CNCF Serverless Workflow specification

This document describes the information about the implementation of the Cloud Native Computing Foundation (CNCF) Serverless Workflow specification. Kogito implements version 0.8 of the Serverless Workflow specification.

The following table shows the implementation status for each Serverless Workflow specification feature.

Kogito does not support Retries, Sleep State, and has a limited support of Timeouts features of Serverless Workflow specification.

Table 1. Implementation status icons
Icon	Description
	Fully implemented feature and compliant with the Serverless Workflow specification
	Partially implemented feature
	Not implemented

Table 2. Implementation status of the Serverless Workflow specification features
Feature	Status	Reference
States		Workflow States
Functions		Function Definition
Events		Event Definition
Serverless Workflow data		Workflow Data
Expressions		Workflow Expressions
Error handling		Workflow Error Handling
Retries		Retry Definition
Timeouts		Workflow Timeouts
Compensation		Workflow Compensation
Constants		Workflow Constants
Secrets		Workflow Secrets

States

The Parallel State of the workflow states feature works in a single thread. This means that a Parallel State does not create one thread per branch, simulating an actual parallel behavior.

If an exclusive property is set to `false`, you should not use the link:{spec_doc_url}#event-state[Event State] of the workflow states feature as the starting state. In case, if it is specified that way, then it will behave as if an exclusive property was set to `true`.

Kogito does not support the Sleep State feature. However, this feature will be supported in a future release.

The following table shows all the workflow states that Kogito supports in the Serverless Workflow specification 0.8 version:

Table 3. Workflow States implementation status
State	Status	Reference
Event		Event State
Operation		Operation State
Switch		Switch State
Sleep		Sleep State
Parallel		Parallel State
Inject		Inject State
ForEach		ForEach State
Callback		Callback State

Functions

The following table shows the status of the workflow functions that Kogito supports:

Table 4. Workflow Functions implementation status
Function	Status	Reference
REST		Using Functions for RESTful Service Invocations
RPC		Using Functions for RPC Service Invocations
Expression		Using Functions for Expression Evaluation
Async API		Using Functions for Async API Service Invocations
GraphQL		Using Functions for GraphQL Service Invocations
OData		Using Functions for OData Service Invocations
Custom		Defining custom function types

For additional functions, the Serverless Workflow specification support the custom function type, such as sysout and java. For more information about these custom function types, see Custom functions for your Serverless Workflow service.

Events

Kogito supports events of the workflow model as defined in the Serverless Workflow specification definition, except the following:

resultEventRef property in EventRefDefinition is not implemented and, if specified, this property is ignored. Same functionality can be achieved by using Callback state.
Correlation has limited support, that evaluates correlation rules matching a single event consumed per time with a workflow instance. The correlation among N different events to be consumed and matched with a workflow instance is not supported in Kogito. For more information about event correlation, see Event correlation in Serverless Workflow.

Serverless Workflow data

Kogito supports the implementation of the data manipulation feature using jq (default) or jsonpath expressions.

jsonpath expressions can have limitations in specific scenarios of complex data manipulation. For more information, see Workflow Data section in the Serverless Workflow specification documentation.

Kogito also supports data filtering. For more information about the data filtering, see State data filters section in the Serverless Workflow specification documentation.

Expressions

Kogito supports jq or jsonpath to define the workflow expressions. jq is the default expression language, however, if you want to use jsonpath, set the expressionLang attribute in the workflow definition as shown in the following example:

Example of changing the default expression language

{
  "id": "myworkflow",
  "version": "1.0",
  "expressionLang": "jsonpath",
  "name": "Workflow example",
  "description": "An example of how to use jsonpath expressions"
}

jsonpath expression is not recommended because of the limitations of the language for data manipulation. Instead, you can use jq in various use cases.

Error handling

Kogito supports the error handling feature as described in the Serverless Workflow specification definition.

For more information about error handling, see Error handling in Serverless Workflow.

Retries

Kogito does not support Retries feature, however, it will be implemented in a future release.

Alternatively, you can use Error handling in Serverless Workflow.

Timeouts

Kogito has limited support for the timeouts feature, which covers only workflow and event timeouts.

For start event state the exclusive property is not supported if set to false, therefore the timeout is not supported for the event state when starting a workflow.

For more information about timeouts, see Timeouts on events for Serverless Workflow.

Compensation

Kogito supports workflow compensation as described in the Serverless Workflow specification definition.

For more information about compensations, see Saga orchestration example in Serverless Workflow.

Constants

Kogito supports workflow constants as described in the Serverless Workflow specification definition.

Secrets

Kogito supports workflow secrets as described in the Serverless Workflow specification definition.

Secrets are associated with the Quarkus Configuration module. This means that any form of configuration that Quarkus supports is also supported in Kogito.

Additional resources

Found an issue?

If you find an issue or any misleading information, please feel free to report it here. We really appreciate it!