhances the JavaScript engine's ability to handle byte array manipulations.
* Add tests for JavaScript byte and string conversions.
Introduce integration tests to verify byte array to string, string to byte array, byte array to Base64, and Base64 to byte array conversions using JavaScript functions. Ensure accurate transformation of data within different encoding scenarios.
* Add meaningful summaries to Jint engine configuration handlers
Updated comment summaries in four handler classes to provide clear and concise descriptions of their purpose. This helps improve code readability and understanding for future developers.
* Rename and merge variable and input/output handlers
Merged the variable accessor logic into the input/output handler and renamed the class to reflect its broader functionality. This consolidation ensures the accessors are registered in the right order.
* Add string base64 conversion functions
Introduced `stringToBase64` and `stringFromBase64` functions to handle base64 encoding and decoding of strings. Updated corresponding provider, handler, and added tests to ensure functionality.
Introduced `ConverterHelper` and `ExpandoObjectConverter` to handle conversions from .NET dynamic objects to JavaScript objects. Updated integration tests to validate the conversion and sorting functionalities.
Fixes#5895
* Add optional refresh parameter
Renamed several `RefreshDescriptors` methods to `RefreshDescriptorsAsync` to reflect their asynchronous nature consistently. Added an optional refresh parameter in the `ListActivityDescriptorsRequest` class and updated related endpoints to handle this parameter, triggering a registry refresh if necessary.
* Update version to 3.2.0-rc5 in GitHub Actions
Modified the `packages.yml` workflow to change the version from `3.2.0-rc4` to `3.2.0-rc5` for non-tagged releases. This ensures version consistency with the new release candidate.
Switched to dictionary index access for "ResumeAt" in Delay.cs to ensure more efficient updating. Added handling for byte[] in ObjectConverter.cs to correctly deserialize from base64 strings.
Introduced customizable database exception handlers across various modules, enabling specific error handling strategies for different DbContexts. Implemented a generic interface for exception handling and provided specific implementations for PostgreSQL and a No-op handler. Updated the code to utilize these handlers in various persistence features, enhancing error management and modularity.
* Refactor workflow context extension structure
Removed unused WorkflowContextWorkflowDefinitionExtensions and renamed several extension and middleware classes for clarity and consistency. Simplified context provider type handling using JsonArray in custom properties.
* Refactor context provider and update email workflow
Refactored `CustomerWorkflowContextProvider` to use constructor parameter directly, removing redundant fields and constructors. Updated comments and list syntax in `CustomerCommunicationsWorkflow`. Improved README.md documentation for workflow contexts and context providers.
* Adjust logging levels and improve workflow logic
Modified `appsettings.json` to reduce the verbosity of logging by adjusting log levels. Updated `CustomerCommunicationsWorkflow` to include a timer activity and simplified the workflow context parameter setting.
* Remove PropertyBag and related functionalities.
Deleted the PropertyBag class and extension methods, and removed all instances where PropertyBag was utilized across various modules. This simplifies the workflow configuration and management by relying solely on CustomProperties for storing metadata.
* Add workflow context management extensions and optimizations
Introduces `WorkflowExecutionContextExtensions` for loading and saving workflow contexts more efficiently. Replaces `JsonArray` usage with `List<object>` for context provider types, simplifying the type management in the workflow execution context. Streamlines `WorkflowContextActivityExecutionMiddleware` by removing redundant service scope creation and enhancing readability.
* Remove redundant timer in CustomerCommunicationsWorkflow
The timer activity that was set to not start the workflow has been removed to streamline the workflow process. Updated the README to include instructions on how to start the CustomerCommunicationsWorkflow using the REST API or Elsa Studio.
* Refactor middleware to streamline context management
Removed dependency on `IServiceScopeFactory` and refactored middleware to directly load and save workflow contexts. Simplified workflow context provider invocation, enhancing code readability and maintainability.
* Remove redundant PropertyBag property
The PropertyBag property was commented out and is no longer needed in the WorkflowDefinition class. This change removes the commented-out code to clean up the class definition.
* Add JsonSerializerOptions for deserialization
Introduce JsonSerializerOptions with camelCase policy and case insensitivity for property names. This ensures consistent deserialization of custom properties in activity workflow context settings.
* Add activity execution log sink
Introduced an interface `IActivityExecutionLogSink` for storing workflow execution log records and implemented it as `StoreActivityExecutionLogSink`. This centralizes and streamlines the handling of persisting execution logs into the workflow's runtime, thus making the process more modular and maintainable. The `PersistActivityExecutionLogMiddleware` has also been refactored to use this sink, replacing the direct usage of the workflow execution log store.
* Refactor logging interfaces and implementations
Replaced specific logging interfaces and implementations with generic ones. The individual interfaces for workflow and activity execution logging have been replaced with a single ILogRecordExtractor and ILogRecordSink interface. Specific implementations have been adjusted to use these new interfaces. This allows for greater flexibility and reuse of logging code.
* Introduce ILogRecordStore interface
The `ILogRecordStore` interface has been introduced to provide a centralized place for handling log records. Both `IActivityExecutionStore` and `IWorkflowExecutionLogStore` have been updated to inherit from this new interface. As a result of this change, the `SaveManyAsync` methods in these two interfaces have been removed to avoid redundancy.
