How can we differentiate System Test Cases from the Functional Test Cases?

Replies to This Discussion

Permalink Reply by Ram on August 4, 2009 at 11:47am

Shruti,
couple follow up questions before I directly answer your questions...
1. what do you clearly mean by "execute all test cases" and "develop a set of system test cases"?
2. "save my time" is a relative - compared to what? are you saying you rather execute a subset of all athe available test cases and there by saving your time as oppose to executing all the available test cases?

to answer as per the way I understood your questions:
Understand the scope of the application and review the test cases that you mentioned as already available. I would also highly encourage the use of Heuristics to approach the regression. Depending on the system configurability and the functionality, it would also be ideal to prepare a chart or a matrix of scenarios that helps you judge the application scope for coverage. It is like a saying of "How do you eat an elephant? ans: one bite at a time".
With the matrix you have, try to map those related test cases and see if you still get the coverage (as you define). Again, coverage is relative and contextual too. Matrix may not tell you everything, but at least guides you through the flow and permutations and combinations.

Now, I know there will be number follow up questions on what I have just said and can be discussed further.

-Ram

▶ Reply to This

Permalink Reply by Shruti on August 4, 2009 at 12:03pm

Hi Ram...

Thanks for your Reply.
And Sorry for the Confusion...

1) Exceute all test cases means - Test cases that are already developed (exists with me)to test the modulewise functionality while regression testing. and Develope a set of system test cases means - With the help of existing test cases i need to diffrentiate/(convert) Functional and System Test case that i can show my manager that how can i reduce the number of test cases to cover all scenarios for the system testing.
2) By doing above thing, i would like to save my time for regession testing. for eg: If previously i need 20-25 days for regression testing, then with the help of the System Test cases i can do it within 10 -15 days etc.

I hope this clears your and may be others doubts too...

▶ Reply to This

Permalink Reply by Ram on August 4, 2009 at 12:24pm

Thank you, Shruthi.
I often run into this discussion with my coworkers around identififying the subset for the regression. My question to them, always is "How did you decide on that subset?" Some would argue and give a straight forward answer that they have picked all the priority 1 test cases. Well, the follow up question is "how did you derive at the priority 1 test cases?"

We came up with a process guidelines to define "priority 1" - again this is by context. Every application is different, every scenario is different and the judgement is primarily based on the guidelines we have defined in our group to categorize that way.

That doesn't mean all priority 1 scenarios are always the right candidate for regression, 'coz it depends on the code being changed and the areas directly impacted and not so directly impacted and least impacted. I learned this from a an interview candidate while interviewing for a tester position. I asked him as about coming up with the right set of test scenarios for a given feature and explain why and how he came up with those scenarios. The answer was with an analogy, and here it goes:
" Assume a feature set under test is like a dead body lying in a park waiting for investigation. What investigation crew does to find the possible clues is mark a matrix around the body and to some distance around it. that some distance comes from the experience and certain guidelines. And this what we say the priority in our case. Divide the Matrix into number of cells to find the clues. You may find a clue right by the body or within 2 feet of radius or 10 feet of radius, but gives an indication and guidelines to what to look for "

This story was appealing to me, in a way it made sense to me. So while building the regression suite,
a. see what functionality is critical for application to perform and
b. application to serve the purpose for customers.
c. What sub features can wait even in case of remote defect and
d. what sub features are vulnerable

note again: some of these may or may not directly applies to every unique situation or applications. They are contextual and but the principles / guidelines can be adopted as and when and where it is applicable.

▶ Reply to This

Permalink Reply by Maksim Grinevich on August 4, 2009 at 2:05pm

As I understand you just need to create some kind of "Smoke regression tests" document. Which consists test cases selected by some criteria.
What criteria to be used nobody can exactly say to you, 'coz it depends on.... (Ram already wrote about it)

So select list of criteria with your boss and go ahed.

You always can say to your boss:
You know why we pass just selected test cases and it takes us for X days, if you want to pass all test cases it takes us for 2*X (or Y) days.

▶ Reply to This

Permalink Reply by Yeshwanthrao Tharkeswar on August 5, 2009 at 6:09pm

System testing is nothing but End to End test which covers all the functionality of the application and performance (which includes load,stress,volume and performance).

A few organisation conducts performance test seperately but undertake a few important (call us benchmark test) performance test to make the system test pass.

▶ Reply to This

Permalink Reply by Yeshwanthrao Tharkeswar on August 5, 2009 at 6:20pm

Shruti,

The Subject line and the detail line doesn't match?
-----------------------------------------------------------------------------------------------------------------------------
Subject Line: How can we differentiate System Test Cases from the Functional Test Cases?
Detail Line:
HI everyone,

I am a manual tester, while doing the Regression Testing for my application, i need to execute all test cases developed for the application testing. but now, i want to develop a set of system test cases which will help me in the regression testing and by which i can save my time. For that, i think, i need to modify some test cases and their sequence that would help me in regression testing to test the application as a whole System. Can somebody help me or give some idea for the same?
--------------------------------------------------------------------------------------------------------

Based on the other replies, I have given my answer.

What to Test?
Start with large areas of coverage
Drill down into more focused areas within each of those large areas.
Followed by subsequent levels of detail until get down to specific testable units.

Why to Test?
What risks are associated with the most used features?
What specific constraints are concerning, how the supporting platform or code base changed over time, or how the use changed?
Which features are most critical to the business or the purpose of the application?
What has to be tested because of legal requirements or SLAs?
What will happen if it doesn't work?
Where's the technical risk and how has it changed over time?
What have been told to test?

Put all activity together meaningfully.
Prioritise the work
Evaluate what can realistically cover given the time, tools and resources
Review and agreement on coverage.
Schedule

Note: Need to be reviewed periodically.

▶ Reply to This