The `open_actor_cluster()` teardown hangs
intermittently on UDS when `gather_contexts(mngrs=())`
raises `ValueError` mid-setup; likely a race in the
actor-nursery cleanup vs UDS socket shutdown. TCP
passes reliably (5/5 runs).
- Add `tpt_proto` fixture param to the test
- `pytest.skip()` on UDS with a TODO for deeper
investigation of `._clustering`/`._supervise`
teardown paths
(this patch was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Factor the CPU-freq-scaling helper out of
`test_legacy_one_way_streaming` into `conftest.py`
alongside a new `cpu_scaling_factor()` convenience fn
that returns a latency-headroom multiplier (>= 1.0).
Apply it to the two other flaky-timeout tests,
- `test_cancel_via_SIGINT_other_task`: 2s -> scaled
- `test_example[we_are_processes.py]`: 16s -> scaled
Deats,
- add `get_cpu_state()` + `cpu_scaling_factor()` to
`conftest.py` so all test mods can share the logic.
- catch `IndexError` (empty glob) in addition to
`FileNotFoundError`.
- rename `factor` var -> `headroom` at call sites for
clarity on intent.
(this patch was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Add `get_cpu_state()` helper to read CPU freq settings
from `/sys/devices/system/cpu/` and use it to compensate
the perf time-limit when `auto-cpufreq` (or similar)
scales down the max frequency.
Deats,
- read `*_pstate_max_freq` and `scaling_max_freq`
to compute a `cpu_scaled` ratio.
- when `cpu_scaled != 1.`, increase `this_fast` limit
proportionally (factoring dual-threaded cores).
- log a warning via `test_log` when compensating.
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Move the `Arbiter` class out of `runtime._runtime` into its
logical home at `discovery._registry` as `Registrar(Actor)`.
This completes the long-standing terminology migration from
"arbiter" to "registrar/registry" throughout the codebase.
Deats,
- add new `discovery/_registry.py` mod with `Registrar`
class + backward-compat `Arbiter = Registrar` alias.
- rename `Actor.is_arbiter` attr -> `.is_registrar`;
old attr now a `@property` with `DeprecationWarning`.
- `_root.py` imports `Registrar` directly for
root-actor instantiation.
- export `Registrar` + `Arbiter` from `tractor.__init__`.
- `_runtime.py` re-imports from `discovery._registry`
for backward compat.
Also,
- update all test files to use `.is_registrar`
(`test_local`, `test_rpc`, `test_spawning`,
`test_discovery`, `test_multi_program`).
- update "arbiter" -> "registrar" in comments/docstrings
across `_discovery.py`, `_server.py`, `_transport.py`,
`_testing/pytest.py`, and examples.
- drop resolved TODOs from `_runtime.py` and `_root.py`.
(this patch was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Adjust all `tractor._state`, `tractor._addr`,
`tractor._supervise`, etc. refs in tests and examples
to use the new `runtime/`, `discovery/`, `spawn/` paths.
Also,
- use `tractor.debug_mode()` pub API instead of
`tractor._state.debug_mode()` in a few test mods
- add explicit `timeout=20` to `test_respawn_consumer_task`
`@tractor_test` deco call
(this patch was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
- Use `bidict.forceput()` in `register_actor()` to handle
duplicate addr values from stale entries or actor restarts.
- Fix `uid` annotation to `tuple[str, str]|None` in
`maybe_open_portal()` and handle the `None` return from
`delete_addr()` in log output.
- Pass explicit `registry_addrs=[reg_addr]` to `open_nursery()`
and `find_actor()` in `test_stale_entry_is_deleted` to ensure
the test uses the remote registrar.
- Update `query_actor()` docstring to document the new
`(addr, reg_portal)` yield shape.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
By spawning an actor task that immediately shuts down the transport
server and then sleeps, verify that attempting to connect via the
`._discovery.find_actor()` helper delivers `None` for the `Portal`
value.
Relates to #184 and #216
- `test_inter_peer_cancellation`: swap all `.uid` refs
on `Actor`, `Channel`, and `Portal` to `.aid.uid`
- `test_legacy_one_way_streaming`: same + fix `print()`
to multiline style
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Use 6s timeout on non-linux (vs 4s) in
`test_cancel_while_childs_child_in_sync_sleep()` to avoid
flaky `TooSlowError` on slower CI runners.
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Removes the `pytest` deprecation warns and attempts to avoid
some de-registration raciness, though i'm starting to think the
real issue is due to not having the fixes from #366 (which handle
the new dereg on `OSError` case from UDS)?
- use `.channel.aid.uid` over deprecated `.channel.uid`
throughout `test_discovery.py`.
- add polling loop (up to 5s) for subactor de-reg check
in `spawn_and_check_registry()` to handle slower
transports like UDS where teardown takes longer.
