Files
dimgigov 2d09edd9f7 fix+refactor: soft keywords as identifiers, full test wiring, ORC crash docs
- parser: clause keywords (header, format, status, user, csv, ...) now work
  as identifiers everywhere; IMPORT/EXPORT accept FORMAT csv/HEADER true
- nimble test + CI run all 13 test suites (650 checks green)
- ExecutionContext.registry is now {.cursor.} (breaks registry<->ctx cycle)
- ORC crash reproduced and bisected (tests/orc_repro.py); ARC stays the MM
2026-07-30 13:11:28 +03:00

58 lines
2.2 KiB
Python

#!/usr/bin/env python3
"""Reproducer for the ORC cycle-collector crash (markGray/trace SIGSEGV).
Usage:
1. Build the server with ORC:
nim c -d:ssl --threads:on --path:src --mm:orc -o:/tmp/baradadb_orc src/baradadb.nim
2. Start it:
BARADB_PORT=39472 BARADB_DATA_DIR=/tmp/baradb_orc_data BARADB_LOG_LEVEL=error /tmp/baradadb_orc
3. Run this script (from the repo root):
python3 tests/orc_repro.py
Expected (as of 2026-07-30, Nim 2.2.10): the server dies with
orc.nim markGray -> trace -> SIGSEGV
triggered from core/server.nim handleClient, and this script fails with
ConnectionResetError. Under ARC (the default in nim.cfg) the same load passes.
Bisect results: 200 pings + 200 SELECTs over TCP are fine; the crash lands
somewhere between 20 and 500 sequential INSERTs (INSERT path only).
Failed root-cause attempts: callback cycle breaks (ed5a719), {.cursor.} on
ExecutionContext.registry (the registry<->ctx cycle), guarding ctx.onChange
against zero WS subscribers. Conclusion: deep ORC+async issue (possibly
upstream Nim), not a single app-level cycle. ARC remains the supported MM.
"""
import asyncio
import sys
sys.path.insert(0, "clients/python")
from baradb import Client
PORT = 39472
async def worker(w: int) -> None:
base = 1000 + w * 100
async with Client("127.0.0.1", PORT) as c:
for i in range(100):
await c.query(f"INSERT INTO orc_stress (id, val) VALUES ({base + i}, 'v{base + i}')")
async def main() -> None:
async with Client("127.0.0.1", PORT) as c:
await c.query("CREATE TABLE orc_stress (id INT PRIMARY KEY, val STRING)")
# Original report: SIGSEGV after ~20 sequential INSERTs under async load
for i in range(1000):
await c.query(f"INSERT INTO orc_stress (id, val) VALUES ({i}, 'x{i}')")
r = await c.query("SELECT COUNT(*) AS n FROM orc_stress")
print("after 1000 sequential INSERTs:", r.rows)
await asyncio.gather(*[worker(w) for w in range(10)])
async with Client("127.0.0.1", PORT) as c:
r = await c.query("SELECT COUNT(*) AS n FROM orc_stress")
print("final count:", r.rows)
print("ping:", await c.ping())
print("ORC STRESS OK")
asyncio.run(main())