Client IPC Interface

[NikVolf]

IPC interface for Client

from example test to run client:

    let remote_client = nanoipc::init_client::<RemoteClient<_>>(socket_path).unwrap();
    assert!(remote_client.handshake().is_ok());

from example test to run worker:

    let mut worker = nanoipc::Worker::new(&client);
    worker.add_reqrep(&socket_path).unwrap();
    while !stop.load(Ordering::Relaxed) {
        worker.poll();
    }

Progress tracker

• limiting errors to crate-level error types Client API refactoring - limiting errors to crate-level error types #1525
• void rpc methods missing IPC codegen enhancement - allow void methods #1540
• interface tweaks mapping
• some docs missing
• pretty name configuration change IPC codegen to allow attributes #1500
• simple rpc test

[NikVolf]

@arkpar .rs.in -> .rs

[NikVolf]

@gavofyork merged with master

[rphmeier]

still says there are merge conflicts

[arkpar]

File containing most of the clients code is called client_ipc.rs?
We do still want to have no-ipc build, right?
Client code should be ipc-independent ideally

[NikVolf]

@rphmeier
because i have to merge conflicts every time someone else commits to the client.rs :)
merged again

@arkpar
everything is completely independent, if you create just Client, you work without any ipc
if you create RemoteClient, you work via ipc

[gavofyork]

i have to agree with @arkpar - why is Client class in a file called client_ipc.rs? unless there's good reasons, filenames should reflect the main class in them.

[NikVolf]

@arkpar @gavofyork
right, didn't got the concerns right
actually extra file can be avoided completely
and diff if much cleaner now

[gavofyork]

indentation messed up.

[gavofyork]

looks good apart from the errant indent.

[NikVolf]

@gavofyork fixed