#What is CO

CO is an elegant and efficient C++ base library that supports Linux, Windows and Mac platforms. It implements a series of high-quality base components, such as go-style coroutine, coroutine-based network programming framework, command line parameter and config file parsing library, high-performance log library, unit testing framework, JSON library, etc.

CO is an open source project on github under the MIT license. It contains code from some other projects, which may have different licenses, see LICENSE file for details.

#History of CO

Alvin(idealvin) has started to develop CO since 2013. The original purpose is to reduce third-party dependencies in C++ projects and improve the efficiency of C++ development. Beginning in 2015, Alvin introduced CO into industrial projects for use by himself and his colleagues, which greatly reduced the development cycle of C++ projects.

After years of accumulation and precipitation, by 2019, Alvin implemented the coroutine mechanism in golang with C++ and provided a set of network programming framework based on coroutine. At the beginning of its birth, it was used in the development of embedded network programming and achieved immediate results.

As of 2021, CO coroutines have made considerable progress. At present, hooks are supported on Linux/Windows/Mac platforms, and coroutine lock, coroutine synchronization event, coroutine pool, and channel and waitgroup in golang have been implemented. Users can use CO to write C++ like golang.

#Quick Start


It is recommended to install xmake and run the following command in the root directory of CO to build all sub-projects:

xmake -a

If you need to use http::Client, SSL or HTTPS features, you can use the following command to build:

xmake f --with_libcurl=true --with_openssl=true
xmake -a

Xmake will automatically install libcurl and openssl from the network. Depending on the network, this process may be slow. xmake -a will build libco, gen, co/unitest and co/test. Users can run test programs in CO with the following commands:

xmake r unitest -a
xmake r flag
xmake r log -cout
xmake r co

#Develop C++ projects with CO

The simplest, you can directly include co/all.h and use all the features in CO. If you are worried about the compiling speed, you can also include only the header files that you need, such as including co/co.h, you can use co/flag, co/log and all features related to coroutines.

#include "co/all.h"

DEF_string(s, "nice", "");

int main(int argc, char** argv) {
    flag::init(argc, argv);

    LOG << FLG_s;
    return 0;

The above is a simple example. The first two lines of the main function are used to initialize the flag and log libraries. Some components in CO use co/flag to define config items and use co/log to print logs. Therefore, it is generally necessary to call flag::init() and log::init() at the beginning of the main function for initialization.

Users can also use the macro DEF_main to define the main function:

#include "co/all.h"

DEF_string(s, "nice", "");

DEF_main(argc, argv) {
    LOG << FLG_s;
    return 0;

DEF_main has already called flag::init() and log::init() internally, and users do not need to call them again. In addition, DEF_main makes code in the main function also run in coroutine, which is consistent with golang. Some coroutine-related components in CO must be used in coroutine. When CO is used to develop coroutine-based applications, it is recommended to define the main function with DEF_main.

#Core features


co/flag is a simple and easy-to-use command line and config file parsing library. Some components in CO use it to define config items.

co/flag provides a default value for each config item. Without config parameters, the program can run with the default config. Users can also pass in config parameters from command line or config file. When a config file is needed, users can run ./exe -mkconf to generate a config file.

#include "co/flag.h"
#include "co/log.h"

DEF_bool(x, false, "bool x");
DEF_bool(y, false, "bool y");
DEF_uint32(u32, 0, "...");
DEF_string(s, "hello world", "string");

int main(int argc, char** argv) {
    flag::init(argc, argv);

    COUT << "x: "<< FLG_x;
    COUT << "y: "<< FLG_y;
    COUT << "u32: "<< FLG_u32;
    COUT << FLG_s << "|" << FLG_s.size();

    return 0;

The above is an example of using co/flag. The macro at the beginning of DEF_ in the code defines 4 config items. Each config item is equivalent to a global variable. The variable name is FLG_ plus the config name. After the above code is compiled, it can be run as follows:

./xx                  # Run with default configs
./xx -xy -s good      # single letter named bool flags, can be set to true together
./xx -s "I'm ok"      # string with spaces
./xx -u32 8k          # Integers can have units: k,m,g,t,p, not case sensitive

./xx -mkconf          # Automatically generate a config file: xx.conf
./xx xx.conf          # run with a config file
./xx -config xx.conf  # Same as above


co/log is a high-performance local log library, some components in CO use it to print logs.

co/log divides the log into five levels: debug, info, warning, error, and fatal. Printing a fatal level log will terminate the program. Users can print logs of different levels as follows:

DLOG << "hello "<< 23; // debug
LOG << "hello "<< 23;  // info
WLOG << "hello "<< 23; // warning
ELOG << "hello "<< 23; // error
FLOG << "hello "<< 23; // fatal

co/log also provides a series of CHECK macros, which can be regarded as an enhanced version of assert, and they will not be cleared in debug mode.

void* p = malloc(32);
CHECK(p != NULL) << "malloc failed..";
CHECK_NE(p, NULL) << "malloc failed..";

When the CHECK assertion fails, co/log will print the function call stack information, and then terminate the program.

co/log is very fast. The following are some test results, for reference only:

  • co/log vs glog (single thread)

    platform google glog co/log
    win2012 HHD 1.6MB/s 180MB/s
    win10 SSD 3.7MB/s 560MB/s
    mac SSD 17MB/s 450MB/s
    linux SSD 54MB/s 1023MB/s
  • co/log vs spdlog (Windows)

    threads total logs co/log time(seconds) spdlog time(seconds)
    1 1000000 0.103619 0.482525
    2 1000000 0.202246 0.565262
    4 1000000 0.330694 0.722709
    8 1000000 0.386760 1.322471
  • co/log vs spdlog (Linux)

    threads total logs co/log time(seconds) spdlog time(seconds)
    1 1000000 0.096445 2.006087
    2 1000000 0.142160 3.276006
    4 1000000 0.181407 4.339714
    8 1000000 0.303968 4.700860


co/unitest is a simple and easy-to-use unit test framework. Many components in CO use it to write unit test code, which guarantees the stability of CO.

