test_perf_extreme_code_porcelain.py
python
sha256:c5131d76c6eada02939111fda4aa8e51b0c1456b9983727cfd6be101916de14e
merge: pull local/dev — resolve trivial _EXT_MAP symbol con…
Sonnet 4.6
patch
12 days ago
| 1 | """Extreme performance tests for Muse code domain porcelain commands. |
| 2 | |
| 3 | Builds a large synthetic repository (100 Python files, 100 commits, ~500 |
| 4 | symbols per snapshot) and enforces per-command wall-clock budgets. These |
| 5 | are intentionally generous: the goal is to catch commands that have O(N²) |
| 6 | or worse scaling, not to micro-optimise. |
| 7 | |
| 8 | Tiered budgets |
| 9 | -------------- |
| 10 | Fast (< 5 s): commands that touch only the current snapshot or a small index |
| 11 | Medium (< 15 s): commands that walk history but have bounded output |
| 12 | Slow (< 45 s): commands that do deep analysis across the full commit graph |
| 13 | |
| 14 | The repo fixture is built once per module (session-scoped) so it is shared |
| 15 | across all tests to avoid the dominant cost being fixture creation. |
| 16 | |
| 17 | Note: these tests are marked `perf` — run them explicitly with |
| 18 | pytest tests/test_perf_extreme_code_porcelain.py -v -m perf |
| 19 | to avoid slowing the standard CI gate. |
| 20 | """ |
| 21 | |
| 22 | from __future__ import annotations |
| 23 | |
| 24 | import datetime |
| 25 | import json |
| 26 | import pathlib |
| 27 | import time |
| 28 | import pytest |
| 29 | |
| 30 | from muse.core.types import fake_id, blob_id |
| 31 | from muse.core.object_store import write_object as _write_obj_store |
| 32 | from muse.core.paths import heads_dir, muse_dir |
| 33 | from tests.cli_test_helper import CliRunner |
| 34 | |
| 35 | cli = None |
| 36 | runner = CliRunner() |
| 37 | |
| 38 | # --------------------------------------------------------------------------- |
| 39 | # Perf marker — tests can be excluded with `-m "not perf"` on slow CI hosts. |
| 40 | # --------------------------------------------------------------------------- |
| 41 | pytestmark = pytest.mark.perf |
| 42 | |
| 43 | _FAST_S: float = 5.0 |
| 44 | _MEDIUM_S: float = 15.0 |
| 45 | _SLOW_S: float = 45.0 |
| 46 | |
| 47 | _N_FILES: int = 100 |
| 48 | _N_COMMITS: int = 100 |
| 49 | _SYMBOLS_PER_FILE: int = 5 |
| 50 | |
| 51 | |
| 52 | # --------------------------------------------------------------------------- |
| 53 | # Large repo fixture |
| 54 | # --------------------------------------------------------------------------- |
| 55 | |
| 56 | def _env(root: pathlib.Path) -> Manifest: |
| 57 | return {"MUSE_REPO_ROOT": str(root)} |
| 58 | |
| 59 | |
| 60 | def _store_object(root: pathlib.Path, content: bytes) -> str: |
| 61 | oid = blob_id(content) |
| 62 | _write_obj_store(root, oid, content) |
| 63 | return oid |
| 64 | |
| 65 | |
| 66 | def _make_py_source(file_idx: int, commit_idx: int) -> bytes: |
| 67 | """Generate a unique Python source file with _SYMBOLS_PER_FILE functions.""" |
| 68 | lines = [f"# file {file_idx} commit {commit_idx}\n"] |
| 69 | for sym_idx in range(_SYMBOLS_PER_FILE): |
| 70 | lines.append( |
| 71 | f"def func_{file_idx}_{sym_idx}():\n" |
