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
//! Single integration-test binary for `yah-cloud` (R743-T7).
//!
//! `autotests = false` in `Cargo.toml` turns off cargo's per-file test-target
//! autodiscovery; this file is the `main` target and `mod`s in the siblings, so
//! four former binaries link once instead of four times. Each module keeps its
//! own file — the live `@yah:` annotation blocks in `mesofact_static_e2e.rs`
//! stay where the harvester expects them, and `CARGO_MANIFEST_DIR` (which three
//! of the four walk up from) is unchanged.
//!
//! Test *names* gain a module prefix (`whisper_derive_e2e::derive_pipeline_…`),
//! so filters written against the old flat names still match — `cargo test -p
//! yah-cloud --test main -- derive_pipeline` is unaffected. What does change is
//! `--test <file-stem>`: use `--test main -- <module>::` instead.
//!
//! # `pond_smoke` is deliberately NOT in here
//!
//! It stays its own `[[test]]` target. Three reasons, strongest first:
//!
//! 1. **A committed pipeline addresses it by target name.**
//! `.yah/qed/pond-smoke.toml`'s `pond-spinup-budget` step shells out to
//! `cargo test --release --locked -p cloud --test pond_smoke -- --nocapture`.
//! Folding it into `main` turns that step into a hard cargo error, and the
//! pipeline file is outside this crate.
//!
//! 2. **It is the only test in the crate that reaches for a fixed external
//! address.** MinIO at `http://127.0.0.1:9000` (`pond_smoke.rs`'s
//! `minio_endpoint`) plus named docker containers it `docker rm -f`s in a
//! `Drop` guard. Its blast radius on failure is outside the process, so a
//! standalone runnable handle is worth keeping. (xtask's census —
//! `xtask/src/lib.rs`, `@yah:assumes` — independently found this to be the
//! only real port bind across all ten in-scope crates; nothing else here has
//! global state to collide over.)
//!
//! 3. **It is benchmark-shaped, and libtest parallelises within a binary.**
//! Both its tests assert wall-clock budgets (`WARM_RESTART_BUDGET` = 3 s,
//! `WARDEN_COLD_BUDGET` = 5 s). Merging would drop `whisper_derive_e2e`
//! (two in-process axum servers + BLAKE3 + tempdir IO, ungated) and
//! `live_workspace_smoke` (a full `CloudConfig::load` walk of the real
//! workspace, ungated) into the measurement window on sibling threads. The
//! pipeline also runs it `--nocapture` for the timing report, which merging
//! would interleave with every other test's output.
//!
//! The five modules below have no such conflict: `live_workspace_smoke` is
//! read-only filesystem, `mesofact_static_e2e` binds `127.0.0.1:0` and is gated
//! on `YAH_RECONCILER_E2E_BIN`, `pg_driver_live` is `#[ignore]`d and tempdir
//! -scoped, `passway_apex_live` is `#[ignore]`d and reaches the network
//! read-only (R859-F1), and `whisper_derive_e2e` is entirely in-process on
//! ephemeral ports inside a tempdir. No shared ports, no shared paths, no
//! process-global state.