Hi:
I would like to thank you for your diligent work at spider UCI of making the tacoco tool. I am currently using tacoco together with tarantula to assist an experiment studying fault localization and regression testing techniques.
Problem Description
In the compact-cov-matrix.json file generated by tacoco, there is a testStmtMatrix attribute for each source file covered by a test class's run. The testStmtMatrix is in the form of Boolean [test ][stmt ], indicating for a single test case which line in the source file has been covered. This information is important to Tarantula to output the suspiciousness score of a statement. However, the problem is when I look at the line with the highest suspiciousness in a buggy source file, those line numbers are not accurate. Currently, the approach I am taking to map an index i in the testStmtMatrix to the actual line in the source file is by adding i with the firstline recorded by tacoco. As a result, the line number with the highest suspicousness seems to be irrelevant with the failing test cases, for example, unlikely-covered methods or even brackets.
My Thought
I guess the problem could be that my experiment subject files are heavily commented, which makes it inaccurate for Tarantula to map a suspicousness score to the correct line in the source file. Therefore, I would like to ask that do you have any comments or thoughts on how to utilize tacoco correctly to assist Tarantula? I would like to ask for a new feature on mapping the line index in testStmtMatrix to the actual line in a source file.
I look forward to hearing from you. Thank you very much!
Regards,
Zhenming Wang
@VijayKrishna @kajdreef @tariq1890 @nish2292 @jycamus90
Hi: I would like to thank you for your diligent work at spider UCI of making the tacoco tool. I am currently using tacoco together with tarantula to assist an experiment studying fault localization and regression testing techniques.
Problem Description
In the compact-cov-matrix.json file generated by tacoco, there is a
testStmtMatrix
attribute for each source file covered by a test class's run. ThetestStmtMatrix
is in the form ofBoolean [test ][stmt ]
, indicating for a single test case which line in the source file has been covered. This information is important to Tarantula to output the suspiciousness score of a statement. However, the problem is when I look at the line with the highest suspiciousness in a buggy source file, those line numbers are not accurate. Currently, the approach I am taking to map an indexi
in thetestStmtMatrix
to the actual line in the source file is by addingi
with thefirstline
recorded by tacoco. As a result, the line number with the highest suspicousness seems to be irrelevant with the failing test cases, for example, unlikely-covered methods or even brackets.My Thought
I guess the problem could be that my experiment subject files are heavily commented, which makes it inaccurate for Tarantula to map a suspicousness score to the correct line in the source file. Therefore, I would like to ask that do you have any comments or thoughts on how to utilize tacoco correctly to assist Tarantula? I would like to ask for a new feature on mapping the line index in
testStmtMatrix
to the actual line in a source file.I look forward to hearing from you. Thank you very much!
Regards, Zhenming Wang @VijayKrishna @kajdreef @tariq1890 @nish2292 @jycamus90