Is it bad form to count on the order of your NUnit unit tests -

I am making a unit test like crazy and have found that I have to set up some tests in a test that I Tearing up in the previous exam is it appropriate to make some (such as a database record) in a test (for example an entry test) and then use it for later testing (e.g., a deletion test)?

Can you also schedule the trials in NUIT or are they always alphabetically?

Note: I am specifically asking about

Update:

Update < / Strong> Thanks for everyone answering - a lot of things have a lot of good consensus and the consensus of the group I have chosen John Nolan's answer because he has given the most complete details and many links That you have guessed, despite this thinking of me to break this rule The temptation Nbirta, because John said it "smelly". Thanks to Fortyrunner for adding the unit-test tag.

The sequence of your tests indicates that you are making the state in all the tests

A cleaner way of testing, where you only rely on a piece of functionality that you want to see in your behavior. Normally you need to get your method to work in other tests for other items.

There is a good pattern to think about the coming of unit tests.

Below is a snippet of Carl Seguin's excellent free I have interpreted the systematic, act and fear. [TestFixture] Public square carteast [[Test] Public Zero to Vectorrolls Updateed Albexingkar () {// Systematic Mock Repository Socks = New Mok Repository (); IDataAccessDatax = mocks.CreateMock & lt; IDataAccess & gt; (); ObjectFactory.InjectStub (IDataAccess, Data Entry); // act car car = new car (); Expect.Call (dataAccess.Save (car)) Return (389). Mocks.ReplayAll (); Car.Save (); Mocks.VerifyAll (); // Assign fears.Arequal (38 9, car. ID); ObjectFactory.ResetDefaults (); }}