| 72 | f" return {file_idx * 1000 + sym_idx * 100 + commit_idx}\n\n" |
| 73 | ) |
| 74 | return "".join(lines).encode() |
| 75 | |
| 76 | |
| 77 | @pytest.fixture(scope="module") |
| 78 | def large_repo(tmp_path_factory: pytest.TempPathFactory) -> pathlib.Path: |
| 79 | """Build a {_N_FILES}-file × {_N_COMMITS}-commit repo. |
| 80 | |
| 81 | Layout: |
| 82 | - 100 Python source files (src/file_00.py … src/file_99.py) |
| 83 | - 100 commits; each commit mutates a rotating subset of files (10 per commit) |
| 84 | - Total symbols ≈ 100 × 5 × 100 = 50 000 symbol-commit entries in the index |
| 85 | """ |
| 86 | root = tmp_path_factory.mktemp("large_repo") |
| 87 | dot_muse = muse_dir(root) |
| 88 | dot_muse.mkdir() |
| 89 | repo_id = fake_id("repo") |
| 90 | (dot_muse / "repo.json").write_text( |
| 91 | json.dumps({ |
| 92 | "repo_id": repo_id, |
| 93 | "domain": "code", |
| 94 | "default_branch": "main", |
| 95 | "created_at": "2025-01-01T00:00:00+00:00", |
| 96 | }), |
| 97 | encoding="utf-8", |
| 98 | ) |
| 99 | (dot_muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 100 | (dot_muse / "refs" / "heads").mkdir(parents=True) |
| 101 | (dot_muse / "snapshots").mkdir() |
| 102 | (dot_muse / "commits").mkdir() |
| 103 | (dot_muse / "objects").mkdir() |
| 104 | (root / "src").mkdir() |
| 105 | |
| 106 | from muse.core.commits import ( |
| 107 | CommitRecord, |
| 108 | write_commit, |
| 109 | ) |
| 110 | from muse.core.snapshots import ( |
| 111 | SnapshotRecord, |
| 112 | write_snapshot, |
| 113 | ) |
| 114 | from muse.core.ids import hash_commit as compute_commit_id, hash_snapshot as compute_snapshot_id |
| 115 | |
| 116 | # Current manifest: maps file_path → object_id. |
| 117 | manifest: Manifest = {} |
| 118 | parent_id: str | None = None |
| 119 | ref_file = heads_dir(root) / "main" |
| 120 | |
| 121 | for commit_idx in range(_N_COMMITS): |
| 122 | # Each commit touches 10 files (rotating window). |
| 123 | changed_files = [commit_idx % _N_FILES + i for i in range(10)] |
| 124 | changed_files = [f % _N_FILES for f in changed_files] |
| 125 | for file_idx in changed_files: |
| 126 | src = _make_py_source(file_idx, commit_idx) |
| 127 | oid = _store_object(root, src) |
| 128 | rel_path = f"src/file_{file_idx:02d}.py" |
| 129 | manifest[rel_path] = oid |
| 130 | (root / rel_path).write_bytes(src) |
| 131 | |
| 132 | snap_id = compute_snapshot_id(dict(manifest)) |
| 133 | committed_at = datetime.datetime( |
| 134 | 2025, 1, 1, tzinfo=datetime.timezone.utc |
| 135 | ) + datetime.timedelta(hours=commit_idx) |
| 136 | msg = f"commit {commit_idx:04d}: rotate {len(changed_files)} files" |
| 137 | commit_id = compute_commit_id( |
| 138 | parent_ids=[parent_id] if parent_id else [], |
| 139 | snapshot_id=snap_id, |
| 140 | message=msg, |
| 141 | committed_at_iso=committed_at.isoformat(), |
| 142 | ) |
| 143 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=dict(manifest))) |
| 144 | write_commit(root, CommitRecord( |
| 145 | commit_id=commit_id, |
| 146 | branch="main", |
| 147 | snapshot_id=snap_id, |
| 148 | message=msg, |
| 149 | committed_at=committed_at, |
| 150 | parent_commit_id=parent_id, |
| 151 | )) |
