6 Banks & Financial Institutions — Assistant Programmer — Exam: 18.03.2021প্রশ্ন ও সমাধান

Correct answer: (b) Random

An array keeps its elements in one contiguous block of memory, and every element has the same size. The address of element i is therefore computed in one step: address = base address + i × element size. Any element can be read or written in O(1) time without visiting the elements before it. This is called random (direct) access.

Why the other options are wrong

• (a) Sequential: this is how a linked list or a tape file is accessed, where you must pass through the earlier items first. An array can be scanned in order, but that is not its defining access method.
• (c) Rational: not an access method at all; the word belongs to numbers (rational numbers), not to data access.
• (d) Stochastic: means governed by probability; it is not a term used for accessing data in a structure.

Tip: Array → random access by index, O(1). Linked list → sequential access, O(n).

Source: Arrays (C++) | Microsoft Learn

Correct answer: (c) "X extends Y" is correct if X and Y are either both classes or both interfaces

The JLS fixes what may follow extends:

• In a class declaration, the extends clause must name a class type (§8.1.4). Interfaces go in the implements clause (§8.1.5).
• In an interface declaration, every type in the extends clause must be an interface type (§9.1.3), and an interface may extend several interfaces.

So class X extends Y (Y a class) and interface X extends Y (Y an interface) are the only legal pairings: both classes or both interfaces. Other rules can still block a particular case (for example, Y cannot be a final class), but they do not change which kinds may be paired.

Why the other options are wrong

• (a) …if and only if X is a class and Y is an interface: A class cannot extend an interface; it must use class X implements Y.
• (b) …if and only if X is an interface and Y is a class: An interface can never extend a class; it can extend only interfaces.
• (d) …for all combinations of X and Y being classes and/or interfaces: The mixed combinations in (a) and (b) are compile errors.

Tip: extends stays within one kind (class→class, interface→interface); implements crosses from class to interface.

Source: Chapter 8. Classes, §8.1.4 Superclasses and Subclasses - Java Language Specification (Java SE 8); Chapter 8. Classes, §8.1.5 Superinterfaces - Java Language Specification (Java SE 8); Chapter 9. Interfaces, §9.1.3 Superinterfaces and Subinterfaces - Java Language Specification (Java SE 8)

Correct answer: (c) go()

The java.lang.Thread API includes start(), run(), sleep(long millis), join(), yield(), interrupt(), isAlive(), setPriority() and others, but no go() method. A thread is launched with start(), which makes the JVM call run() on the new thread.

Why the other options are wrong

• (a) sleep(long msec): A real static method, Thread.sleep(long millis). It pauses the current thread for the given number of milliseconds (the parameter name in the option does not matter).
• (b) stop(): A real method, though deprecated because stopping a thread this way is inherently unsafe (it releases all monitors and can leave shared objects inconsistent). Deprecated still means it exists in the class.
• (d) yield(): A real static method that hints to the scheduler that the current thread is willing to give up the processor.

Tip: Thread life cycle: start() → run(); pause with sleep(), wait for another thread with join(), give way with yield(). There is no go().

Source: Thread (Java Platform SE 8) - Oracle API documentation

Correct answer: (c) Merge sort

An in-place sort keeps the sorted items in the same array that held the input, holding at most a constant number of items outside it at any moment. Standard merge sort breaks this rule: to merge two sorted halves it writes the result into a separate buffer of up to n elements and then copies it back, so it needs Θ(n) extra memory.

Why the other options are wrong

• (a) Insertion sort: shifts larger elements one place to the right inside the array, holding just the current key in a temporary variable, so O(1) extra space.
• (b) Selection sort: finds the minimum of the unsorted part and swaps it into position inside the array, so O(1) extra space.
• (d) Heap sort: the heap is built inside the input array itself, and each step only swaps elements within it, so O(1) extra space.

Tip: In-place: insertion, selection, bubble, heap (and quick sort, apart from its recursion stack). Not in-place: merge, counting, radix.

Source: in-place sort — Dictionary of Algorithms and Data Structures (NIST)
Source: Lecture 5: Linear Sorting, MIT 6.006 Introduction to Algorithms (Spring 2020), MIT OpenCourseWare
Source: heapsort — Dictionary of Algorithms and Data Structures (NIST)

Correct answer: (c) Simple Mail Transfer Protocol

SMTP is the Simple Mail Transfer Protocol, defined today in RFC 5321. It is the push protocol of email: a mail client uses it to hand an outgoing message to its mail server, and mail servers use it to relay the message to the recipient's server. Server-to-server SMTP runs on TCP port 25.

