From 5a32e746cf7aec3e4ca76616b6b66f023de53988 Mon Sep 17 00:00:00 2001
From: tv <tv@krebsco.de>
Date: Mon, 26 Oct 2026 22:08:32 +0100
Subject: add delta blobs

---
 src/simtest/deltas.rs | 98 +++++++++++++++++++++++++++++++++++++++++++++++++++
 src/simtest/mod.rs    |  1 +
 2 files changed, 99 insertions(+)
 create mode 100644 src/simtest/deltas.rs

(limited to 'src/simtest')

diff --git a/src/simtest/deltas.rs b/src/simtest/deltas.rs
new file mode 100644
index 0000000..ce74607
--- /dev/null
+++ b/src/simtest/deltas.rs
@@ -0,0 +1,98 @@
+//! Large transfers as deltas (`docs/peers.md`): a node joining a swarm with much
+//! history, and one returning after much was written, get the bulk as a blob, and no
+//! sync message they receive is larger than the threshold.
+//!
+//! The nodes use 1 KiB chunks, so the threshold is 16 KiB, and a few hundred KiB of
+//! writes are well above it.
+
+use std::time::Duration;
+
+use serde_json::json;
+
+use super::{Sim, read, write};
+use crate::node::NodeConfig;
+use crate::state::AppState;
+use crate::swarm::bootstrap::Start;
+
+const SOON: Duration = Duration::from_secs(10);
+/// Catching up a few hundred KiB in 1 KiB chunks.
+const CATCH_UP: Duration = Duration::from_secs(60);
+const CHUNK: u32 = 1024;
+/// 16 chunks (`swarm::blob`).
+const THRESHOLD: u64 = 16 * CHUNK as u64;
+
+fn small_chunks(config: &mut NodeConfig) {
+    config.blob_chunk_bytes = CHUNK;
+}
+
+/// About 1 KB that doesn't compress (Automerge compresses large changes).
+fn noise(seed: u64) -> String {
+    let mut rng = crate::testrand::Rng::new(seed);
+    (0..63)
+        .map(|_| format!("{:016x}", rng.next_u64()))
+        .collect()
+}
+
+/// `count` writes of about 1 KB each, under `prefix`.
+async fn write_much(node: &AppState, prefix: &str, count: usize) {
+    for i in 0..count {
+        write(node, &format!("{prefix}/{i}"), json!(noise(i as u64))).await;
+    }
+}
+
+fn largest_sync_received(node: &AppState) -> u64 {
+    node.stats.snapshot().largest_sync_received
+}
+
+fn blob_bytes_received(node: &AppState) -> u64 {
+    node.traffic.total().received.blob.bytes
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 4)]
+async fn joining_node_gets_the_history_as_a_delta() {
+    let sim = Sim::new(41);
+    let a = sim.start_with("a", &[], Start::Genesis, small_chunks);
+    sim.wait_ready(&[&a], SOON).await;
+    write_much(&a, "history", 200).await;
+
+    let b = sim.start_with("b", &["a"], Start::Join, small_chunks);
+    sim.wait_ready(&[&b], CATCH_UP).await;
+    sim.wait_converged(&[&a, &b], CATCH_UP).await;
+    assert_eq!(read(&b, "history/199").await, Some(json!(noise(199))));
+
+    assert!(
+        blob_bytes_received(&b) > 100 * 1_000,
+        "{} blob bytes received",
+        blob_bytes_received(&b)
+    );
+    assert!(
+        largest_sync_received(&b) <= THRESHOLD,
+        "a sync message of {} bytes",
+        largest_sync_received(&b)
+    );
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 4)]
+async fn returning_node_gets_what_it_missed_as_a_delta() {
+    let sim = Sim::new(42);
+    let a = sim.start_with("a", &[], Start::Genesis, small_chunks);
+    let b = sim.start_with("b", &["a"], Start::Join, small_chunks);
+    let (a, b) = (&a, &b);
+    sim.wait_ready(&[a, b], SOON).await;
+    write(a, "early", json!(1)).await;
+    sim.wait_converged(&[a, b], SOON).await;
+
+    sim.net.take_down("b");
+    write_much(a, "while/away", 200).await;
+    let blobs_before = blob_bytes_received(b);
+    sim.net.bring_up("b");
+
+    sim.wait_converged(&[a, b], CATCH_UP).await;
+    assert_eq!(read(b, "while/away/199").await, Some(json!(noise(199))));
+    assert!(blob_bytes_received(b) - blobs_before > 100 * 1_000);
+    assert!(
+        largest_sync_received(b) <= THRESHOLD,
+        "a sync message of {} bytes",
+        largest_sync_received(b)
+    );
+}
diff --git a/src/simtest/mod.rs b/src/simtest/mod.rs
index f7af61b..5e9cc2f 100644
--- a/src/simtest/mod.rs
+++ b/src/simtest/mod.rs
@@ -9,6 +9,7 @@
 //! `RUST_LOG=continuum=debug cargo test simtest -- --nocapture` shows the nodes' logs
 //! (all nodes interleaved; messages name peers by ID).
 
+mod deltas;
 mod remakes;
 mod soak;
 mod timing;
-- 
cgit v1.3.1