net-mesh 0.24.0

High-performance, schema-agnostic, backend-agnostic event bus
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484

//! End-to-end nRPC duplex integration tests against real
//! `MeshNode`s. Composes Phase B's client-streaming (request
//! direction) with the existing server-streaming response path
//! (response direction) — both run interleaved over one
//! `call_duplex` / `serve_rpc_duplex` pair.
//!
//! Coverage:
//! 1. Bidirectional echo — caller streams N requests, server
//!    emits one response per request + a final summary; caller
//!    observes interleaved responses while still sending.
//! 2. `finish_sending` keeps the response stream open — caller
//!    closes upload; server keeps emitting; caller drains until
//!    terminal End.
//! 3. Server terminates first; caller's subsequent send surfaces
//!    a clean error.
//! 4. `into_split` lets sink + stream live in independent tokio
//!    tasks; CANCEL only fires when BOTH halves drop.
//! 5. CANCEL from either side closes both halves cleanly.

#![cfg(all(feature = "net", feature = "cortex"))]

use std::net::SocketAddr;
use std::sync::atomic::{AtomicBool, AtomicUsize, Ordering};
use std::sync::Arc;
use std::time::Duration;

use bytes::Bytes;
use futures::StreamExt;
use net::adapter::net::cortex::{
    RequestStream, RpcDuplexHandler, RpcHandlerError, RpcResponseSink, RpcStreamingContext,
};
use net::adapter::net::mesh_rpc::{CallOptions, CodecDirection, RpcError};
use net::adapter::net::{EntityKeypair, MeshNode, MeshNodeConfig, SocketBufferConfig};

const TEST_BUFFER_SIZE: usize = 256 * 1024;
const PSK: [u8; 32] = [0x42u8; 32];

fn test_config() -> MeshNodeConfig {
    let addr: SocketAddr = "127.0.0.1:0".parse().unwrap();
    let mut cfg = MeshNodeConfig::new(addr, PSK)
        .with_heartbeat_interval(Duration::from_millis(200))
        .with_session_timeout(Duration::from_secs(5))
        .with_handshake(3, Duration::from_secs(2))
        .with_capability_gc_interval(Duration::from_millis(250));
    cfg.socket_buffers = SocketBufferConfig {
        send_buffer_size: TEST_BUFFER_SIZE,
        recv_buffer_size: TEST_BUFFER_SIZE,
    };
    cfg
}

async fn build_node() -> Arc<MeshNode> {
    let cfg = test_config();
    let keypair = EntityKeypair::generate();
    Arc::new(MeshNode::new(keypair, cfg).await.expect("MeshNode::new"))
}

async fn handshake_pair(a: &Arc<MeshNode>, b: &Arc<MeshNode>) {
    let a_id = a.node_id();
    let b_id = b.node_id();
    let b_pub = *b.public_key();
    let b_addr = b.local_addr();
    let b_clone = b.clone();
    let accept = tokio::spawn(async move { b_clone.accept(a_id).await });
    a.connect(b_addr, &b_pub, b_id)
        .await
        .expect("connect failed");
    accept
        .await
        .expect("accept task panicked")
        .expect("accept failed");
    a.start();
    b.start();
}

// ============================================================================
// Test handlers.
// ============================================================================

/// Echo handler: for every inbound request body, emits one
/// response body of the form `"echo:<body>"`. On EOS, emits a
/// summary chunk `"total:<count>"` and returns Ok.
struct EchoDuplexHandler;

#[async_trait::async_trait]
impl RpcDuplexHandler for EchoDuplexHandler {
    async fn call(
        &self,
        _ctx: RpcStreamingContext,
        mut requests: RequestStream,
        responses: RpcResponseSink,
    ) -> Result<(), RpcHandlerError> {
        let mut count: u64 = 0;
        while let Some(req) = requests.next().await {
            let mut body = b"echo:".to_vec();
            body.extend_from_slice(&req);
            responses.send(body);
            count += 1;
        }
        responses.send(format!("total:{count}").into_bytes());
        Ok(())
    }
}

/// Emits N responses immediately, returns Ok. Used to verify
/// that `finish_sending` doesn't cut off response delivery.
struct EmitNThenWaitForEosHandler {
    n: usize,
}

