spock | Automatically perform git commits , pushes , and other cli | Runtime Evironment library

Aside from updating the other dependencies to later versions and aligning groovy versions, which ought to resolve it, a quick fix is to declare a system property:

You can parallelise Spock testing on several levels, i.e. per specification (test class) or per feature (test method). The per-method setting also means that in iterated (unrolled) tests, iterations can run in parallel. Here is some proof:

The same author also published Fake SFTP Server Lambda, which is independent of the test framework in contrast to the JUnit 4 rule you use.

If you want to stick with the old tool, Spock 1.3 can also use JUnit 4 rules, and in Spock 2.x it might also work with the JUnit 4 compatibility layer.

Update: Here is an example program using the SSHJ library for downloading a file from an SFTP server, so we have a subject under test:

nasch Are you sharing actual code/error snippets Few observations

1. Caused by: org.springframework.beans.factory.NoSuchBeanDefinitionException: No qualifying bean of type 'com.package.repository.ChannelBalanceAdjustmentRepository' how you can have a package name containing package.. its a reserverd keyword and will not be allowed.
2. Your repository which is in a different package and without public keyword how its accessible in the service package.

You are encountering a side-effect of Spock's magic.

Your method basically gets transformed to this:

Your tests start an app that listens on a real port. And you use TestRestTemplate to make HTTP calls. It's the same as if you ran your test from a remote machine - would you expect @Transactional on such tests to be some how applied to the app?

@Transactional will work only if you invoke your endpoint directly, without any network calls:

• either inject endpoint object directly to your test and call its method
• or use MockMvc (or RestAssured+MockMvc) - it will also eventually call the endpoint directly

Both of these options will simplify debugging - you'll be able to see in call stack which test is calling your production code at the moment.

PS: also it shouldn't be a problem when your data is kept between test runs. You can isolate your tests with randomization.

I finally managed to find a solution after some more frantic googling. Although, this is not as clean as I would have wished, it works!

The only way I've found it possible to manipulate the request and response objects in a controller is by calling RequestContextHolder.setRequestAttributes() with a new GrailsWebRequest. The downside to this is that the response object also has to be overwritten. This is not a big problem however, as it is manipulated "in-place" when calling render(), so we can just check the "would be" response status on our newly created object. My Spock test now looks like this:

After some research I have stumbled over the following setting that needs to be applied to the application-test.properties or application-test.yml file.

The thing that breaks you code, is that mock interactions are the first thing that is evaluated in every then block, even if you list other things first. So for it to work, you need to use a second then block as explained in the chapter invocation order in the docs. Since, the mocks are evaluated first, the execution doesn't even reach result.get() that would wait long enough so that the save interaction could be triggered.

This should fix your problem: