Which way to use DaprClient is recommended?

Hi, there are 2 ways to use DaprClient in current demos.

first one is to create a new DaprClient instance whenever you need to talk to dapr runtime:

try (DaprClient client = (new DaprClientBuilder()).build()) {
 // ...
}

second one is to create a DaprClient bean before usage, then inject it into code:

@Autowired
private DaprClient client;

since our project is Spring-based, the second way seems attractive, however, is it thread-safe to use DaprClient in the second way, since Spring’ bean is singleton by default.

[artursouza]

The second way is preferred because you reuse the client. The client is thread-safe and designed to be reused.

[darren-wang]

ources that might not be freed by taking this approach, even after the JVM shut down (i.e. used ports)?

that’s what confusing me too. i guess the only way to find it out is to read the source codes.