#[async_trait::async_trait]
impl RpcDuplexHandler for EmitNThenWaitForEosHandler {
    async fn call(
        &self,
        _ctx: RpcStreamingContext,
        mut requests: RequestStream,
        responses: RpcResponseSink,
    ) -> Result<(), RpcHandlerError> {
        for i in 0..self.n {
            responses.send(format!("pre-{i}").into_bytes());
        }
        // Wait for caller's EOS, then return.
        while requests.next().await.is_some() {}
        Ok(())
    }
}

/// Emits one response, then returns immediately — terminal
/// RESPONSE flies before the caller has finished its sends.
struct ServerTerminatesFirstHandler;

#[async_trait::async_trait]
impl RpcDuplexHandler for ServerTerminatesFirstHandler {
    async fn call(
        &self,
        _ctx: RpcStreamingContext,
        _requests: RequestStream,
        responses: RpcResponseSink,
    ) -> Result<(), RpcHandlerError> {
        responses.send(Bytes::from_static(b"only"));
        // Return immediately — fold emits terminal End; further
        // caller sends arrive at a closed-down call.
        Ok(())
    }
}

/// Loops forever observing cancellation. Used to verify the
/// CANCEL-from-caller path.
struct ForeverHandler {
    observed_cancel: Arc<AtomicBool>,
    consumed: Arc<AtomicUsize>,
}

#[async_trait::async_trait]
impl RpcDuplexHandler for ForeverHandler {
    async fn call(
        &self,
        ctx: RpcStreamingContext,
        mut requests: RequestStream,
        responses: RpcResponseSink,
    ) -> Result<(), RpcHandlerError> {
        loop {
            tokio::select! {
                _ = ctx.cancellation.cancelled() => {
                    self.observed_cancel.store(true, Ordering::SeqCst);
                    return Ok(());
                }
                maybe = requests.next() => {
                    match maybe {
                        Some(body) => {
                            self.consumed.fetch_add(1, Ordering::SeqCst);
                            responses.send(body);
                        }
                        None => {
                            // EOS without cancel — still wait for
                            // cancellation so the test can drive
                            // the assertion deterministically.
                            ctx.cancellation.cancelled().await;
                            self.observed_cancel.store(true, Ordering::SeqCst);
                            return Ok(());
                        }
                    }
                }
            }
        }
    }
}

// ============================================================================
// Tests.
// ============================================================================

/// 1/5 — Bidirectional echo. Caller streams 5 requests, server
/// emits one response per request + a final "total:5" summary
/// before EOS. Caller collects 6 chunks then sees clean EOF.
#[tokio::test]
async fn duplex_interleaves_send_and_recv() {
    let server = build_node().await;
    let caller = build_node().await;
    handshake_pair(&caller, &server).await;

    let _serve = server
        .serve_rpc_duplex("echo", Arc::new(EchoDuplexHandler))
        .expect("serve_rpc_duplex");

    let mut call = caller
        .call_duplex(server.node_id(), "echo", CallOptions::default())
        .await
        .expect("call_duplex");

    // Send 5 items, interleaved with response polling. The
    // server emits one Resp per Req, so by the time we finish
    // sending 5 we should have collected ~5 responses (plus the
    // final "total:5" summary after our EOS).
    for i in 0..5u8 {
        call.send(Bytes::copy_from_slice(&[b'a' + i]))
            .await
            .expect("send");
    }
    call.finish_sending().await.expect("finish_sending");

    let mut collected: Vec<String> = Vec::new();
    while let Some(item) = call.next().await {
        let chunk = item.expect("chunk must be Ok");
        collected.push(String::from_utf8(chunk.to_vec()).unwrap());
    }
    assert_eq!(collected.len(), 6, "5 echoes + 1 summary");
    for (i, label) in (0..5u8).zip(["a", "b", "c", "d", "e"]) {
        assert_eq!(collected[i as usize], format!("echo:{label}"));
    }
    assert_eq!(collected[5], "total:5");
}

