What C++ Needs to be Safe
summary
Bloomberg shows a concrete roadmap: macrobased contracts in production since 2004 (open sourced 2012), funded clang prototypes and library work to land contracts by C++26 and finer control by C++29, and prototypes for ghost data and runtime exclusivity. Measured impact: simple contracts and bounds checks catch ~65% of bugs, contract violations ~20%, erroneous-behavior fixes ~10%, lifetimes/data races ~2% each. Work includes PMR allocators, relocation and reflection; systems run in production and catch real defects while keeping hot paths efficient.
AI-written summary · about
chapters (11)
- 0:00C++ safety roadmap intro
- 1:13C++ safety criticism
- 3:01Safety strategy and goals
- 7:19Idea to production
- 13:01Defining C++ safety
- 20:27Contract assertions and ghost code
- 27:12Runtime vs compile-time
- 31:07Ghost data benefits
- 33:38Safety priorities
- 37:52Bloomberg’s safety priorities
- 40:25Summary and call to action
related talks
01h 02mTowards Correctness, Safety and Security in C++26
01h 02mTowards Safety and Security in C++26
01h 29mC++ Safety And Security Panel 2024
36mTowards Safety & Security in C++26
01h 03mLifetime Safety in C++: Past, Present and Future
01h 26mA Path to Practically Safe C++
01h 04mContracts for C++
01h 26mVulnerable C++