Template Metaprogramming: Type Traits Part 1 CppCon 2020 1 Jody Hagins
 jhagins@maystreet.com 
 coachhagins@gmail.comTemplate Metaprogramming: Type Traits CppCon 2020 2 IntroductionIntended Audience oriented: shallow depth, slow current • Not necessarily beginner to C++, but beginner to traditional template metaprogramming techniques 3Intended Audience • Beginner/Intermediate • Gentle entry: swimming oriented: shallow depth, slow current • Not necessarily beginner to C++, but beginner to traditional template metaprogramming techniques • Type traits part of standard library for ~10 years 3Intended Audience

1 / 58Template Metaprogramming (TMP) Template Metaprogramming (TMP) template using meta_fun = this_talk; 2 / 58❓static_assert(Template::Metaprogramming::is_hard); // ✔ 3 / 58❓s 58❓static_assert(Template::Metaprogramming::is_hard); // ✔ ❓static_assert(Template::Metaprogramming::is_powerful); // ✔ 3 / 58❓static_assert(Template::Metaprogramming::is_hard); // ✔ ❓static_assert( ❓static_assert(Template::Metaprogramming::is_powerful); // ✔ ❓static_assert(( Template::Metaprogramming::is_easy and Template::Metaprogramming::is_powerful and Template::Metaprogramming::has_nice_error_messages

What are the responsibilities of a “linear algebra library”? • Layer -1: Fundamentals • Multidimensional arrays & iteration • Layer 0: Performance primitives • Vector: dot, norm, vector sum, plane update • Layer 1: Low-level math problems • Linear systems 𝐴𝑥 = 𝑏 (& determinants, etc.) • Least-squares problems min 𝑥 𝐴𝑥 − 𝑏 2 • Eigenvalue & singular value problems 𝐴𝑥 = 𝜆𝑥 • Layer 2: Higher-level What are the responsibilities of a “linear algebra library”? • Layer -1: Fundamentals • Multidimensional arrays & iteration • Layer 0: Performance primitives • Vector: dot, norm, vector sum, plane

asynchronous operations enables the layered construction of operations whose power increases with each layer Templates & named type requirements decouple composed asynchronous operations from concrete types std::net::steady_timer Possibilities: EFVI-based multicast socket Database access layer Printer spool 7async_event template struct basic_async_event { using executor_type auto get_executor() const noexcept; std::size_t notify_one(); std::size_t notify_all(); template decltype(auto) async_wait(CompletionToken&& token); }; using async_event

(?) FUNCTORS A functor is a parameterized type (a class template) of one (free) argument. A functor is a parameterized type (a class template) of one (free) argument. Some examples of functors: Some function is The characteristic function is fmap fmap.. 8fmap fmap IN C++ IN C++ template template typename Functor, typename T> constexpr auto fmap(F f, Functor A monad is A monad is also also a parameterized type (a class template) of one (free) argument. a parameterized type (a class template) of one (free) argument. Some examples of monads: Some examples

Overview of ROS 2What is ROS 2? ● has a middleware layer that allows for message passing between different processesWhat is ROS 2? ● has a middleware layer that allows for message passing between different service server, act as an action client, or act as an action serverWhat is ROS 2? ● has a middleware layer that allows for message passing between different processes ● systems are made up of nodes that can provide configurable parameters which can be adjusted at run-timeWhat is ROS 2? ● has a middleware layer that allows for message passing between different processes ● systems are made up of nodes that can

get the trades – and to be accurate! 11 Ultra low-latency execution (< 1μs) Low-latency strategy layer (1-10 μs)Order book deep dive “The order book is the heart of any trading system.” Larry Harris Volume, std::greater> mBidLevels; std::map> mAskLevels; template typename T::iterator AddOrder(T& levels, Price price, Volume volume) { auto [it, inserted] Volume, std::greater> mBidLevels; std::map> mAskLevels; template typename T::iterator AddOrder(T& levels, Price price, Volume volume) { auto [it, inserted]

alignment cohesionMORE BEHAVIOURS MORE BEHAVIOURS We can arbitrarily layer more behaviours into the calculation. We can arbitrarily layer more behaviours into the calculation. Obstacle avoidance, collision relies on the composable properties of This relies on the composable properties of bool bool.. template struct FallbackAllocator : Primary, Fallback { auto allocate(size_t turn a binary pattern into an n-ary pattern: We can turn a binary pattern into an n-ary pattern: template struct FallbackAllocator : As... { auto allocate(size_t n) -> Blk { Blk r{};

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 14.6 Template system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . of the recipe revision and package revision (RREV and PREV). Creating a basic package Create a template package: $ conan new /@[/] # / is not specified␣ for a basic test␣ ˓→to verify the package was created successfully [-s] # Create a recipe/source template␣ ˓→for a package with local source code Build a package from a recipe and store it in the local

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 14.6 Template system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . of the recipe revision and package revision (RREV and PREV). Creating a basic package Create a template package: $ conan new /@[/] # / is not specified␣ for a basic test␣ ˓→to verify the package was created successfully [-s] # Create a recipe/source template␣ ˓→for a package with local source code Build a package from a recipe and store it in the local