Preexam 2 Discussion Exam 2 Thurs Nov 15

Pre-exam 2 Discussion Exam 2 Thurs, Nov 15 th 150 pts, 2 crib sheets allowed 841 f 07 exam 2 preview 23 nov 13 11

Parnas - question In Parnas' Paper, the table 1 - you said in the class we should be able to estimate "h“ but in the article "h" is given, so how we can estimate "h"? do you mean that you will give us "the failure probability", and we can calculate "h" based on "failure probability"? or there is another way to estimate "h"? Yes, you could calculate h based on likelihood, but I think I said that I would not ask you to calculate anything based on the number of tests. 841 f 07 exam 2 preview 23 nov 13 22

Van der Meulen What is the effectiveness of diversity in calculating failure probability E(Θ) and how does it influence in the calculation of E(Θ)? 841 f 07 exam 2 preview 23 nov 13 33

Van der Meulen Comparison of formulae’s for E(Θ) for Frankl’s and Van der meulen article? 841 f 07 exam 2 preview 23 nov 13 44

How does this match Frankl’s (1) 841 f 07 Meulen 16 oct 18 55

6

Van der Meulen 841 f 07 exam 2 preview 23 nov 13 77

Hamlet u Does Hamlet create the alternatives before or after seeing the input? 841 f 07 exam 2 preview 23 nov 13 8

Hamlet Can the term rewriting system be developed for any sorts of ADT specifications such as ADT specification of program in OCL, Alloy or some other formal languages? 841 f 07 exam 2 preview 23 nov 13 99

Hamlet How does the diversity effect E(Θ) calculation in Hamlet paper? 841 f 07 exam 2 preview 23 nov 13 10 10

Hamlet (page 187 second paragraph) explains that a conventional oracle is not needed to obtain the 10^-4 bound in the testing phase (first phase). He explains that in this phase he uses a BTRS to generate a test set of terms. If the starting point of these terms were 0, main should evaluate EVERY test term to 0 841 f 07 exam 2 preview 23 nov 13 11 11

Hamlet In (page 187 first paragraph), he says that if three randomly chosen terms agree, the probability if all of them are failures is less than 10^-12. Why is he always considering that all the bad case is when all of them are failures? i. e. page 198 second paragraph: "The effectiveness of TRM on TCASis measured as the probability that these ten randomly generated formulas will ALL take the same incorrect value. 841 f 07 exam 2 preview 23 nov 13 12 12

Hamlet Equation (2) on page 193. I'm unsure why he uses the value N/2 n 841 f 07 exam 2 preview 23 nov 13 13 13

Hamlet says that a typical practical reliability is 10^-4 but what is the typical confidence on this reliability? 841 f 07 exam 2 preview 23 nov 13 14 14

All Comparison between reliability estimate E(Θ) in Hamlet paper, Frankl and Van der meulen and what is the difference between them? 841 f 07 exam 2 preview 23 nov 13 15 15

Bertolino’s Does the bertolino’s dream of achieving 100% testing can be achieved using Van der meulen approach of generating equivalent terms for a program without using test oracles. Briefly explain? 841 f 07 exam 2 preview 23 nov 13 16 16

Bertolino’s Diagram Where does all the methods of reliability estimates by Frankl, Hamlet and Van der meulen lie in the Bertolino’s diagram and under what goal/challenge do they fall under? 841 f 07 exam 2 preview 23 nov 13 17 17

841 f 07 exam 2 preview 23 nov 13 18 18

Parnas’s article 30 pts u Reliability u Terms u Concepts u Evaluation 841 f 07 exam 2 preview 23 nov 13 19

Van Meulen’s article 40 pts u Terms u Approach/experiment u Diversity u Diagrams u Results/conclusions 841 f 07 exam 2 preview 23 nov 13 20

Hamlet’s article 50 pts u Terms u Approach u ADT, axioms u Diversity u Conclusions/results u Evaluation 841 f 07 exam 2 preview 23 nov 13 21

Bertolino’s article 30 pts u Terms u Practical, Effective u Discussion u Coonclusions u Diagrams 841 f 07 exam 2 preview 23 nov 13 22 22

Questions u What questions, if we discussed them in class would help you prepare for the exam? u Send by 5 pm Sunday, Nov 11 th 841 f 07 exam 2 preview 23 nov 13 23

Unresolved Issues 841 f 07 exam 2 preview 23 nov 13 24 24

Hamlet The original ADT program is released in a modified form that evaluates a number of equivalent terms for each original term in the program. Thus in execution, every term is checked. In the example of Figure 2, if main correctly implements the stack, every equivalent term generated by the equations of Figure 1 treated as a BTRS will evaluate to the same value. 841 f 07 exam 2 preview 23 nov 13 25 25

841 f 07 exam 2 preview 23 nov 13 26 26