| 152 | ref_file.parent.mkdir(parents=True, exist_ok=True) |
| 153 | ref_file.write_text(commit_id, encoding="utf-8") |
| 154 | parent_id = commit_id |
| 155 | |
| 156 | return root |
| 157 | |
| 158 | |
| 159 | # --------------------------------------------------------------------------- |
| 160 | # Timing helper |
| 161 | # --------------------------------------------------------------------------- |
| 162 | |
| 163 | def _run_timed(root: pathlib.Path, args: list[str], budget_s: float) -> None: |
| 164 | t0 = time.monotonic() |
| 165 | r = runner.invoke(cli, args, env=_env(root)) |
| 166 | elapsed = time.monotonic() - t0 |
| 167 | assert elapsed < budget_s, ( |
| 168 | f"Command {args[:4]} took {elapsed:.2f}s > budget {budget_s}s on " |
| 169 | f"the {_N_FILES}-file × {_N_COMMITS}-commit repo" |
| 170 | ) |
| 171 | assert r.exception is None, ( |
| 172 | f"Command raised unexpectedly: {r.exception}\n{r.output[-500:]}" |
| 173 | ) |
| 174 | |
| 175 | |
| 176 | # --------------------------------------------------------------------------- |
| 177 | # Fast-tier tests (< _FAST_S seconds) |
| 178 | # --------------------------------------------------------------------------- |
| 179 | |
| 180 | class TestFastTierPerf: |
| 181 | """Commands that touch only the current snapshot or a small index.""" |
| 182 | |
| 183 | def test_symbols_perf(self, large_repo: pathlib.Path) -> None: |
| 184 | _run_timed(large_repo, ["code", "symbols", "--json"], _FAST_S) |
| 185 | |
| 186 | def test_grep_perf(self, large_repo: pathlib.Path) -> None: |
| 187 | _run_timed(large_repo, ["code", "grep", "func_0", "--json"], _FAST_S) |
| 188 | |
| 189 | def test_query_perf(self, large_repo: pathlib.Path) -> None: |
| 190 | _run_timed(large_repo, ["code", "query", "kind=function", "--json"], _FAST_S) |
| 191 | |
| 192 | def test_cat_perf(self, large_repo: pathlib.Path) -> None: |
| 193 | _run_timed( |
| 194 | large_repo, ["code", "cat", "src/file_00.py::func_0_0", "--json"], _FAST_S |
| 195 | ) |
| 196 | |
| 197 | def test_languages_perf(self, large_repo: pathlib.Path) -> None: |
| 198 | _run_timed(large_repo, ["code", "languages", "--json"], _FAST_S) |
| 199 | |
| 200 | def test_api_surface_perf(self, large_repo: pathlib.Path) -> None: |
| 201 | _run_timed(large_repo, ["code", "api-surface", "--json"], _FAST_S) |
| 202 | |
| 203 | def test_deps_perf(self, large_repo: pathlib.Path) -> None: |
| 204 | _run_timed(large_repo, ["code", "deps", "src/file_00.py", "--json"], _FAST_S) |
| 205 | |
| 206 | def test_impact_perf(self, large_repo: pathlib.Path) -> None: |
| 207 | _run_timed( |
| 208 | large_repo, |
| 209 | ["code", "impact", "src/file_00.py::func_0_0", "--json"], |
| 210 | _FAST_S, |
| 211 | ) |
| 212 | |
| 213 | def test_breakage_perf(self, large_repo: pathlib.Path) -> None: |
| 214 | _run_timed(large_repo, ["code", "breakage", "--json"], _FAST_S) |
| 215 | |
| 216 | |
| 217 | # --------------------------------------------------------------------------- |
| 218 | # Medium-tier tests (< _MEDIUM_S seconds) |
| 219 | # --------------------------------------------------------------------------- |
| 220 | |
| 221 | class TestMediumTierPerf: |
