Yarc is a C++-based Linux & Windows client for Redis. It directly supports...
- RESP, RESP3 (attributes, streaming, maps, etc.)
- Pipelining
- Pub-Sub
- Cluster
- Transactions
- Connection Pooling
Other Redis features would be exposed indirectly through the raw-use of the Redis server communication protocol.
Here is the hello-world of the Yarc library...
#include <yarc_simple_client.h>
#include <yarc_protocol_data.h>
#include <yarc_byte_stream.h>
using namespace Yarc;
int main()
{
// This connects to 127.0.0.1 on port 6379 by default.
auto client = Yarc::SimpleClient::Create();
client->MakeRequestAsync(ProtocolData::ParseCommand("SET greeting \"Hello, world!\""));
ProtocolData* responseData = nullptr;
if (client->MakeRequestSync(ProtocolData::ParseCommand("GET greeting"), responeData))
{
BlobStringData* blobStringData = Cast<BlobStringData>(responseData);
if (blobStringData)
std::cout << "greeting = " << blobStringData->GetValue() << std::endl;
Yarc::ProtocolData::Destroy(responseData); // Be sure to free it in the proper heap!
}
Yarc::SimpleClient::Destroy(client); // Again, free in proper heap.
return 0;
}Redis Cluster support is obtained by simply replacing SimpleClient with ClusterClient, and then replacing #include <yarc_simple_client.h> with #include <yarc_cluster_client.h>. It has not been well tested! I wouldn't trust it.
The simple client is fairly stable, and I have a few production use-cases for it. As for the cluster client, the tester application thrashes a mini local cluster of 6 nodes (3 masters, 3 slaves, 1 slave per master) all while performing live resharding of the hash slots. This is as far as the cluster client has been tested, but that was a long time ago, and I'm not sure if it still works. In any case, I would revisit and revamp the code before I ever tried to use it for some application.
One obvious downside to the API shown above is that it's using an old callback style for asynchronous processing. A more modern API would use promises or the asyc/await keywords.