/// 2/5 — `finish_sending` closes the upload but keeps the
/// response stream alive. Server emits N=3 responses BEFORE the
/// caller finishes sending; caller calls finish_sending without
/// any sends, then drains all 3 responses + sees clean EOF.
#[tokio::test]
async fn duplex_finish_sending_keeps_response_stream_open() {
    let server = build_node().await;
    let caller = build_node().await;
    handshake_pair(&caller, &server).await;

    let _serve = server
        .serve_rpc_duplex("emit_n", Arc::new(EmitNThenWaitForEosHandler { n: 3 }))
        .expect("serve_rpc_duplex");

    let mut call = caller
        .call_duplex(server.node_id(), "emit_n", CallOptions::default())
        .await
        .expect("call_duplex");
    // finish_sending immediately — degenerate "no upload" path.
    call.finish_sending().await.expect("finish_sending");

    let mut collected: Vec<String> = Vec::new();
    while let Some(item) = call.next().await {
        let chunk = item.expect("Ok chunk");
        collected.push(String::from_utf8(chunk.to_vec()).unwrap());
    }
    assert_eq!(collected, vec!["pre-0", "pre-1", "pre-2"]);
}

/// 3/5 — Server emits terminal RESPONSE before the caller is
/// done. Subsequent sends fail with a transport error (the
/// server's reply path is torn down once the terminal frame is
/// emitted; further REQUEST_CHUNK frames arrive at a closed
/// entry on the server side, are silently dropped, but the
/// caller's send / publish_to_peer either succeeds locally or
/// surfaces the transport-level failure). Either way the
/// response stream's terminal End is observed.
#[tokio::test]
async fn duplex_server_terminates_first() {
    let server = build_node().await;
    let caller = build_node().await;
    handshake_pair(&caller, &server).await;

    let _serve = server
        .serve_rpc_duplex("term_first", Arc::new(ServerTerminatesFirstHandler))
        .expect("serve_rpc_duplex");

    let mut call = caller
        .call_duplex(server.node_id(), "term_first", CallOptions::default())
        .await
        .expect("call_duplex");

    // Send one item so the initial REQUEST flies. The server
    // emits one response + terminates. Drain the response stream
    // until None.
    call.send(Bytes::from_static(b"hello"))
        .await
        .expect("send hello");
    let mut collected: Vec<Bytes> = Vec::new();
    while let Some(item) = call.next().await {
        match item {
            Ok(body) => collected.push(body),
            Err(_) => break,
        }
    }
    // Server emitted exactly "only" + terminal End.
    assert_eq!(collected.len(), 1);
    assert_eq!(&collected[0][..], b"only");
}

/// 4/5 — `into_split` lets the sink + stream live in separate
/// tokio tasks. The sink task encodes 5 requests then
/// finish_sending; the stream task collects 6 responses (5
/// echoes + 1 summary) then EOF. CANCEL only fires when BOTH
/// halves drop — here neither half drops early (both run to
/// clean completion), so the server-side cancellation token
/// stays unfired.
#[tokio::test]
async fn duplex_into_split_lets_halves_run_independently() {
    let server = build_node().await;
    let caller = build_node().await;
    handshake_pair(&caller, &server).await;

    let _serve = server
        .serve_rpc_duplex("echo_split", Arc::new(EchoDuplexHandler))
        .expect("serve_rpc_duplex");

    let call = caller
        .call_duplex(server.node_id(), "echo_split", CallOptions::default())
        .await
        .expect("call_duplex");
    let (mut sink, mut stream) = call.into_split();

    let sender = tokio::spawn(async move {
        for i in 0..5u8 {
            sink.send(Bytes::copy_from_slice(&[b'A' + i]))
                .await
                .expect("send");
        }
        sink.finish_sending().await.expect("finish_sending");
    });

    let receiver = tokio::spawn(async move {
        let mut count = 0;
        while let Some(item) = stream.next().await {
            std::hint::black_box(item.expect("Ok"));
            count += 1;
        }
        count
    });

    sender.await.expect("sender task");
    let count = receiver.await.expect("receiver task");
    assert_eq!(count, 6, "5 echoes + 1 summary");
}