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Add UDS skip-guard to `test_streaming_to_actor_cluster()`
and plumb `tpt_proto` through the `@tractor_test` wrapper
so transport-parametrized tests can receive it.
Deats,
- skip cluster test when `tpt_proto == 'uds'` with
descriptive msg, add TODO about `@pytest.mark.no_tpt`.
- add `tpt_proto: str|None` param to inner wrapper in
`tractor_test()`, forward to decorated fn when its sig
accepts it.
- register custom `no_tpt` marker via `pytest_configure()`
to avoid unknown-marker warnings.
- add masked todo for `no_tpt` marker-check code in `tpt_proto` fixture
(needs fn-scope to work, left as TODO).
- add `request` param to `tpt_proto` fixture for future
marker inspection.
Also,
- add doc-string to `test_streaming_to_actor_cluster()`.
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Namely the workaround expected exc branches added in ef7ed7a for the UDS
parametrization. With the new boxing of the underlying CREs as
tpt-closed, we can expect the same exc outcomes as in the TCP cases.
Also this tweaks some error report logging content used while debugging
this,
- properly `repr()` the `TransportClosed.src_exc`-type from
the maybe emit in `.report_n_maybe_raise()`.
- remove the redudant `chan.raddr` from the "closed abruptly"
header in the tpt-closed handler of `._rpc.process_messages()`,
the `Channel.__repr__()` now contains it by default.
Forward the `tpt_proto` fixture val into spawned daemon
subprocesses via `run_daemon(enable_transports=..)` and
sync `_runtime_vars['_enable_tpts']` in the `tpt_proto`
fixture so sub-actors inherit the transport setting.
Deats,
- add `enable_transports={enable_tpts}` to the daemon
spawn-cmd template in `tests/conftest.py`.
- set `_state._runtime_vars['_enable_tpts']` in the
`tpt_proto` fixture in `_testing/pytest.py`.
(this patch was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
For more reliability with the oob registrar using tests
via the `daemon` fixture,
- increase spawn-wait to `2` in CI, `1` OW; drop
the old py<3.7 branch.
- move `_ci_env` to module-level (above `_non_linux`)
so `_PROC_SPAWN_WAIT` can reference it at parse time.
- add `test_log` fixture param to `daemon()`, use it
for the error-on-exit log line instead of bare `log`.
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Add `expect_timeout: float` param to `_spawn()`
so individual tests can tune `pexpect` timeouts
instead of relying on the hard-coded 3/10 split.
Deats,
- default to 4s, bump by +6 on non-linux CI.
- use walrus `:=` to capture resolved timeout and assert
`spawned.timeout == timeout` for sanity.
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Event on linux i was noticing lotsa false negatives based on sub
teardown race conditions, so this tries to both make way for
(eventually?) expanding the set of suite cases and ensure the current
ones are more reliable on every run.
The main change is to hange the `error_in_child=False` case to use
parent-side-cancellation via a new `trio.move_on_after(timeout)` instead
of `actor.cancel_soon()` (which is now toggled by a new `self_cancel:
bool` but unused rn), and add better teardown assertions.
Low level deats,
- add `rent_cancel`/`self_cancel` params to
`crash_and_clean_tmpdir()` for different cancel paths;
default to `rent_cancel=True` which just sleeps forever
letting the parent's timeout do the work.
- use `trio.move_on_after()` with longer timeouts per
case: 1.6s for error, 1s for cancel.
- use the `.move_on_after()` cancel-scope to assert `.cancel_called`
pnly when `error_in_child=False`, indicating we
parent-graceful-cancelled the sub.
- add `loglevel` fixture, pass to `open_nursery()`.
- log caught `RemoteActorError` via console logger.
- add `ids=` to parametrize for readable test names.
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Add a 6s timeout guard around `test_streaming_to_actor_cluster()`
to catch hangs, and nest the `async with` block inside it.
Found this when running `pytest tests/ --tpt-proto uds`.
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Deliver `(LinkedTaskChannel, Any)` instead of the prior `(first, chan)`
order from `open_channel_from()` to match the type annotation and be
consistent with `trio.open_*_channel()` style where the channel obj
comes first.
- flip `yield first, chan` -> `yield chan, first`
- update type annotation + docstring to match
- swap all unpack sites in tests and examples
(this patch was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Convert every remaining `to_trio`/`from_trio` fn-sig style
to the new unified `chan: LinkedTaskChannel` iface added in
prior commit (c46e9ee8).