Moved specific services from Scoped to Singleton in SchedulingFeature and SystemClockFeature for better performance and consistency. Refactored DefaultWorkflowScheduler to use a constructor with an IScheduler parameter and removed redundant private field. Removed attribute RequiresUnreferencedCode in WorkflowManagementFeature.
* Rename and refactor journal update endpoint
Replaced `/workflow-instances/{id}/journal/has-updates` endpoint with `/workflow-instances/{id}/updated-at` to simplify API responses. Deleted `HasUpdates` related classes and introduced `GetUpdatedAtResponse` for consistency and clarity. Updated client contracts accordingly.
* Remove HasUpdates endpoint and refactor workflow observer
Deleted the HasUpdates endpoint and refactored related code to use an updated timestamp approach instead. Improved nullable handling in WorkflowInstanceDesigner and ensured proper observer disposal to avoid memory leaks. Updated workflow observer factory and observer implementations to support observer names and enhanced logging.
* Rename updated workflow instance endpoint and handle execution state
Renamed the endpoint from "/updated-at" to "/execution-state" to better reflect its purpose. Updated related response models and documentation to capture workflow execution state details such as status, sub-status, and last updated timestamp.
* Enable SignalR for real-time workflows
Add a flag to use SignalR and activate real-time workflows when enabled. Refactor code to wrap SignalR setup in conditional checks based on the new flag. This enhances the application's interactivity through real-time capabilities.
* Remove obsolete endpoints and rename execution state paths
Deleted the outdated Api1 and DynamicWorkflows endpoints under Elsa.Server.Web. Also, renamed paths related to execution state models and endpoint to remove "Journal" from the namespace for better clarity and organization.
* Add reloading logic for workflow definitions
Introduced a new mechanism to handle reloaded workflow definitions. This includes creating a "ReloadedWorkflowDefinition" model, updating the caching logic, and modifying notification handlers to work with the enhanced workflow reloading logic. This ensures workflow definitions are updated and managed correctly when published, retracted, or deleted.
* Improve RefreshActivityRegistry documentation
Updated the XML documentation to clarify that `RefreshActivityRegistry` refreshes the `IActivityRegistry` for `WorkflowDefinitionActivityProvider` whenever workflow definitions are reloaded, instead of when they are published, retracted, or deleted.
* Add 'materializer' to user dictionary
The term 'materializer' has been added to the user dictionary to improve code spelling and naming consistency. This change ensures that 'materializer' is recognized as a correct term in the codebase.
* Organize test files by adding a 'Workflows' directory
Renamed 'http-workflow.json' to indicate it belongs under 'Workflows'. This improves file organization and clarity within the 'WorkflowDefinitionReload' scenario.
* Add TestWorkflowMaterializer and TestWorkflowProvider
Introduce `TestWorkflowMaterializer` for deserializing workflows from `TestWorkflowProvider`. Added integration of these new components in the `ReloadWorkflowTests` and `WorkflowServer`. Also renamed `RemoveReloadWorkflowTests` to `ReloadWorkflowTests`.
* Rename and expand workflow reload tests
Renamed `RemoveReloadWorkflowTests` to `ReloadWorkflowTests` to better reflect its purpose and expanded with additional test cases. Added tests to verify workflow and activity registry updates after source provider changes and workflow reloads.
* Update copy settings for workflow test scenarios
Reorganized and added 'CopyToOutputDirectory' settings for JSON files in workflow test scenarios. Ensured all necessary files are correctly included and copied during output directory builds to maintain test consistency.
Introduced three new functions: `getWorkflowDefinitionId`, `getWorkflowDefinitionVersionId`, and `getWorkflowDefinitionVersion`. These functions provide easy access to specific workflow definition details within the JavaScript evaluator. Removed redundant constructor in CommonFunctionsDefinitionProvider to streamline the code.
Replaced the use of ActivityId with ActivityNodeId when retrieving distinct activity IDs. This change ensures that we are using the correct node identifier for logging workflow progress.
Enhanced the Revert endpoint to return a summary of the newly created workflow definition and include its location. Adjusted the API contract and endpoint configuration accordingly.
This commit introduces the ActivityNodeIds filter to the journal request models. This enhancement allows filtering logs based on specific activity node IDs, providing more granular control over workflow instance logs. Additionally, unnecessary summary comments have been removed for clarity.
* Fix issues with workflow deletion, newline, and DB context.
Retract published workflows before deletion to avoid exceptions. Added missing newline at EOF for consistency. Removed redundant DisposeAsync call in migration service.
* Add PropertyBag for storing custom workflow properties
Replaced `CustomProperties` with `PropertyBag` class for more structured property management across workflows. Adjusted serializations, extensions, and middleware for seamless integration with the new `PropertyBag` structure.
Introduced a new service, WorkflowExecutionLogRecordExtractor, to abstract the logic for extracting workflow execution logs records from the WorkflowExecutionContext. Updated StoreWorkflowExecutionLogSink to use this new service, which simplifies the execution log persistence method. This modification enhances code readability and enables potential reuse of the extraction logic.