| 222 | """Commands that walk history but have bounded output size.""" |
| 223 | |
| 224 | def test_hotspots_perf(self, large_repo: pathlib.Path) -> None: |
| 225 | _run_timed( |
| 226 | large_repo, |
| 227 | ["code", "hotspots", "--top", "20", "--max-commits", "50", "--json"], |
| 228 | _MEDIUM_S, |
| 229 | ) |
| 230 | |
| 231 | def test_stable_perf(self, large_repo: pathlib.Path) -> None: |
| 232 | _run_timed( |
| 233 | large_repo, ["code", "stable", "--top", "20", "--json"], _MEDIUM_S |
| 234 | ) |
| 235 | |
| 236 | def test_coupling_perf(self, large_repo: pathlib.Path) -> None: |
| 237 | _run_timed( |
| 238 | large_repo, |
| 239 | ["code", "coupling", "--top", "20", "--min", "2", "--json"], |
| 240 | _MEDIUM_S, |
| 241 | ) |
| 242 | |
| 243 | def test_blast_risk_perf(self, large_repo: pathlib.Path) -> None: |
| 244 | _run_timed( |
| 245 | large_repo, |
| 246 | ["code", "blast-risk", "--top", "10", "--max-commits", "30", "--json"], |
| 247 | _MEDIUM_S, |
| 248 | ) |
| 249 | |
| 250 | def test_age_perf(self, large_repo: pathlib.Path) -> None: |
| 251 | _run_timed( |
| 252 | large_repo, |
| 253 | [ |
| 254 | "code", "age", "src/file_00.py::func_0_0", |
| 255 | "--max-commits", "30", "--json", |
| 256 | ], |
| 257 | _MEDIUM_S, |
| 258 | ) |
| 259 | |
| 260 | def test_velocity_perf(self, large_repo: pathlib.Path) -> None: |
| 261 | _run_timed( |
| 262 | large_repo, |
| 263 | ["code", "velocity", "--top", "10", "--max-commits", "30", "--json"], |
| 264 | _MEDIUM_S, |
| 265 | ) |
| 266 | |
| 267 | def test_entangle_perf(self, large_repo: pathlib.Path) -> None: |
| 268 | _run_timed( |
| 269 | large_repo, |
| 270 | ["code", "entangle", "--top", "10", "--max-commits", "30", "--json"], |
| 271 | _MEDIUM_S, |
| 272 | ) |
| 273 | |
| 274 | def test_find_symbol_perf(self, large_repo: pathlib.Path) -> None: |
| 275 | _run_timed( |
| 276 | large_repo, |
| 277 | ["code", "find-symbol", "--name", "func_0_0", "--limit", "50", "--json"], |
| 278 | _MEDIUM_S, |
| 279 | ) |
| 280 | |
| 281 | def test_symbol_log_perf(self, large_repo: pathlib.Path) -> None: |
| 282 | _run_timed( |
| 283 | large_repo, |
| 284 | ["code", "symbol-log", "src/file_00.py::func_0_0", "--max", "30", "--json"], |
| 285 | _MEDIUM_S, |
| 286 | ) |
| 287 | |
| 288 | def test_blame_perf(self, large_repo: pathlib.Path) -> None: |
| 289 | _run_timed( |
| 290 | large_repo, |
| 291 | ["code", "blame", "src/file_00.py::func_0_0", "--max", "30", "--json"], |
| 292 | _MEDIUM_S, |
| 293 | ) |
| 294 | |
| 295 | def test_detect_refactor_perf(self, large_repo: pathlib.Path) -> None: |
| 296 | _run_timed( |
| 297 | large_repo, |
| 298 | ["code", "detect-refactor", "--max-commits", "30", "--json"], |
| 299 | _MEDIUM_S, |
| 300 | ) |
| 301 | |
| 302 | def test_compare_perf(self, large_repo: pathlib.Path) -> None: |
| 303 | _run_timed( |
| 304 | large_repo, ["code", "compare", "HEAD~10", "HEAD", "--json"], _MEDIUM_S |
| 305 | ) |
| 306 | |
| 307 | def test_predict_perf(self, large_repo: pathlib.Path) -> None: |
| 308 | _run_timed( |
| 309 | large_repo, |
| 310 | ["code", "predict", "--top", "10", "--max-commits", "30", "--json"], |
| 311 | _MEDIUM_S, |
| 312 | ) |
| 313 | |
| 314 | |
