This just in:
- It fixes several bugs
- It is able to run under .net 2.0 beta 1
If you were wondering, yes, I've been raising my suggestions about the rollback and database restore attributes in the Nunit-Developers mailing list and they have been received quite well - that is - people were interested, but Nunit version 3 will probably go on a different route than the one products like MbUnit is going in. Mainly - it will be kept more “simple” but will allow a large set of extensibility points, to which people can add Nunit Extensions. the Nunit Extension project will be a separate project to which community members will contribute (I'll add that all the info I know right now is based only on some emails and might change) and will be distributed along with the NUnit release (sorta like NAnt and NAnt-contrib projects). If that would be the case than that's good news. It should be pretty easy to develop such extensions (though I don't know exactly how that's going to work yet, there are several ways being considered).
Anyway - I think soon the Unit Testing Framework market is going to get a little crowded. I think its going to become a little like the IM market - best features + simplicity win, but the tools that can provide support for all frameworks (see Trillian) will also start to pop out. Problems might arise:
- Once you start with a framework on your project - how hard will it be to replace it (and would you want to do such a thing?).
- How complicated will those frameworks be?
- Unit testing will start to be part of formal training for people, since this is starting to be main stream (see Visual Studio 2005 Team System - with unit testing abilities).
- How will those frameworks match against the built in support in VS 2005? will they be competitors or will they be welcomed with open arms by the extensibility team?
Answers for these burning questions and more, on the next episode of “Roy Osherove - Testing Philosopher”