You are viewing the documentation for Blueriq 17. Documentation for other versions is available in our documentation directory.
Introduction
This page describes the mechanism of loading the right case with the right application in Blueriq so that it is possible to:
- Deploy from different branches without breaking the runtime application.
- Have lists that show cases and tasks from different projects.
Processes are made unique in Blueriq by a so called application ID (applicationId). This ID is an aggregation of the branch version and the project name. In reality an example would look like:
export-TestProject-1.0-Trunk
In this article we will use the notation "a:1" in the schema of "<name>:<version>" . This ID is used to make sure to always address the correct case from a case list or work list. The same behavior applies using message events. So having an unique identity is a good thing, but it also limits the way of deploying models and expanding an application using more than one projects.
With this mechanism we are able to fulfill the following requirements:
As business I want to be able to deploy from different branches without breaking the process, so that it is easier to manage my deployment strategy.
Behavior
This behavior is different to how it previously worked! Before R11 clicking an application determined what case could be run. The drawback of this approach was that it needs a polling mechanism.
Since R11 the mechanism is as follows:
- a case that has to be started triggers the application(s) that is/are needed to be able to run.
This behavior has a few pros:
- The system doesn't have to poll every time, it can now push
But it also has some cons:
- Determining which application is relevant for a specific case might take longer.
Without options
Without any options the name and version of a case will be matched against the list of all applications. When there is an exact match, that application will be picked.
Settings
No additional settings are set.
Example
Cases | Applications | Comments | |
---|---|---|---|
a:1 | → | d:1 | case a:1 will trigger:
|
b:1 | b:1 | ||
c:1 | a:1 |
Ignore Options
It is possible for every application to configure what should be ignored. You would ignore this version when ... $TODO ...
The following ignore options will influence the behavior of loading the application ID.
Ignore setting
Ignore setting | Description | Match on | Comment |
---|---|---|---|
none | The default behavior. The application Id is still regarded by lists and messages. | name and version | This is default behavior like prior to Blueriq 9.6. So no migration is needed. |
project | Ignores the project name part of the applicationId. This can be used when multiple projects on the same branch share the same process model. | only version | This setting was introduced in Blueriq 9.8.5. |
version | Ignores the version part of the application Id. This can be used when models are deployed from different branches. | only name | Typically for deploying from hot fix branches and different deployment strategies. |
all | Ignores version and project name from the application Id. This can be used when there are pages that wants to collect information from different projects. | nothing | Typically used in a dashboard for DCM applications. |
Example
In the following example we use all different kinds of settings to illustrate what their effect is.
Cases | Applications | Ignore | Comments | |
---|---|---|---|---|
a:1 | → | d:2 | all | case a:1 will trigger:
note that application b:1 will not be triggered, because they don't match on name and on version. |
b:1 | b:1 | none | ||
c:1 | a:3 | version | ||
d:1 | e:1 | project |
Levels
It is also possible to determine on a higher level how the mechanism should work in general, so this is then for applications overall.
Level | Description | Comment |
---|---|---|
Server | The settings apply for the complete application. | A general setting if all behavior for the application is the same |
Project | A setting per project if settings only apply for a single project. | These settings overrule the server setting. It is possible to have a mix. So if no project setting is available, then the server setting applies. |
Components
On top of all properties we have the following:
Component | Description | Comment |
---|---|---|
Process engine | Controls which part of the application ID is ignored by the process engine when evaluating timer nodes and handling message events. | |
Process list | Controls which part of the application ID is ignored when displaying cases in Case Lists and tasks in Work Lists. For backward compatibility, when process list settings are not configured, the process engine settings apply. | This component was introduced in Blueriq 9.8.5. |
The settings are part of the application.properties
for Java.
Rules for process engine
When handling timers and message events, the Process Engine of an application A uses the following rules to determine which part of the application Id should be ignored:
- check the
process engine
setting for application A; if not configured then... - check the global
process engine
setting; if not configured then... - use
none
as default
Rules for process list
- check the
process list
setting for application A; if not configured then... - check the global
process list
setting; if not configured then... - check the
process engine
setting for application A; if not configured then... - check the global
process engine
setting; if not configured then... - use
none
as default
Configuration
Work list and case list
The default behavior for building a work list or a case list is a query on the database which uses an application Id. If the version Id and the project name of the current model definition do not match the stored version Id and project name, no tasks and cases will be shown in the corresponding lists. With the new settings it is possible to get the required behavior.
Below an overview is given of the interaction between the different components.
Throw message event
The default behavior for throwing a message event is that it is thrown to the version and project of the corresponding application Id. So if we have a solution which consists of multiple projects, a broadcasted message should be broadcasted within the solution. It is possible that model wise the process is modeled in one module which is a library and imported in multiple projects. With the new settings it is possible to get the required behavior.
Warning
If there is a deployment from a different version which is missing models compared to the current application, there is a possibility functionality won't work anymore. So ignoring versions can complicate things. For example a work list shows all tasks within a case ignoring version. But a task definition has been deleted and is not deployed from a new version. The task is still part of the process database, so it will show up in the work list because the version is ignored. Starting the task from the list will result in an error, because it cannot find its model definition.
Example
Below an example is given to get a good understanding on how to address the properties:
# global setting, accepted values = none, project, version, all blueriq.processengine.default-app-id-ignore-mode = none # per-applicaton setting; format = blueriq.processengine.app-id-ignore.<application name> = none | project | version | all blueriq.processengine.app-id-ignore-modes.studio-ACO-First = all blueriq.processengine.app-id-ignore-modes.studio-ACO-Second = all