Software Architecture Lessons from Multidimensional Spaces Explorations
summary
Proposes that std::mdspan breaks into element type, extent, layout and accessor and that std::mdspan is a promising start but currently omits iterators and risks limiting Genericity Performance Expressivity. Demonstrates using C++20 features — generic NTTPs, concepts, generic lambdas in unevaluated contexts — to build symbolic EDSLs and template classes-as-mini-compilers, claims meta‑programming achieves near zero‑overhead and advises bottom‑up examples for genericity and top‑down APIs for expressivity.
AI-written summary · about
chapters (14)
- 0:00Start
- 1:04Multi-Dimensional Arrays
- 3:26Multidimensional Arrays
- 6:35The Difference between Good Code and Bad Code
- 18:40Haskell
- 20:39Handle Geneticity
- 21:01Design Abstractions
- 21:40Bottom-Up Approach
- 23:15Bottom Up Approach
- 33:42Design Limits
- 36:26Meta Programming
- 37:39Expressivity
- 42:41Symbolic Calculus
- 54:09Gpu Principle
related talks
01h 32mHow C++23 changes the way we write code
49mApplied Modern C++: beyond type erasure and visitor pattern
06mLightning Talk: Cute Approach for Polymorphism in C++
26mInfinite Possibilities
41mEnough TMP to write a Variant
57mDesign Patterns - The Most Common Misconceptions (2 of N)
01h 05mThe Imperatives Must Go!
49mEnough TMP to write a Variant