without this change we can't have a contract that will generate a stub but not a test. Of course that's for a reason since it's overriding the very essence of contract test.
with this change we assume that the users know what they're doing and that they will use this flag rarerly. Cause it means that you can have a false positive.
related to gh-881
* Generate stubs at runtime
without this change the producer side must publish the stubs for the consumer side to use them
with this change the consumer can toggle a switch so that the stubs get generated at runtime. This might of course lead to false positivies but we assume that the users know what they're doing
fixes gh-881
without this change if one does $(regex("[0-9]")) we have no knowledge of whether the result should be text or a number. What we do ATM is we always generate a String
with this change once can pass the type of regular expression and we will generate the concrete value of that given type
fixes gh-768
* Introduces explicit mode of stub downloading
you have to explicitly provide the mode [CLASSPATH, REMOTE, LOCAL] of how you want to fetch and register stubs
fixes gh-287
Never got to the bottom of this, but the test failure with the
dropped connection (EOF on socket) seems to go away if we use
the Apache client instead of the JDK.
The problem is that it is a Spring Boot app (really tiny one), so
it binds to server.* in the environment, if you try to use
the stock ServerProperties. We only need a subset of the features
so it's actually relatively easy to wrap it and use a delegate.
* Added options to pass props via @AutoConfigureStubRunner
it's much easier to pass props via the annotation instead of property files. With this change the user can provide the properties inside the test via the annotation. The only thing that has to be passed via props is repositoryRoot (typically it's a very constant property that you set once).
fixes#46
