Se

1. CPM / PERT / Project Scheduling (Guaranteed Numericals)

You need to calculate ES, EF, LS, LF, and Slack. The Trick: Two passes. Forward pass for Earliest times, Backward pass for Latest times.

• Earliest Start (ES): Max of the Earliest Finish (EF) of all immediate predecessors. (Start at 0).

• Earliest Finish (EF): $ES + \text{Duration}$.

• Minimum Time (MT) / Critical Path: The max EF at the very end of the network.

• Latest Finish (LF): Min of the Latest Start (LS) of all immediate successors. (Start at the end using MT).

• Latest Start (LS): $LF - \text{Duration}$.

• Slack Time: $LS - ES$ (or $LF - EF$). If Slack is 0, that activity is on the Critical Path.

2. COCOMO & Cost Estimation

Always identify the software class first:

• Organic: Small team, familiar environment.

• Semi-detached: Mixed experience.

• Embedded: Complex hardware/real-time constraints.

The Formulas:

• $Effort = a1 \times (KLOC)^{a2}$ (in Person-Months)

• $Tdev = b1 \times (Effort)^{b2}$ (in Months)

The EAF Trick (Intermediate COCOMO): If they give you Cost Drivers (like Pool 1 vs Pool 2 developers), multiply the base Effort by the Effort Adjustment Factor (EAF).

• $Effort_{new} = Effort_{nominal} \times EAF$.

• Note: Time ($Tdev$) is calculated using the new Effort.

3. Putnam Model & Schedule Compression

• The Trick: Putnam models effort over time using a Rayleigh curve.

• The Core Equation: $K = L^{3} / ({C_{k}}^{3} \times {t_{d}}^{4})$. (Where $K$ is effort, $L$ is size, $t_{d}$ is time, $C_{k}$ is environment constant) .

• Exam Phrase to Drop: “Extreme penalty for schedule compression.” Because $t_{d}$ is to the power of 4 in the denominator, reducing the schedule slightly causes effort and cost to explode exponentially.

4. Control Flow Graph (CFG) & Cyclomatic Complexity

If they ask you to estimate independent paths (like in the Binary Search PYQ):

• The Formula: $V(G) = E - N + 2$ (Edges - Nodes + 2).
• The Quick Trick: Count the number of predicates (condition nodes like if, while, for) and add 1.
• Example: A while loop containing an if and an else if = 3 conditions. Complexity = 4. This is your upper bound for independent paths.

5. Coupling and Cohesion

• Coupling (Between modules - Want LOW):

• Best to Worst: Data $\rightarrow$ Stamp $\rightarrow$ Control $\rightarrow$ External $\rightarrow$ Common $\rightarrow$ Content.

• Keywords: Data (passing only simple parameters) , Stamp (passing whole data structures) , Control (passing flags to change logic) , Common (shared global variables).

• Cohesion (Inside module - Want HIGH):

• Best to Worst: Functional $\rightarrow$ Sequential $\rightarrow$ Communicational $\rightarrow$ Procedural $\rightarrow$ Temporal $\rightarrow$ Logical $\rightarrow$ Coincidental.

• Keywords: Functional (one single task) , Sequential (output of A is input to B) , Temporal (tasks executed at the same time, e.g., initialization).

6. Testing Terminology

• Verification vs. Validation: Verification is phase-containment (did we build the phase right?). Validation checks the final product against the SRS (did we build the right product?).

• Why Branch > Statement Coverage:

• Proof: Consider an if (x > 0) statement without an else. Executing with x = 1 gives 100% statement coverage. However, it ignores what happens if x = -1. Branch coverage forces you to test both the True and the implicit False branches, catching errors Statement coverage misses.

• Black Box: Equivalence Partitioning (grouping inputs into valid/invalid sets) and Boundary Value Analysis (testing the exact edges of those sets, e.g., if range is 1-5000, test 0, 1, 5000, 5001).

7. Software Maintenance

Memorize these three buckets:

• Corrective: Fixing actual bugs and failures.
• Adaptive: Porting software to a new OS, database, or hardware environment.
• Perfective: Adding new features or improving performance based on user requests.