Teknium
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).
93 lines
3.3 KiB
Python
93 lines
3.3 KiB
Python
"""Regression test for issue #51587.
|
|
|
|
MCP tools connected/enabled but never surfaced into the agent's session
|
|
toolset on the desktop app + dashboard WebUI.
|
|
|
|
Root cause: there are two independent background MCP discovery thread owners
|
|
by surface:
|
|
|
|
* ``tui_gateway.entry`` — the stdio ``hermes --tui`` path.
|
|
* ``hermes_cli.mcp_startup`` — the desktop app + dashboard WebSocket sidecar
|
|
(``tui_gateway/ws.py``) and ``hermes dashboard``.
|
|
|
|
The late-refresh scheduler (``tui_gateway.server._schedule_mcp_late_refresh``)
|
|
gates on ``tui_gateway.entry.mcp_discovery_in_flight()``. Before the fix that
|
|
function read ONLY ``tui_gateway.entry._mcp_discovery_thread``. On the
|
|
desktop/dashboard surfaces that thread is ``None`` (the thread lives on
|
|
``hermes_cli.mcp_startup``), so the scheduler bailed immediately and a slow MCP
|
|
server's tools never surfaced for the whole session — even after a container
|
|
restart. The fix makes ``mcp_discovery_in_flight`` / ``join_mcp_discovery``
|
|
consult BOTH thread owners.
|
|
"""
|
|
|
|
import threading
|
|
|
|
import pytest
|
|
|
|
import hermes_cli.mcp_startup as startup
|
|
import tui_gateway.entry as entry
|
|
|
|
|
|
@pytest.fixture
|
|
def clean_discovery_globals():
|
|
"""Snapshot and restore both modules' discovery-thread globals."""
|
|
saved_entry = entry._mcp_discovery_thread
|
|
saved_startup = startup._mcp_discovery_thread
|
|
entry._mcp_discovery_thread = None
|
|
startup._mcp_discovery_thread = None
|
|
try:
|
|
yield
|
|
finally:
|
|
entry._mcp_discovery_thread = saved_entry
|
|
startup._mcp_discovery_thread = saved_startup
|
|
|
|
|
|
def _alive_thread(stop: threading.Event) -> threading.Thread:
|
|
t = threading.Thread(target=lambda: stop.wait(5.0), daemon=True)
|
|
t.start()
|
|
return t
|
|
|
|
|
|
def test_entry_in_flight_sees_startup_thread(clean_discovery_globals):
|
|
"""Desktop/dashboard surface: discovery thread lives on hermes_cli.mcp_startup.
|
|
|
|
The entry-level in-flight check must report True so the late-refresh
|
|
scheduler does not bail (the #51587 bug).
|
|
"""
|
|
stop = threading.Event()
|
|
startup._mcp_discovery_thread = _alive_thread(stop)
|
|
try:
|
|
# Entry's own thread is None, but the startup thread is alive.
|
|
assert entry._mcp_discovery_thread is None
|
|
assert entry.mcp_discovery_in_flight() is True
|
|
finally:
|
|
stop.set()
|
|
startup._mcp_discovery_thread.join(timeout=2.0)
|
|
|
|
# After the thread exits, neither owner is in flight.
|
|
assert entry.mcp_discovery_in_flight() is False
|
|
|
|
|
|
def test_no_mcp_threads_not_in_flight(clean_discovery_globals):
|
|
"""No discovery anywhere → not in flight, join reports done immediately."""
|
|
assert entry.mcp_discovery_in_flight() is False
|
|
assert entry.join_mcp_discovery(timeout=0.1) is True
|
|
|
|
|
|
def test_startup_module_exposes_in_flight_helpers(clean_discovery_globals):
|
|
"""hermes_cli.mcp_startup gains the in-flight/join helpers entry delegates to."""
|
|
assert startup.mcp_discovery_in_flight() is False
|
|
assert startup.join_mcp_discovery(timeout=0.1) is True
|
|
|
|
stop = threading.Event()
|
|
t = _alive_thread(stop)
|
|
startup._mcp_discovery_thread = t
|
|
try:
|
|
assert startup.mcp_discovery_in_flight() is True
|
|
assert startup.join_mcp_discovery(timeout=0.1) is False
|
|
finally:
|
|
stop.set()
|
|
t.join(timeout=2.0)
|
|
assert startup.mcp_discovery_in_flight() is False
|
|
assert startup.join_mcp_discovery(timeout=2.0) is True
|