Why the other options are wrong

• (a) Single Mail Text Protocol: "Single" and "Text" are wrong; the words are Simple and Transfer.
• (b) Single Mail Transfer Problem: "Single" is wrong, and "Problem" is not even the word "Protocol".
• (d) Simple Mail Textual Protocol: "Textual" replaces "Transfer"; SMTP is about transferring mail, not about text formatting.

Tip: SMTP sends mail (Simple Mail Transfer), POP3/IMAP fetch it.

Sources: RFC 5321: Simple Mail Transfer Protocol; Service Name and Transport Protocol Port Number Registry, search "smtp" - IANA

Correct answer: (c) 9

x > ANY (subquery) is true when x is greater than at least one value the subquery returns.

1. The subquery returns all 10 salaries, and they are all different (UNIQUE, NOT NULL).
2. Every salary except the lowest is greater than at least the lowest salary, so the test is true for those 9 rows.
3. The lowest salary is not greater than any value (only equal to itself), so that one row fails.
4. COUNT(*) = 10 − 1 = 9.

SQLite does not support ANY, so the logic was run with an equivalent form on 10 unique salaries:

SELECT COUNT(*) FROM Employee
WHERE SALARY > (SELECT MIN(SALARY) FROM Employee);   -- returns 9

Why the other options are wrong

• (a) 0: That is the result of > ALL, where no salary can beat its own value.
• (b) 5: Nothing in the query compares with a middle value; only the single minimum is excluded.
• (d) 10: The lowest salary has nothing smaller to beat, so all 10 cannot qualify.

Tip: > ANY → all rows except the minimum (n − 1 with unique values); > ALL → 0 rows when the list includes every value.

Source: PostgreSQL Documentation: Subquery Expressions

Correct answer: (b) A logical product of N with 0Fh

0Fh = 0000 1111: upper 4 bits 0, lower 4 bits 1. A logical product is a bitwise AND, and bit by bit x AND 1 = x, x AND 0 = 0. So N AND 0Fh copies the lower 4 bits of N unchanged and clears the upper 4 bits.

Example with N = A5h = 1010 0101:

• AND 0Fh → 0000 0101 (lower 4 bits 0101, unchanged)
• XOR 0Fh → 1010 1010 (lower bits inverted)
• NAND 0Fh → 1111 1010 (lower bits inverted)
• OR 0Fh → 1010 1111 (lower bits forced to 1)

A Python check over all 256 values of N confirms that only the AND leaves the lower 4 bits unchanged every time.

Why the other options are wrong

• (a) Exclusive logical sum (XOR) with 0Fh: x XOR 1 = x′, so every one of the lower 4 bits is toggled.
• (c) Negative logical product (NAND) with 0Fh: NOT(x AND 1) = x′, so the lower 4 bits are inverted (and the upper 4 become 1). Even if read as AND with the complement F0h, the lower 4 bits are cleared to 0000.
• (d) Logical sum (OR) with 0Fh: x OR 1 = 1, so the lower 4 bits are all forced to 1.

Exam note: The options print "0Fn"; this is a typo for 0Fh (h = hexadecimal), corrected here. The terms used are logical product = AND, logical sum = OR, exclusive logical sum = XOR and negative logical product = NAND.

Tip: AND with 1 keeps a bit and AND with 0 clears it; OR with 1 sets it; XOR with 1 toggles it.

Source: Microsoft Learn: C Bitwise Operators

Correct answer: (d) {(False, False), (True, True)}

Decision/condition coverage needs two things at once:

1. the whole decision (A OR B) must come out True at least once and False at least once (so both the X = X+1 and the X = X−1 branches run);
2. each condition, A and B, must itself take both True and False.

Check option (d):

• (False, False) → A OR B = False → else branch, X = X−1
• (True, True) → A OR B = True → if branch, X = X+1
• A takes F and T; B takes F and T

Both requirements are met with only 2 test cases.

Why the other options are wrong

• (a) {(False, True)}: one test; the decision is only True, A is only False and B only True.
• (b) {(False, True), (True, False), (True, True)}: every case gives A OR B = True, so the else branch never runs.
• (c) {(False, True), (True, False)}: A and B each take both values, but the decision is True both times; this satisfies condition coverage only, not decision coverage.

Tip: For A OR B, the only way to get a False decision is (F, F); any branch-coverage set must include it.

Exam note: The stem equates decision/condition coverage with branch coverage. Strictly, branch (decision) coverage needs only the two decision outcomes, while decision/condition coverage also needs each condition both ways; option (d) satisfies both. Option (b) was printed with a stray "*" before True, which has been removed.

Source: A Practical Tutorial on Modified Condition/Decision Coverage (NASA/TM-2001-210876); ISTQB Certified Tester Foundation Level Syllabus v4.0.1