When I vented my feelings on Jeff Atwood’s naive view of commercial software development, I did not have any statistical data. For example, in his post Jeff talked about how MS Test sucks vs. NUnit and how nobody is using it.
And now, thanks to the poll that Roy Osherove did, there are some data at least to argue that claim. In his original post, he asked what what unit test framework people use, what mocking framework and what threading objects are most frequently used.
Now that the votes are all counted, everyone can see that MS Test beats everything but NUnit and was used by 27% of those responded (NUnit was 47%). Of course, you can discard the poll as not representative enough, but Roy's readers are pretty advanced bunch.
The data in the poll surely undermines Jeff’s claim that “Everybody I know who has tried to use the MSTest flavor of unit tests has eventually thrown up their arms and gone back to NUnit”. I have used both NUnit and MSTest, and while the latter may lack some features for advanced users, in general it is very decent framework with decent integration into Visual Studio.
What I really wanted to say is that it pays to experience things first-hand; if you evaluate something, do not be swayed by popular opinion without evaluating software and do make your very own opinion.
And another data in Roy’s poll (somewhat unrelated to the topic of the post) is the relative usage of threading primitives. Who could have thought that in year 2008 the close second is Thread.Sleep? Underlines the importance of code reviews, that :)