v1.71 and 1.80 SetInstance breaks object creation about automoq

[darrencauthon]

Thank you for the failing test. I've started a pull request to resolve it. I'll post an update as soon as I have a fix.

[darrencauthon]

@kszepieto The issue seems to be calling Create twice, so I believe that's something I'll be able to resolve.

I haven't seen this in any test because I've never come across that... calling Create<T> twice. In my thousands of tests written with this tool, I've always had only one call to that method per test. But it should work! This test of yours will be great to prevent the regression.

If you are stuck in a bad spot right now, you can downgrade to 1.7.1 or use Moq to create the stub object. I do this occassionally:

var foo = new Mock<IFoo>().Object;

[darrencauthon]

@kszepieto I'm still investigating.

It is a matter of calling Resolve twice during a test -- not the SetInstance method. If you use Resolve only to register the class under test, you should be fine.

[darrencauthon]

I believe I resolved the issue here:

e7d3977

That fixed the new failing tests. I'm still getting an error around abstract dependency resolution, which I'm still in the process of resolving.

The ultimate fail-safe is to remove the refactorings I made to isolate the "mocking" and "ioc" work. But I really don't want to give them up quite yet -- they're essential to the next place I want to take AutoMoq. Still working...

[darrencauthon]

The fix for this has been published as 1.8.1.

Thanks @kszepieto for the catch! It's strange, in all of the years of using this library I've never called Create<T> twice in one test! Your unit tests are part of the test suite to prevent regressions.