gabriel / muse public
test_wire_step2a_pack.py python
166 lines 5.3 KB
Raw
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d feat(pack): delta-encode snapshots in MPackBundle wire format Sonnet 4.6 minor ⚠ breaking 122 days ago
1 """Wire protocol step 2A — Pack performance gate (client-side).
2
3 Ticket #45, Step 2A: BFS walk + bundle serialization must complete in under 5s.
4
5 This test measures the two operations that constitute the client-side pack phase:
6
7 build_mpack(root, [head], have=[]) — BFS walk + collect all objects
8 msgpack.packb(bundle, use_bin_type) — serialize bundle to wire bytes
9
10 Total < 5s
11
12 Repo size: 100 commits, 600 unique objects, 4 KiB blobs (~2.4 MB raw).
13 600 objects sits just above _PRESIGN_OBJECT_THRESHOLD (500) — the size class
14 that triggers the new bundle upload path instead of N individual PUTs.
15
16 If the assertion fails, step 2A is not done. Do not move to step 2B.
17 """
18 from __future__ import annotations
19
20 import datetime
21 import pathlib
22 import time
23
24 import msgpack
25 import pytest
26
27 from muse.core.object_store import write_object
28 from muse.core.pack import build_mpack
29 from muse.core.paths import muse_dir
30 from muse.core.snapshot import compute_commit_id, compute_snapshot_id
31 from muse.core.store import (
32 CommitRecord,
33 SnapshotRecord,
34 write_branch_ref,
35 write_commit,
36 write_snapshot,
37 )
38 from muse.core.types import blob_id
39
40
41 # ---------------------------------------------------------------------------
42 # Gate constants
43 # ---------------------------------------------------------------------------
44
45 _N_COMMITS = 100
46 _N_OBJECTS = 600 # just above _PRESIGN_OBJECT_THRESHOLD = 500
47 _BLOB_SIZE = 4096 # 4 KiB per object → ~2.4 MB raw total
48 _TOTAL_GATE_S = 5.0
49
50
51 # ---------------------------------------------------------------------------
52 # Repo fixture
53 # ---------------------------------------------------------------------------
54
55 def _make_repo(tmp: pathlib.Path) -> pathlib.Path:
56 tmp.mkdir(parents=True, exist_ok=True)
57 dot = muse_dir(tmp)
58 dot.mkdir()
59 (dot / "repo.json").write_text('{"repo_id":"step2a","owner":"gabriel"}')
60 for d in ("commits", "snapshots", "objects"):
61 (dot / d).mkdir()
62 (dot / "refs" / "heads").mkdir(parents=True)
63 (dot / "HEAD").write_text("ref: refs/heads/main\n")
64 (dot / "config.toml").write_text("")
65 return tmp
66
67
68 def _populate(repo: pathlib.Path) -> str:
69 """Write _N_OBJECTS blobs + _N_COMMITS chain; return tip commit ID."""
70 blobs: dict[str, str] = {}
71 for i in range(_N_OBJECTS):
72 data = f"step2a-{i:08d}-".encode() + b"x" * _BLOB_SIZE
73 oid = blob_id(data)
74 write_object(repo, oid, data)
75 blobs[f"file_{i:04d}.py"] = oid
76
77 sid = compute_snapshot_id(blobs)
78 write_snapshot(repo, SnapshotRecord(snapshot_id=sid, manifest=blobs))
79
80 parent: str | None = None
81 tip = ""
82 for i in range(_N_COMMITS):
83 ts = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc)
84 msg = f"commit-{i:05d}"
85 cid = compute_commit_id(
86 parent_ids=[parent] if parent else [],
87 snapshot_id=sid,
88 message=msg,
89 committed_at_iso=ts.isoformat(),
90 author="gabriel",
91 )
92 rec = CommitRecord(
93 repo_id="step2a",
94 commit_id=cid,
95 branch="main",
96 snapshot_id=sid,
97 message=msg,
98 committed_at=ts,
99 parent_commit_id=parent,
100 parent2_commit_id=None,
101 author="gabriel",
102 metadata={},
103 structured_delta=None,
104 sem_ver_bump="none",
105 breaking_changes=[],
106 agent_id="",
107 model_id="",
108 toolchain_id="",
109 prompt_hash="",
110 signature="",
111 signer_key_id="",
112 )
113 write_commit(repo, rec)
114 parent = cid
115 tip = cid
116
117 write_branch_ref(repo, "main", tip)
118 return tip
119
120
121 # ---------------------------------------------------------------------------
122 # THE test
123 # ---------------------------------------------------------------------------
124
125 def test_pack_step2a_performance_gate(tmp_path: pathlib.Path) -> None:
126 """Step 2A gate: build_mpack + msgpack.packb must complete in under 5s.
127
128 This is a pass/fail performance gate, not a benchmark. If the assertion
129 fails, step 2A is not done and step 2B must not be attempted.
130 """
131 repo = _make_repo(tmp_path / "repo")
132 head = _populate(repo)
133
134 t0 = time.perf_counter()
135
136 bundle = build_mpack(repo, [head], have=[])
137 wire_bytes = msgpack.packb(bundle, use_bin_type=True)
138
139 total_s = time.perf_counter() - t0
140
141 assert isinstance(wire_bytes, bytes), "msgpack.packb returned non-bytes"
142 assert len(wire_bytes) > 0, "bundle serialized to empty bytes"
143
144 n_commits = len(bundle.get("commits", []))
145 n_objects = len(bundle.get("objects", []))
146 wire_kb = len(wire_bytes) / 1024
147
148 assert n_commits == _N_COMMITS, (
149 f"expected {_N_COMMITS} commits in bundle, got {n_commits}"
150 )
151 assert n_objects == _N_OBJECTS, (
152 f"expected {_N_OBJECTS} objects in bundle, got {n_objects}"
153 )
154
155 assert total_s < _TOTAL_GATE_S, (
156 f"Step 2A FAIL: build_mpack + packb took {total_s:.2f}s — gate is {_TOTAL_GATE_S}s "
157 f"({n_commits} commits, {n_objects} objects, {wire_kb:.1f} KiB wire)"
158 )
159
160 print(
161 f"\n Step 2A — Pack (client-side)\n"
162 f" Commits: {n_commits}\n"
163 f" Objects: {n_objects}\n"
164 f" Wire size: {wire_kb:.1f} KiB\n"
165 f" Total time: {total_s:.3f}s (gate {_TOTAL_GATE_S}s) ✅"
166 )
File History 1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d feat(pack): delta-encode snapshots in MPackBundle wire format Sonnet 4.6 minor ⚠ 122 days ago