Re: SuiteAttribute, RowTest and IterativeTest
"Charlie Poole" <[email protected]>
| Newsgroups | gmane.comp.windows.dotnet.nunit.devel |
|---|---|
| Message-ID | <00bf01c883a9$fc476c30$6501a8c0@ferrari> |
Hi Kelly, > > SuiteAttribute is a hold-over from an older style of framework. > > NUnit 1.x got the idea from the original JUnit. NUnit 2.x > > discouraged its use, since we added automatic suite > creation based on > > the namespace hierarchy. > > > > Some people use it to create arbitrary combinations of tests that > > can be loaded from the command line using the /fixture option. In > > fact, this was the original and only purpose of the > /fixture option - > > we changed it so it would accept normal test fixtures and then > > namespaces. > > > > Mostly, it's of historical interest but it did establish > the pattern > > that we only call static methods at test load time. > > So, would it possibly be deprecated in 3.0? Could be. > As for the pattern that we only call static methods at test > load time, I think that might have to change as we go towards > a full Theory implementation. I guess that remains to be seen - bear in mind that I'm only talking about user fixtures here. Although they shouldn't, users often write test fixtures that create strange side effects in their constructors - it's really weird when a large machine in the corner starts up simply because you looked at the structure of the tests. I saw that happen in one shop, back when there was an error in NUnit causing the instance to be created too early. So, if we are going to create any user-instances at load time, I would rather they be some other object than the test fixture class. Charlie > -Kelly > ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/