C++ coroutines: The mental model for coroutine promises

by Raymond Chen

My earlier series on getting started with awaitable objects looked at how you produce awaitable objects, but up until now we haven’t been looking at the other side: How can you be an object that can await other objects?

Creating a task completion source for a C++ coroutine: Failing to produce a result

by Raymond Chen

So far, we’ve been working on building a result_holder that can hold any type of result. But what about errors?

Because maybe you have code that’s waiting for a result, and the code that’s supposed to produce the result realizes that it messed up and wants to say, “Sorry, no result today.”...

Creating a task completion source for a C++ coroutine: Producing nothing

by Raymond Chen

Last time, we created a result_holder that can hold a reference, and we solved it by using a wrapper. But there’s another type that we can’t put in a result_holder, not even with the help of a wrapper. That type is void...

Creating a task completion source for a C++ coroutine: Producing a result with references

by Raymond Chen

Last time, we created a result_holder that can be awaited until it is assigned a result, and I noted that the code was broken.

One of the reasons that it’s broken is that it doesn’t handle references properly...

GotW #101 Solution: Preconditions, Part 2

by Herb Sutter

We covered some basics of preconditions in GotW #100. This time, let’s see how we can use preconditions in some practical examples…

Creating a task completion source for a C++ coroutine: Producing a result

by Raymond Chen

We’ve been looking at creating different types of awaitable synchronization objects. This time, we’ll create something analogous to what C# calls a Task­Completion­Source and what PPL calls a task_completion_event. For lack of a better name, I’m going to call it a result_holder...

Ticket Maps

by Anthony Williams

It has been an increasingly common scenario that I've encountered where you have some ID that's monotonically increasing, such as a subscription or connection index, or user ID, and you need your C++ program to hold some data that's associated with that ID value. The program can then pass round the ID, and use that ID to access the associated data at a later point...

Creating other types of synchronization objects that can be used with co_await, part 10

by Raymond Chen

We’ve been looking at creating different types of awaitable synchronization objects. This time, we’re going to create something that doesn’t exist in the normal Win32 repertoire: An event where you can wait for the event to be in a desired state, either set or reset. Normal Win32 events allow you to wait for them to be set, but you cannot wait for Win32 event to be reset. The usual workaround is to have two events...

An Infinite Data Stream with Coroutines in C++20

by Rainer Grimm

In this post, I analyze the new keyword co_yield. Thanks to co_yield, you can create an infinite data stream in C++20...

constexpr Dynamic Memory Allocation, C++20

by Bartlomiej Filipek

constexpr has become a major feature for compile-time programming in C++. Introduced in a simple form in C++11 evolved into almost another “sub-language”, an alternative to regular template code. In C++20 you can even use std::vector and std::string in constexpr context!

In this article, I’d like to discuss constexpr memory allocations, a building block for std::vector. Let’s see how this feature works and see its limitations...