| 315 | # --------------------------------------------------------------------------- |
| 316 | # Slow-tier tests (< _SLOW_S seconds) |
| 317 | # --------------------------------------------------------------------------- |
| 318 | |
| 319 | class TestSlowTierPerf: |
| 320 | """Commands that do deep graph analysis or full-history traversal.""" |
| 321 | |
| 322 | def test_narrative_perf(self, large_repo: pathlib.Path) -> None: |
| 323 | _run_timed( |
| 324 | large_repo, |
| 325 | [ |
| 326 | "code", "narrative", "src/file_00.py::func_0_0", |
| 327 | "--max-commits", "50", "--json", |
| 328 | ], |
| 329 | _SLOW_S, |
| 330 | ) |
| 331 | |
| 332 | def test_gravity_perf(self, large_repo: pathlib.Path) -> None: |
| 333 | _run_timed( |
| 334 | large_repo, |
| 335 | [ |
| 336 | "code", "gravity", "src/file_00.py::func_0_0", |
| 337 | "--max-commits", "30", "--json", |
| 338 | ], |
| 339 | _SLOW_S, |
| 340 | ) |
| 341 | |
| 342 | def test_contract_perf(self, large_repo: pathlib.Path) -> None: |
| 343 | _run_timed( |
| 344 | large_repo, |
| 345 | [ |
| 346 | "code", "contract", "src/file_00.py::func_0_0", |
| 347 | "--max-commits", "30", "--json", |
| 348 | ], |
| 349 | _SLOW_S, |
| 350 | ) |
| 351 | |
| 352 | def test_dead_perf(self, large_repo: pathlib.Path) -> None: |
| 353 | _run_timed( |
| 354 | large_repo, ["code", "dead", "--workers", "4", "--json"], _SLOW_S |
| 355 | ) |
| 356 | |
| 357 | def test_codemap_perf(self, large_repo: pathlib.Path) -> None: |
| 358 | _run_timed( |
| 359 | large_repo, ["code", "codemap", "--top", "30", "--json"], _SLOW_S |
| 360 | ) |
| 361 | |
| 362 | def test_clones_perf(self, large_repo: pathlib.Path) -> None: |
| 363 | _run_timed(large_repo, ["code", "clones", "--json"], _SLOW_S) |
| 364 | |
| 365 | def test_semantic_test_coverage_perf(self, large_repo: pathlib.Path) -> None: |
| 366 | _run_timed( |
| 367 | large_repo, |
| 368 | ["code", "semantic-test-coverage", "--max-commits", "30", "--json"], |
| 369 | _SLOW_S, |
| 370 | ) |
| 371 | |
| 372 | def test_lineage_perf(self, large_repo: pathlib.Path) -> None: |
| 373 | _run_timed( |
| 374 | large_repo, |
| 375 | ["code", "lineage", "src/file_00.py::func_0_0", "--json"], |
| 376 | _SLOW_S, |
| 377 | ) |
| 378 | |
| 379 | def test_coverage_perf(self, large_repo: pathlib.Path) -> None: |
| 380 | _run_timed( |
| 381 | large_repo, |
| 382 | ["code", "coverage", "src/file_00.py::func_0_0", "--json"], |
| 383 | _SLOW_S, |
| 384 | ) |
File History
5 commits
sha256:c5131d76c6eada02939111fda4aa8e51b0c1456b9983727cfd6be101916de14e
merge: pull local/dev — resolve trivial _EXT_MAP symbol con…
Sonnet 4.6
patch
12 days ago
sha256:9c33d61749fff814c5226d5386aa2af7064c2c02788594a25fdd709358132eea
fix: _PROPOSAL_PREFIX_RESOLVE_LIMIT 200 → 100 to match hub …
Sonnet 4.6
19 days ago
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
22 days ago
sha256:c06a9b9b9fee26c68ea725b44d54b2c0a171301ce9de746d5b656617b4463a9a
fix: repair four test failures from post-migration audit
Sonnet 4.6
patch
28 days ago
sha256:1900655993c83c4107067375548a7be823e471d2515830842f1a12cba4bd3cdf
fix: unified object store migration — idempotent writes, JS…
Sonnet 4.6
minor
⚠
29 days ago