#include "co/unitest.h"
#include "co/os.h"

namespace test {
DEF_test(os) {
    DEF_case(homedir) {
        EXPECT_NE(os::homedir(), "");

    DEF_case(cpunum) {
        EXPECT_GT(os::cpunum(), 0);
} // namespace test

The above is a simple example. The DEF_test macro defines a test unit, which is actually a function (a method in a class). The DEF_case macro defines test cases, and each test case is actually a code block. Multiple test units can be put in the same C++ project, the main function is simple as below:

#include "co/unitest.h"

int main(int argc, char** argv) {
    flag::init(argc, argv);
    return 0;

co/unitest contains the unit test code in CO. Users can run unitest with the following commands:

xmake r unitest -a   # Run all test cases
xmake r unitest -os  # Run test cases in the os unit


CO has implemented a go-style coroutine, which has the following features:

  • Multi-thread scheduling, the default number of threads is the number of system CPU cores.
  • Shared stack, coroutines in the same thread share several stacks (the default size is 1MB), and the memory usage is low. Simple test on Linux shows that 10 millions of coroutines only take 2.8G of memory (for reference only).
  • There is a flat relationship between coroutines, and new coroutines can be created from anywhere (including in coroutines).
  • Support system API hook (Windows/Linux/Mac), you can directly use third-party network library in coroutine.
  • Coroutineized socket API.
  • Coroutine synchronization event co::Event.
  • Coroutine lock co::Mutex.
  • Coroutine pool co::Pool.
  • channel co::Chan.
  • waitgroup co::WaitGroup.

#Create a coroutine

go(ku);           // void ku();
go(f, 7);         // void f(int);
go(&T::f, &o);    // void T::f(); T o;
go(&T::f, &o, 7); // void T::f(int); T o;
    LOG << "hello go";

The above is an example of creating coroutines with go(). go() is a function that accepts 1 to 3 parameters. The first parameter f is any callable object, as long as f(), (*f)(), f(p), (*f)(p), (o->*f)() or (o->*f)(p) can be called.

The coroutines created by go() will be evenly distributed to different scheduling threads. If users want some coroutines to run in the same thread, they can create coroutines in the following way:

auto s = co::next_scheduler();

If users want to create coroutine in all scheduling threads, the following way can be used:

auto& s = co::all_schedulers();
for (size_t i = 0; i <s.size(); ++i) {


co::Chan, similar to the channel in golang, can be used to transfer data between coroutines.

#include "co/co.h"

DEF_main(argc, argv) {
    co::Chan<int> ch;
    go([ch]() {
        ch << 7;

    int v = 0;
    ch >> v;
    LOG << "v: "<< v;

    return 0;

The read and write operations of channel must be performed in coroutine, so the main function is defined with DEF_main in above code, so that code in the main function also runs in coroutine.

The channel object in the code is on the stack, while CO uses a shared stack implementation. Data on the stack of one coroutine may be overwritten by other coroutines. Generally, we cann’t share data on stack between coroutines. Therefore, we capture by value in the lambda to copy the channel and pass it to the newly created coroutine. The copy operation just increases the internal reference count by 1 and will not affect the performance.

When creating a channel, we can add a timeout as follows:

co::Chan<int> ch(8, 1000);

After read or write operation, we can call co::timeout() to determine whether it has timed out. This method is simpler than the select-based implementation in golang.

The channel in CO is implemented based on memory copy. The data type passed can be a built-in type, a pointer type, or a structure type with simple memory copy semantics for the copy operation. Like std::string or container types in STL, the copy operation is not a simple memory copy, and generally cannot be passed directly in channel. For details, see Document of co::Chan.


co::WaitGroup, similar to sync.WaitGroup in golang, can be used to wait for the exit of coroutines or threads.

#include "co/co.h"

DEF_main(argc, argv) {
    FLG_cout = true;

    co::WaitGroup wg;

    for (int i = 0; i <8; ++i) {
        go([wg]() {
            LOG << "co: "<< co::coroutine_id();

    return 0;

#network programming

CO provides a set of coroutineized socket APIs, most of them are consistent with the native socket APIs in form. In addition, CO has also implemented higher-level network programming components, including TCP, HTTP and RPC framework based on JSON, they are IPv6-compatible and support SSL at the same time, which is more convenient than socket APIs. Here is just a brief demonstration of the usage of HTTP, and the rest can be viewed in the documents.

Static web server

#include "co/flag.h"
#include "co/log.h"
#include "co/so.h"

DEF_string(d, ".", "root dir"); // Specify the root directory of the web server

int main(int argc, char** argv) {
    flag::init(argc, argv);

    so::easy(FLG_d.c_str()); // mum never have to worry again

    return 0;

HTTP server

http::Server serv;

    [](const http::Req& req, http::Res& res) {
        if (req.is_method_get()) {
            if (req.url() == "/hello") {
                res.set_body("hello world");
            } else {
        } else {
            res.set_status(405); // method not allowed

serv.start("", 80);                                    // http
serv.start("", 443, "privkey.pem", "certificate.pem"); // https

HTTP client

void f() {
    http::Client c("");

    LOG << "response code: "<< c.response_code();
    LOG << "body size: "<< c.body_size();
    LOG << "Content-Length: "<< c.header("Content-Length");
    LOG << c.header();"/hello", "data xxx");
    LOG << "response code: "<< c.response_code();