Stage 2 fixes: the two tests the review named, and the docs drift

From docs/reviews/2026-10-04-stage2-3-review.md (Stage 3 done; Stage 2
done-with-fixes). Its findings 1, 2, 3 and 5, which it called the only
things standing before merge.

(A) REPLAY, THROUGH THE SINK. What existed was
    `test_every_keyboard_verb_is_cached_for_the_same_reason`, which
    asserts the two verbs are in MUTATING_VERBS -- and the verification
    doc had already named that as the false pass: "asserting
    `deduplicated` and not the sink. A handler that types and then sets
    the flag stays green." Two integration tests now, modelled on
    `test_a_retried_click_with_the_same_request_id_clicks_once`:

      test_a_retried_type_with_the_same_request_id_types_once
        `keyboard.type` of "ab" twice with request_id="retry-me".
        Asserts `deduplicated`, chars == 2 on both, and that the sink
        still has exactly 4 key events -- two characters, a down and an
        up each -- so a replay that typed again is caught by the count
        and not by the flag.

      test_a_retried_unmappable_type_stays_refused_and_types_nothing
        `café` twice with one id. Both refusals are `invalid_argument`,
        the second carries `deduplicated` (a recorded failure is
        replayed as the failure it was), and the sink is empty after
        both -- there is no partial type on either call.

    Mutation-checked rather than assumed: removing `keyboard.key` and
    `keyboard.type` from MUTATING_VERBS fails both.

(B) KILL, THROUGH kill(). `test_killing_the_holder_lifts_its_keys` calls
    `_drop_lease_for` directly, so it proves that helper lifts keys and
    says nothing about whether `kill` still calls it. The integration
    kill test now holds `Shift_L` before the kill and asserts the SINK
    saw `("key", 42, 0)` and `("modifiers", 0, 0, 0, 0)` -- the lease
    forgetting a key is not the compositor being told it came up, and
    only the second matters to the application. Code 42 because `type`
    never emits it, so the release cannot be a leftover.

    Mutation-checked: stubbing out the `_drop_lease_for` call in
    `kill()` fails it. Before this it stayed green.

(C) DOCS. The plan's stage table and risk paragraph still said acquire
    is one warp and that `focuswindow` was unbuilt Stage 4 work. The
    contract (v3 §4) describes what shipped; the plan did not. Rewritten
    so they agree, and the plan now says outright that the contract wins
    where they differ:

      * risk 1 marked RESOLVED, describing the two warps, the pinned
        `follow_mouse_threshold`, and `focuswindow` as the
        nudge-failure fallback selecting on `address:` (not
        `stable_id`, which is not a Hyprland selector)
      * Stage 3 marked Done; Stage 4 struck as not existing -- the
        answer to Stage 0 was the second warp, not a focus dispatch
      * the top banner, the §5 VM assertion list and open question 2
        no longer say "one warp"
      * verification §B3 subtest 1 amended: it is a live assertion
        rather than an expected red, and the check additionally requires
        `focus == "warp"` so the fallback is not accepted as success

    Also amended one sentence the review did not list, because leaving
    it would contradict the edit above inside the same document:
    verification §"Warp does not guarantee wl_keyboard.enter" predicted
    `focuswindow` was "likely mandatory". It was not; the second warp
    was.

    Written down rather than smoothed over: the `focuswindow` branch
    itself has no test, and if both mechanisms fail acquire still
    returns ok with `focus="none"` -- a lease whose first key may go to
    the previously focused window.

(D) The tap docstring claimed the `up` is sent even if the `down` was
    refused after sending. The next lines return early on `not ok`, and
    that early return is right: a refusal means the key was never
    pressed, so an `up` would ask the compositor to release something it
    never had. Sentence replaced with why the early return is correct
    and what actually covers a lost reply (the broker's `release_all`,
    not the CLI, which cannot know whether the request arrived).

Not done, and not claimed: findings 4, 6 and 7 (the broker chord test
checking first/last mask rather than the key; the at-cap test not
looking at the sink; the module-defaults test being a substring match).
The review rates all three as not holes in the claim, and the driver
tests are the locks for the first two.

  nix develop -c python -m pytest tests -q
    907 passed in 67.71s
  nix develop -c python -m pytest tests/pointer tests/sessions/unit -q
    605 passed in 27.16s

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GhNPM6rGcuzxocM1nATnuM
This commit is contained in:
HermenixandClaude Opus 5 committed 2026-10-05 04:43:56 +00:00
1 parent 54dd109dde
commit 561e0c28f5
5 files changed
+321 -15

