Comments on: Introducing Scurvy.Test

By: Björn

Björn — Wed, 30 Jul 2008 14:34:10 +0000

RE firstly: I am not sure that this is a real requirement/limitation of any framework and not only a convention of unit testing.

RE second: makes sense. The way I pondered it was to use the refernece rasterizer and dump the bits into an image and compare that one with a reference image. This would cause issues when some device settings (e.g. MSAA and such) were changed, though, and it might suboptimal for validating animations.

RE lastly: ah, the good old NIH

By: Joel Martinez

Joel Martinez — Wed, 30 Jul 2008 13:20:53 +0000

Good questions, it boils down to three things. Firstly, those unit test frameworks (and all others as far as I know) are tied to being run in the context of an external test runner. I wanted to bring that in-process so that you could take advantage of things like rendering to the graphics device. Second, I also wanted to enable manual testing scenarios where QA testers can visually/manually verify certain things that you can’t verify in code, “does this model *look* right”. And lastly, I kinda wanted to tackle the project on my own. If I would have tried to take the time to learn the internals of those other frameworks and figure out how to work my features in, I can almost guarantee this would have never been released

By: Trevor Longino

Trevor Longino — Wed, 30 Jul 2008 02:27:18 +0000

You trying to run me out of a job?

…not that I work as a tester anymore. But you get the idea.

Seems to be a good idea, although I *do* wonder about Bjorn’s question.

By: Björn

Björn — Mon, 28 Jul 2008 14:32:31 +0000

While I have not taken a look at the code yet (will do soonish ) I cannot stop but wonder why you didn’t extend an exisiting Unit test framework like xUnit.net or NUnit.