Lightning Talk: How we Grade C++ Assignments
summary
Course uses an automated grader that runs about five to seven tests per submission under a blackbox model and penalizes failures by start-with-full-score deductions rather than rewarding passes. Memory and runtime checks use file grind output fed back to students. Human review still inspects every solution and flags common failures: wrong index types, uninitialized variables, misuse of exceptions and partial functions, and creative but incorrect hash and power computations that pass tests but are unsafe.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:35Data Structures Complexity
- 1:39The Challenge
- 2:20The Problem
- 3:40Some Students Dont Understand
- 5:05Running the Test
- 6:30Inspecting the Code
- 7:20Fragments
- 8:00Hash Values
- 8:40Power
- 9:00Big Prime
- 9:20Lower Case to Upper Case
- 9:30Power of Two
- 9:45Partial Functions
related talks
58mWhat Makes Good C++ Programmers: A Continuous Search For C++ Teaching Recipes
01h 26mHow I Teach Modern C++ One Pixel at a Time
03mLightning Talk: C++ Catastrophes: A Poem
57mRefactoring Superpowers: Make Your C++ IDE Do Your Work, Faster and More Safely
56mWill Your C++ Program Still Be Correct Next Year?
01h 01mBack to Basics: Test-driven Development
43mEverything I learned about static analysis and program safety in C++
55mC++ Core and Other C++ Guidelines - The Good, the Bad, the… Questionable?