/// 5/5 — CANCEL from the caller closes both halves. Caller
/// streams 2 chunks then drops the handle without
/// finish_sending. Server's ForeverHandler observes
/// `ctx.cancellation` firing and returns; terminal RESPONSE is
/// `Cancelled` (CANCEL-wins ordering in the fold).
#[tokio::test]
async fn duplex_cancel_from_caller_closes_both_halves() {
    let server = build_node().await;
    let caller = build_node().await;
    handshake_pair(&caller, &server).await;

    let observed = Arc::new(AtomicBool::new(false));
    let consumed = Arc::new(AtomicUsize::new(0));
    let _serve = server
        .serve_rpc_duplex(
            "forever",
            Arc::new(ForeverHandler {
                observed_cancel: observed.clone(),
                consumed: consumed.clone(),
            }),
        )
        .expect("serve_rpc_duplex");

    let mut call = caller
        .call_duplex(server.node_id(), "forever", CallOptions::default())
        .await
        .expect("call_duplex");
    call.send(Bytes::from_static(b"first"))
        .await
        .expect("send 1");
    call.send(Bytes::from_static(b"second"))
        .await
        .expect("send 2");
    drop(call);

    let deadline = std::time::Instant::now() + Duration::from_secs(3);
    while !observed.load(Ordering::SeqCst) && std::time::Instant::now() < deadline {
        tokio::time::sleep(Duration::from_millis(50)).await;
    }
    assert!(
        observed.load(Ordering::SeqCst),
        "handler must observe ctx.cancellation after caller drops the handle"
    );
}

/// `into_split` CANCEL semantics: dropping ONE half (sink OR
/// stream) must NOT fire CANCEL on the wire — the surviving half
/// keeps the call alive. The server only observes CANCEL after
/// BOTH halves have dropped without a clean close.
#[tokio::test]
async fn duplex_into_split_one_half_drop_does_not_cancel() {
    let server = build_node().await;
    let caller = build_node().await;
    handshake_pair(&caller, &server).await;

    let observed = Arc::new(AtomicBool::new(false));
    let consumed = Arc::new(AtomicUsize::new(0));
    let _serve = server
        .serve_rpc_duplex(
            "no_cancel_on_one_drop",
            Arc::new(ForeverHandler {
                observed_cancel: observed.clone(),
                consumed: consumed.clone(),
            }),
        )
        .expect("serve_rpc_duplex");

    let call = caller
        .call_duplex(
            server.node_id(),
            "no_cancel_on_one_drop",
            CallOptions::default(),
        )
        .await
        .expect("call_duplex");
    let (mut sink, stream) = call.into_split();
    // Publish the initial REQUEST so the server registers the
    // call — without this the server doesn't know about the call
    // and "no CANCEL observed" would be a vacuous pass.
    sink.send(Bytes::from_static(b"keepalive"))
        .await
        .expect("send");

    // Drop the sink half ONLY. The stream stays alive — Arc
    // refcount is still > 0 inside DuplexInner.
    drop(sink);

    // Wait long enough that, if CANCEL were going to fire, it
    // would have arrived by now (the server publishes
    // cancellation on the receive path very quickly).
    tokio::time::sleep(Duration::from_millis(200)).await;
    assert!(
        !observed.load(Ordering::SeqCst),
        "server must NOT observe CANCEL while the stream half is alive",
    );

    // Now drop the stream too — both halves gone → CANCEL fires.
    drop(stream);
    let deadline = std::time::Instant::now() + Duration::from_secs(3);
    while !observed.load(Ordering::SeqCst) && std::time::Instant::now() < deadline {
        tokio::time::sleep(Duration::from_millis(50)).await;
    }
    assert!(
        observed.load(Ordering::SeqCst),
        "server must observe CANCEL once BOTH halves drop",
    );
}

/// Regression (cubic-dev-ai bot P2): mirror of the client-stream
/// guard for duplex. `Some(0)` would deadlock the upload sink the
/// same way (initial REQUEST is lazy → server never sees the call
/// → no GRANT ever lands); reject before any wire traffic.
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn call_duplex_rejects_zero_request_window() {
    let caller = build_node().await;
    let target = 0xC0DE_u64;
    let opts = CallOptions {
        request_window_initial: Some(0),
        ..CallOptions::default()
    };
    let err = match caller.call_duplex(target, "anything", opts).await {
        Ok(_) => panic!("Some(0) must be rejected before any wire traffic"),
        Err(e) => e,
    };
    match err {
        RpcError::Codec { direction, message } => {
            assert_eq!(direction, CodecDirection::Encode);
            assert!(
                message.contains("request_window_initial"),
                "diagnostic must name the offending option: {message}",
            );
        }
        other => panic!("expected RpcError::Codec(Encode), got {other:?}"),
    }
}