mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-07-31 19:16:29 +00:00
6b81590c55
Systematic prune per AGENTS.md test policy, one pass over every major test tree (gateway, hermes_cli, tools, agent, run_agent, plugins, cli, cron, tui_gateway, honcho/openviking, root-level): - DELETE: source-reading tests (read_text/getsource on prod files), change-detector tests (exact catalog counts, model-name snapshots, config version literals), mock-echo tests (assert a mock returns what it was told), assertion-free/trivial tests, near-duplicate parametrizations (boundaries + one representative kept), async/sync twin duplicates, cosmetic within-file variations. - KEEP (mandatory): security/redaction/approval guards, message-role alternation invariants, prompt-caching/deterministic-call-id invariants, issue-number regression tests (deduped), E2E tests. - 6 test files deleted outright (script-style/no-assert or fully redundant); conftest.py, fakes/, fixtures/ untouched. - tests/acp/conftest.py added: autouse fixture stubs the live models.dev/GitHub/Copilot/Anthropic inventory fetches that ACP server tests performed on every session create — test_server.py 147s → 3.4s, and the tests are now genuinely hermetic. - Sleep-based slowness shrunk where safe (codex_ttfb_watchdog, compression_concurrent_fork, etc.); no wall-clock assertion tightened. Verification: full hermetic suite via scripts/run_tests.sh — 2439 files, 31,130 tests passed, 0 failed, 0 flaky retries, 315s wall (baseline: 583s wall, 13,564s subprocess CPU).
97 lines
3.4 KiB
Python
97 lines
3.4 KiB
Python
"""note_turn_start / note_turn_persisted — the concurrent-turn tripwire.
|
|
|
|
Two turns interleaving on one session corrupt the durable transcript (flush
|
|
order races, identity-dedup row loss, stale history base). The tripwire does
|
|
not prevent the overlap; it names the occurrence with both turn ids so the
|
|
dispatch route that bypassed the busy guard can be identified from logs.
|
|
"""
|
|
|
|
import logging
|
|
|
|
import pytest
|
|
|
|
from agent import agent_runtime_helpers as _helpers
|
|
from agent.agent_runtime_helpers import note_turn_start, note_turn_persisted
|
|
|
|
|
|
class _FakeAgent:
|
|
session_id = "s1"
|
|
|
|
|
|
@pytest.fixture(autouse=True)
|
|
def _clear_inflight_registry():
|
|
"""Isolate the module-level session registry between tests."""
|
|
with _helpers._INFLIGHT_TURNS_LOCK:
|
|
_helpers._INFLIGHT_TURNS_BY_SESSION.clear()
|
|
yield
|
|
with _helpers._INFLIGHT_TURNS_LOCK:
|
|
_helpers._INFLIGHT_TURNS_BY_SESSION.clear()
|
|
|
|
|
|
def test_clean_serial_turns_no_warning(caplog):
|
|
agent = _FakeAgent()
|
|
with caplog.at_level(logging.WARNING, logger="agent.agent_runtime_helpers"):
|
|
assert note_turn_start(agent, "s1:t1:aaaa") is None
|
|
note_turn_persisted(agent)
|
|
assert note_turn_start(agent, "s1:t2:bbbb") is None
|
|
note_turn_persisted(agent)
|
|
assert not caplog.records
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def test_cross_agent_same_session_overlap_warns(caplog):
|
|
"""#64934 route: two routing keys mapped to one session_id run their
|
|
turns on two different agent objects (the gateway agent cache is keyed
|
|
by routing key), so per-agent state alone can never see the overlap."""
|
|
agent_a, agent_b = _FakeAgent(), _FakeAgent()
|
|
with caplog.at_level(logging.WARNING, logger="agent.agent_runtime_helpers"):
|
|
assert note_turn_start(agent_a, "s1:t1:aaaa") is None
|
|
prev = note_turn_start(agent_b, "s1:t2:bbbb")
|
|
assert prev == "s1:t1:aaaa"
|
|
assert len(caplog.records) == 1
|
|
msg = caplog.records[0].getMessage()
|
|
assert "s1:t1:aaaa" in msg and "s1:t2:bbbb" in msg and "s1" in msg
|
|
assert "different agent object" in msg
|
|
|
|
|
|
|
|
|
|
def test_distinct_sessions_never_cross_warn(caplog):
|
|
"""Concurrent turns on different session_ids are legitimate parallelism."""
|
|
agent_a, agent_b = _FakeAgent(), _FakeAgent()
|
|
agent_b.session_id = "s2"
|
|
with caplog.at_level(logging.WARNING, logger="agent.agent_runtime_helpers"):
|
|
note_turn_start(agent_a, "s1:t1:aaaa")
|
|
assert note_turn_start(agent_b, "s2:t2:bbbb") is None
|
|
assert not caplog.records
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def test_persist_disabled_fork_neither_registers_nor_warns(caplog):
|
|
"""Background-review forks share the live parent's session_id for
|
|
prompt-cache warmth but are _persist_disabled — they can never write
|
|
to the transcript, so they must not trip the cross-agent warning
|
|
against the parent's real in-flight turn (in either direction)."""
|
|
parent, fork = _FakeAgent(), _FakeAgent()
|
|
fork._persist_disabled = True
|
|
with caplog.at_level(logging.WARNING, logger="agent.agent_runtime_helpers"):
|
|
note_turn_start(parent, "s1:t1:aaaa") # real turn in flight
|
|
assert note_turn_start(fork, "s1:tr:ffff") is None # fork: silent
|
|
note_turn_persisted(fork) # fork's funnel still runs
|
|
# Reverse order on the next cycle: fork in flight, then real turn.
|
|
note_turn_persisted(parent)
|
|
note_turn_start(fork, "s1:tr:gggg")
|
|
assert note_turn_start(parent, "s1:t2:bbbb") is None
|
|
assert not caplog.records
|
|
|
|
|