Well I am writing tests for my html5 parser. So the script I have is generating Smalltalk unit test code of all possible combinations a set of html elements. I am using basic data mining to change the expected output. To make it clear. The tests I have the elements html head body and p opening and closing tags. Any output with no p in the input would generate this tree: Html Head Body I added a little bit of "data mining" to check for any p elements in the input and if that is true the output would be Html Head Body P The reason I have many failing tests is because many tests have more than 1 p element and I want to file those out and change the expected output to have 2 p elements ... Keep in mind I am still a student and have no experience in development so this might be the worst possible way to do things. I am also using Java code to generate the Smalltalk code in which I then file in. I hope i did not make it so complicated by explaining... -----Original Message----- From: Pharo-users [mailto:pharo-users-bounces@lists.pharo.org] On Behalf Of Stephan Eggermont Sent: Monday, August 26, 2013 4:11 PM To: pharo-users@lists.pharo.org Subject: Re: [Pharo-users] Categorizing tests according to test result Hi Mohammad, You can take a look at TestResult>>historyFor: Why are you generating unit tests as code with a script? And why do you have many failing tests? Stephan