Deats,
- `to_trio.send_nowait(val)` (1st call) -> `chan.started_nowait(val)`
- `to_trio.send_nowait(val)` (subsequent) -> `chan.send_nowait(val)`
- `await from_trio.get()` -> `await chan.get()`
Converted fns,
- `sleep_and_err()`, `push_from_aio_task()` in
`tests/test_infected_asyncio.py`
- `sync_and_err()` in `tests/test_root_infect_asyncio.py`
- `aio_streamer()` in
`tests/test_child_manages_service_nursery.py`
- `aio_echo_server()` in
`examples/infected_asyncio_echo_server.py`
- `bp_then_error()` in `examples/debugging/asyncio_bp.py`
Also,
- drop stale comments referencing old param names.
(this patch was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
With methods to comms similar to those that exist for the `trio` side,
- `.get()` which proxies verbatim to the `._to_aio: asyncio.Queue`,
- `.send_nowait()` which thin-wraps to `._to_trio: trio.MemorySendChannel`.
Obviously the more correct design is to break up the channel type into
a pair of handle types, one for each "side's" task in each event-loop,
that's hopefully coming shortly in a follow up patch B)
Also,
- fill in some missing doc strings, tweak some explanation comments and
update todos.
- adjust the `test_aio_errors_and_channel_propagates_and_closes()` suite
to use the new `chan` fn-sig-API with `.open_channel_from()` including
the new methods for msg comms; ensures everything added here works e2e.
Reorganize existing msg-related test suites under
a new `tests/msg/` subdir (matching `tests/devx/`
and `tests/ipc/` convention) and add unit tests for
the `_`-prefixed field filtering in `pformat()`.
Deats,
- `git mv` `test_ext_types_msgspec` and `test_pldrx_limiting` into
`tests/msg/`.
- add `__init__.py` + `conftest.py` for the new test sub-pkg.
- add new `test_pretty_struct.py` suite with 8 unit tests:
- parametrized field visibility (public shown, `_`-private hidden,
mixed)
- direct `iter_struct_ppfmt_lines()` assertion
- nested struct recursion filtering
- empty struct edge case
- real `MsgDec` via `mk_dec()` hiding `_dec`
- `repr()` integration via `Struct.__repr__()`
(this patch was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
It seems something is up with their VM-img or wtv bc i keep increasing
the subproc timeout and nothing is changing. Since i can't try
a `-xlarge` one without paying i'm just muting this test for now.
- convert all doc-strings to `'''` multiline style.
- rename `nursery` -> `an`, `n` -> `tn` to match
project-wide conventions.
- add type annotations to fn params (fixtures, test
helpers).
- break long lines into multiline style for fn calls,
assertions, and `parametrize` decorator lists.
- add `ids=` to `@pytest.mark.parametrize`.
- use `'` over `"` for string literals.
- add `from typing import Callable` import.
- drop spurious blank lines inside generators.
(this commit msg was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Via ensuring `all(mark.args)` on wtv expressions are arg-passed to the
mark decorator; use it to skip the `test_subactor_breakpoint` suite when
`ctlc=True` since it seems too unreliable in CI.
There's a very sloppy registrar-actor-bootup syncing approach used in
this fixture (basically just guessing how long to sleep to wait for it
to init and bind the registry socket) using a `global _PROC_SPAWN_WAIT`
that needs to be made more reliable. But, for now i'm just playing along
with what's there to try and make less CI runs flaky by,
- sleeping *another* 1s when run from non-linux CI.
- reporting stdout (if any) alongside stderr on teardown.
- not strictly requiring a `proc.returncode == -2` indicating successful
graceful cancellation via SIGINT; instead we now error-log and only
raise the RTE on `< 0` exit code.
* though i can't think of why this would happen other then an
underlying crash which should propagate.. but i don't think any test
suite does this intentionally rn?
* though i don't think it should ever happen, having a CI run
"error"-fail bc of this isn't all that illuminating, if there is
some weird `.returncode == 0` termination case it's likely not
a failure?
For later, see the new todo list; we should sync to some kind of "ping"
polling of the tpt address if possible which is already easy enough for
TCP reusing an internal closure from `._root.open_root_actor()`.
Namely, after trying to get `test_multi_daemon_subactors` to work for
the `ctlc=True` case (for way too long), give up on that (see
todo/comments) and skip it; the normal case works just fine. Also tweak
the `test_ctxep_pauses_n_maybe_ipc_breaks` pattern matching for
non-`'UDS'` per the previous script commit; we can't use UDS alongside
`pytest`'s tmp dir generation, mega lulz.
To be a null default and set to `0.1` when not passed by the caller so
as to avoid having to pass `0.1` if you wanted the
param-defined-default.
Also,
- in the `spawn()` fixtures's `unset_colors()` closure, add in a masked
`os.environ['NO_COLOR'] = '1'` since i found it while trying to debug
debugger tests.
- always return the `child.before` content from `assert_before()`
helper; again it comes in handy when debugging console matching tests.
Gud Boi
Tyler Goodlet