Smalltalk Unit

CategoryTesting: RegressionTesting

SUnit, formerly known as TestingFramework is KentBeck's framework for automating ProgrammerTest s in SmalltalkLanguage programs.

The Smalltalk version of the TestingFramework can be downloaded from

http://sunit.sourceforge.net/

Version 3.1 (2003) supports

(With Release 2 the software is called SUnit.)

Original documentation is available at

http://www.xprogramming.com/testfram.htm Simple Smalltalk Testing: With Patterns KentBeck, and
Chapter 30 "Simple Smalltalk Testing", in Kent Beck's Guide to Better Smalltalk by KentBeck, Donald G. Firesmith (Editor), Cambridge University Press, December 1998, ISBN 9780521644372

Note that the TestSuite class>>named: method is not supported in 2.3 (and beyond?)

SqueakSmalltalk

SUnit is included in the main distributions of Squeak 3.9, Squeak 3.10 and Pharo 0.1 Source packages can be downloaded from SqueakMap for Squeak versions 3.8 and earlier.

I wrote a simple VA Smalltalk front end to the framework, following Kent's suggestions. I'll be happy to pass it on to anyone who'd like to use or expand on it. -JosephPelrine

See EnhancingSunit

Is there a good pattern for placing tests in ENVY applications? Should my application pre-req SUnitApp, or should I create an <appname>TestApp? that prereqs the real app and the SUnitApp (I think the latter). - SteveCline You sure want the latter as you don't want to pre-req SUnit in a runtime image- DanielEnting

It seems, version 2.6 doesn't work well with name spaces, at least in VisualWorks 5i.1b. It doesn't find classes in namespaces below Smalltalk. I have submitted a test case and fix to SamuelShuster?, the current maintainer. Meanwhile, if you need it, contact me. --HaskoHeinecke

Shouldn't it be named SmallUnit??

See also JeffreyOdell? 's SUnitBrowser at http://www.odellsoft.com/sunitbrowser/

I'm using the SUnit extensions for the RefactoringBrowser. I'm starting to really hate the poor design of the interface.

There's a Debug button and a Run button. If you hit the run button, it suppresses any walkbacks. If you hit run and you get an error or failure which you wish to inspect, hitting debug afterwards will run the test all over again. This isn't any fun when your one measly test takes upwards of 5 minutes to run.

So if I've only got one test selected, for what possible reason am I allowed to use the run button? I can't conceive of a legitimate reason why I would wish to do such a thing.

CategorySmalltalk