The evolving search for effective C++ - Keynote
summary
Argument: emplace vs insert is not universally faster — empirical tests across three libraries (libc++, libstdc++, VS2013) show behavior varies by value category and reallocation, so Howard Hinn's three-implementation experiment invalidates a blanket rule. For unique-key containers (quuck (unique-key) containers) emplace often allocates and constructs a node before duplicate checks, making emplace worse for duplicates. Also, emplace uses direct initialization and permits explicit constructors. Practical tooling notes: compilers, warnings, and clang-format matter; recorded talks reach thousands (e.g. 114k, 11k, 6k views) so design presentations for recordings.
AI-written summary · about
chapters (11)
- 0:00Berlin memories
- 2:48Effective C++ project history
- 14:28Writing effective technical books
- 18:26Case study on emplace vs insert
- 40:58Emplace performance and pitfalls
- 49:04Safety concerns with initialization
- 59:15Modern technical writing techniques
- 1:10:58Using color and formatting
- 1:18:20Effective presentation design
- 1:28:11Audience Q&A
- 1:35:29Q&A: C++ and writing tools
related talks
01h 24mKeynote: Optimizing for Change
01h 51mProg C++
01h 01mLearning (and teaching) modern C++ - Challenges and Resources
40mC++ for C Developers
01h 01mDown the Rabbit Hole: An Exploration of Stack Overflow Questions
01h 08mKEYNOTE - Emotional Code
58mThere is no silver bullet
01h 03mFrom Introductory to Advanced C++ - Learning Guidelines