No files matched your search

@@ -517,14 +517,42 @@ def test_a_button_left_down_is_lifted_when_the_lease_is_released(rig: Rig):
def test_killing_a_session_that_holds_the_cursor_gives_it_back(rig: Rig):
"""Through `kill`, with a button AND a key down.
The key half is here rather than in a unit test on purpose. The unit test
calls `_drop_lease_for` directly, so it proves that helper lifts keys and
says nothing about whether `kill` still calls it -- removing the call would
leave that test green. This goes through the verb, and it asserts on the
SINK rather than on `pointer.state`: the lease forgetting a key is not the
same as the compositor being told it came up, and only the second one
matters to the application.
"""
first = start(rig.daemon, "sleeper")
second = start(rig.daemon, "sleeper")
lease = rig.daemon.call("pointer.acquire", id=first)["lease_id"]
rig.daemon.call("pointer.button", id=first, lease_id=lease, button="right", state="down")
held = rig.daemon.call(
"keyboard.key", id=first, lease_id=lease, key="Shift_L", state="down"
)
assert held["keys"] == ["Shift_L"], held
rig.driver.sink.calls.clear()
rig.daemon.call("kill", id=first, grace_s=0)
# KEY_LEFTSHIFT, and `state=0`. Shift is the key to use because `type`
# never emits code 42, so a release of it cannot be a leftover from
# anything else this rig did.
assert ("key", 42, 0) in rig.driver.sink.calls, (
f"kill did not lift the held key: {rig.driver.sink.calls}. On this "
"compositor a code left in the seat's m_pressed makes the NEXT press "
"of it a no-op, so the key is dead rather than merely stuck."
)
# And the mask with it: a virtual key-up clears no modifier state, so
# without this the next holder inherits Shift.
assert ("modifiers", 0, 0, 0, 0) in rig.driver.sink.calls, rig.driver.sink.calls
assert rig.driver.dispatcher._pressed == {}
assert rig.driver.dispatcher._keys == {}
assert rig.daemon.call("pointer.state")["holder"] is None
assert rig.daemon.call("pointer.acquire", id=second)["ok"]
@@ -561,6 +589,81 @@ def test_a_retried_click_with_the_same_request_id_clicks_once(rig: Rig):
assert len(rig.driver.sink.of("button")) == 2, "one press and one release, not two"
def test_a_retried_type_with_the_same_request_id_types_once(rig: Rig):
"""A retried `type` is the keyboard's version of a second click.
Asserted on the SINK and not on the `deduplicated` flag, because the flag
alone is the false pass the verification doc names: a handler that typed
the string and *then* set the flag would satisfy it. The thing that must
not happen twice is the key events, so those are what is counted.
The cache is not keyboard-specific -- `_dispatch` looks the `request_id` up
before the handler and remembers it after, keyed on the id alone -- so what
this locks is that `keyboard.type` is in the mutating set and therefore
reaches that path at all.
"""
session_id = start(rig.daemon, "sleeper")
lease = rig.daemon.call("pointer.acquire", id=session_id)["lease_id"]
rig.driver.sink.calls.clear()
first = rig.daemon.call(
"keyboard.type", id=session_id, lease_id=lease, text="ab", request_id="retry-me"
)
second = rig.daemon.call(
"keyboard.type", id=session_id, lease_id=lease, text="ab", request_id="retry-me"
)
assert first["ok"] and second["ok"]
assert second.get("deduplicated") is True
assert first["chars"] == 2 and second["chars"] == 2
# Two characters, a down and an up each: four key events for one `type`,
# and the replay must not add a fifth.
assert len(rig.driver.sink.of("key")) == 4, (
f"a replayed type sent {rig.driver.sink.of('key')}; the recorded answer "
"must be returned without typing again"
)
def test_a_retried_unmappable_type_stays_refused_and_types_nothing(rig: Rig):
"""The recorded answer includes a recorded refusal.
An agent that retries a `type` which was refused must see the refusal
again, not a second attempt -- and the sink must still be untouched after
both calls, because validation happens before the first event either time.
"""
session_id = start(rig.daemon, "sleeper")
lease = rig.daemon.call("pointer.acquire", id=session_id)["lease_id"]
rig.driver.sink.calls.clear()
first = rig.daemon.call(
"keyboard.type",
id=session_id,
lease_id=lease,
text="caf\u00e9",
request_id="refuse-me",
)
second = rig.daemon.call(
"keyboard.type",
id=session_id,
lease_id=lease,
text="caf\u00e9",
request_id="refuse-me",
)
assert first["ok"] is False and second["ok"] is False
assert first["error"] == "invalid_argument", first
assert second["error"] == "invalid_argument", second
assert second.get("deduplicated") is True, (
"a recorded failure must be replayed as the failure it was"
)
assert rig.driver.sink.of("key") == [], (
f"an unmappable string typed something: {rig.driver.sink.calls}. There is "
"no partial type, on the first call or the replay."
)
def test_a_restored_bookmark_survives_a_daemon_restart(short_tmpdir: Path):
"""That is what putting it in the record, and not only in the lease, is for."""