Although test type typically corresponds to run time, this is not always the case. Multithreaded unit tests often run long, but are unit tests nonetheless. Some integration (component) tests run as quick as unit tests. Maximum benefit is achieved by running tests as often as is practical, regardless of their type.
If the goal is to group tests by their run times, perhaps one should do just that, i.e, group tests by fast/slow instead of unit/component/system.
Re: test categorization2006-11-08T03:38:59ZThis is the accepted answer. This is the accepted answer.You raise a good point; although, I tend to group multithreaded testing, in general, as a separate category all together due to its inherent challenges (not withstanding the trials of using JUnit in these scenarios).
I've found more often then not that the run time of a test strongly correlates to the three categories outlined in the referenced article; however, if you find short/long more effective then go for it! A specific taxonomy doesn't really matter-- what matters is the strategy placed around defining and ultimately running test groups.
Re: test categorization2006-11-09T03:31:04ZThis is the accepted answer. This is the accepted answer.
- SystemAdmin 110000D4XK