Why is code review alone insufficient to validate transformed code?
Select an answer to reveal the explanation.
Short Explanation
You can read intent and structure, but you only see the real trouble when the code runs. Rounding, data types, and edge cases can change values without looking wrong on the page.
Full Explanation
The correct answer is the option saying a reviewer checks intent and structure, but behavioural differences in arithmetic, rounding, data types and edge cases appear only when the code is executed against real inputs. COBOL and Java differ in numeric representation and default behaviours, and those differences surface as wrong values at particular inputs rather than as visibly wrong code. The option saying generated Java is too long is incorrect because volume is a practical difficulty, not the reason review cannot establish equivalence. The option saying reviewers cannot access the original COBOL source is incorrect because reviewers do work with it. The option saying review comments cannot be recorded against generated files is incorrect because tooling supports review of generated files. For the exam, treat review as complementary to executed tests, never as a substitute.