Verification and Validation
๐
Software verification and validation ๐
Verification
: Check that each development stage is performed correctly.- Are we building the product right?
- Mainly performed through inspections and reviews.
Validation
: Check that each deliverable satisfies the requirements specified in the previous stage.- Are we building the right product?
- Mainly performed through testing.
Inspection ๐
Finding problems by analyzing a static representation of a system: Static verification.
The system does not need to run.
- Inspection is possible before implementation.
Applicable to any system representation:
- Requirements, designs, code, documents, and so on.
Program testing ๐
Finding problems by executing software: Dynamic verification.
Checks whether the program behaves as intended.
Finds defects before the program is used.
Inspection and testing ๐
Inspection
- Requirement specifications.
- Software architecture.
- UML design models.
- Database schemas.
- Programs.
Testing
- System prototypes.
- Programs.
Inspection and testing complement each other.
Both should be used in V&V.
Inspection
cannot establish
:- Whether the system meets customers' real needs.
- Nonfunctional qualities such as performance and usability.
Stages of testing ๐
Development testing
- Test during development to find bugs and defects.
- Developers perform unit and integration tests.
Release testing
- A separate testing team tests a complete version before users receive it.
- Test the full system to verify that all integrated features work correctly.
User testing
- Actual or potential users test the system in their own environments.
- Tests use real user scenarios.
Development testing ๐
Unit testing
- Test individual program units or object classes.
- Focus on the functionality of objects and methods.
Component testing
- Combine individual units to test a component.
- Focus on component interfaces.
System testing
- Combine some or all system components.
- Focus on interactions between components.
Black-box vs. white-box testing ๐
Black-box testing
- Derive test cases from an external view of the object under test.
- Choose valid and invalid inputs and determine the correct outputs.
- Requires no knowledge of internal structure.
White-box testing
- Derive test cases from program structure.
- Use knowledge of the program to identify additional cases.
Equivalence partitioning ๐
Divide the input domain into subsets, or partitions, and test representative values from each.
This reduces the test space while helping discover defects efficiently.
Equivalence-partitioning example ๐
- Partition 1: 0 and 5, invalid values.
- Partition 2: 6 and 10, valid values.
- Partition 3: 11 and 14, invalid values.
Create test cases using
one value
from each partition.Search routine specification ๐
Procedure:
search(Key : ELEM; T : SEQ of ELEM; Found : in out BOOLEAN; L : in out ELEM_INDEX).
Precondition
- The sequence contains at least one element.
- T`FIRST < = L < = T`LAST.
Postcondition
- If Found = TRUE, T(L) = Key.
- If Found = FALSE, Key does not exist in T.
