Prerequisites

Read an overview on Petri’s architecture for a general understanding before moving forward.

Integration For JVM Applications

If you are using a JVM app, please read more here

Integration For Non-JVM Applications

If you don’t run a JVM application you may call Petri server to perform the experiment for you and get the resulting testgroups via a REST API. This can be achieved by deploying the sample app. Note: The URL is defined in laboratory.properties under the WEB-INF folder.

When you use this setup, it is important you understand how Laboratory reads and writes context:

  • Relevant context is read from the HTTP request (user id, language etc). See the relevant code for exact header/cookie/param names
  • Previous conduction values are read from Petri’s cookie, so consistent experience is maintained per user.
  • Conduction values (where relevant) may be written back to the cookies.

This means there are two options for using this service:

  • For issuing calls directly from your client code / the browser

  • From your non-JVM application, and then you may want to pass some of the context to the request via one of these 2 options:

    • Add the userId param for example, or the previous conduction value from the cookie
    • Pass your own custom context and call the conductExperimentWithCustomContext method on the SampleAppController. Combine this with writing your own custom filters .