SiddiqSoft.asynchrony 1.7.3

asynchrony : Add asynchrony to your apps

Motivation

• We needed to add asynchrony to our code.
• The code here is a set of helpers that utilize the underlying deque, semaphore, mutex features found in std.
• Be instructive while providing functional code
• Use only C++20 standard code: jthread, deque, semaphore, barriers and latch
• No external dependency

Usage

Requires C++20 support!

Specifically we require jthread and stop_token support which is unfortunately not available in the Apple Clang 16 version. This library works with gcc 14+ or MSVC 17+ or Clang 18+.

Refer to the documentation for details.

The library uses concepts to ensure the type T meets move construct requirements.

Single threaded worker

#include "siddiqsoft/simple_worker.hpp"

// Define your data
struct MyWork
{
   std::string urlDestination{};
   std::string data{};
   void operator()(){
      magic_post_to(urlDestination, data);
   }
};

void main()
{
   // Declare worker with our data type and the driver function.
   siddiqsoft::simple_worker<MyWork> worker{[](auto& item){
                                              // call the item's operator()
                                              // to invoke actual work.
                                              item();
                                           }};
   // Fire 100 items
   for( int i=0; i < 100; i++ )
   {
      // Queues into the single worker
      worker.queue({std::format("https://localhost:443/test?iter={}",i),
                    "hello-world"});
   }

   // As the user, you must control the lifetime of the worker
   // Trying to delete the worker will cause it to stop
   // and abandon any items in the internal deque.
   std::this_thread::sleep_for(1s);
}

Multi-threaded worker pool

#include "siddiqsoft/simple_pool.hpp"

void main()
{
   // Declare worker with our data type and the driver function.
   siddiqsoft::simple_pool<MyWork> worker{[](auto& item){
                                           // call the item's operator()
                                           // to invoke actual work.
                                           item();
                                        }};
   // Fire 100 items
   for( int i=0; i < 100; i++ )
   {
      // Queues into the single queue but multiple worker threads
      // (defaults to CPU thread cout)
      worker.queue({std::format("https://localhost:443/test?iter={}",i),
                    "hello-world"});
   }

   // As the user, you must control the lifetime of the worker
   // Trying to delete the worker will cause it to stop
   // and abandon any items in the internal deque.
   std::this_thread::sleep_for(1s);
}

Multi-threaded roundrobin pool

#include "siddiqsoft/roundrobin_pool.hpp"

void main()
{
   // Declare worker with our data type and the driver function.
   siddiqsoft::roundrobin_pool<MyWork> worker{[](auto& item){
                                               // call the item's operator()
                                               // to invoke actual work.
                                               item();
                                             }};
   // Fire 100 items
   for( int i=0; i < 100; i++ )
   {
      // Queues into the thread pools individual queue by round-robin
      // across the threads with simple counter.
      // (defaults to CPU thread cout)
      worker.queue({std::format("https://localhost:443/test?iter={}",i),
                    "hello-world"});
   }

   // As the user, you must control the lifetime of the worker
   // Trying to delete the worker will cause it to stop
   // and abandon any items in the internal deque.
   std::this_thread::sleep_for(1s);
}

Implementation note

In order to use std::jthread on Clang 18 and Clang 19, we enable the compiler flag "CMAKE_CXX_FLAGS": "-fexperimental-library" in the CMakeLists.txt. This option will show up in your client library under Clang compilers.

<p align="right"> © 2021 Siddiq Software LLC. All rights reserved. </p>