fabi/Sylpheed
1
0
Fork 0
Code Issues 9 Pull Requests Actions Packages Projects Releases Wiki Activity

tools+docs: the flight pad was writing into a dead file - every autopilot input went nowhere

Browse Source 
flight_probe.Pad opened /tmp/sylph-vgamepad.fifo, "the vgamepad server's FIFO".
That server was removed when the uinput pad was replaced by Canary's --hid=file
driver, because a uinput device is not namespaced and scripted presses leaked to
the host's desktop. The FIFO is now an ordinary 91-byte file nothing reads, and
/tmp/xenia_pad.txt - the file the emulator polls - was 0 bytes while the
"autopilot" was flying. So every axis, trigger and button from pilot.py,
autopilot3.py, aim_probe.py and flight_probe.py went nowhere, silently.

The corpus already carried this trap for the SHELL scripts
(canary-scripted-input-traps.md, "every call here failed silently"). This class
was missed, and every flight tool imports it.

Verified against the oracle rather than by inspection, on runs confirmed
animating at both ends of every phase: before, full stick produced 0.00 degrees
of heading change over 4 s while the ship travelled 350-735 units, and the
attitude matrix at pos-0x70 was byte-identical; after, LX=-1 turns 12.72 degrees
and LX=+1 swings the flight direction from [1,0,0] to [0.13,-0.14,-0.98], with
the matrix moving 0.44/0.31 under stick and 0.0000 at neutral.

That also REFUTES the "stale attitude matrix" suspicion from the earlier pass -
pos-0x70 is live and tracks the ship; it only looked dead because nothing was
turning the ship.

Still open and said plainly: with the pad fixed a 150 s pilot run still fires 0
times, with |aim yaw| still exactly 90.0 and the target 36-43 km away. Steering
works now, so what is left is the stick SIGN against the pilot's error convention
or a target selection that commits to something too far to close. Both are
finally testable.

findrot_global/findself/findspeed/selfstate still write to the dead FIFO and now
say so in place; they are not repaired because none has been re-run since.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PMRJjbxLqZtsb5Vb7KunPE
This commit is contained in:
Sylpheed RE agent
2026-08-23 22:09:27 +00:00
parent d9cf899667
commit 0ddeffa9a6
7 changed files with 207 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

74
docs/re/pilot-never-fires.md
View File

@@ -1,7 +1,11 @@
# `pilot.py` never pulls the trigger — the proximal cause, measured
**Status: ✅ the gate that blocks it is identified and measured; 🔴 the root cause
is not, and one attempt to find it was invalidated by a freeze.**
**Status: ✅ ROOT CAUSE FOUND AND FIXED (2026-08-23, later the same day) — the
flight tools' pad wrote to a FIFO nothing has read since the uinput pad was
removed, so every stick, trigger and button they sent went nowhere.** The gate
measured below is real; it was open because the craft was never being flown at
all. 🟡 The pilot still does not converge on a target in 150 s with the input
working — see the last section.
## The observation
@@ -75,3 +79,69 @@ Re-run `aim_probe.py` on a run confirmed to be animating **at the end of the
probe as well as the start** — the tool should check `frozen.py` itself and
discard the phase otherwise. If the matrix does move under stick, the question
becomes the sign; if it does not, the binding is what to chase.
---
# ✅ ROOT CAUSE — the flight tools' pad was writing into a dead file
`flight_probe.Pad` opened **`/tmp/sylph-vgamepad.fifo`** and wrote
`axis LX 0.850`-style lines to it, described in its own docstring as "the vgamepad
server's FIFO". That server does not exist any anymore: the uinput pad was removed
because a uinput device is **not namespaced** and every scripted press leaked to
the *host's* desktop, and it was replaced by Canary's `--hid=file` driver. The
FIFO is now an ordinary 91-byte file that nothing reads.
Meanwhile `/tmp/xenia_pad.txt` — the file the emulator actually polls — was
**0 bytes** while the "autopilot" was supposedly flying.
So every axis, trigger and button from `pilot.py`, `autopilot3.py`,
`aim_probe.py` and `flight_probe.py` itself went into a dead file, silently, and
the craft flew on its own for the whole of every run. That is why the target sat
astern at 90.0° forever and the range grew to 49 km: **the ship was never being
turned.**
**The corpus already carried this trap — for the other half of the toolkit.**
[`canary-scripted-input-traps.md`](canary-scripted-input-traps.md) records
"pad.py, NOT `vgamepad`: that command no longer exists … and every call here
failed silently, so the whole scripted route pressed nothing". The **shell**
scripts were fixed then. This class was not, and every flight tool imports it.
## Measured, before and after
Same probe, on runs confirmed animating at both ends of every phase
([`aim_probe.py`](../../tools/re-capture/aim_probe.py) brackets each phase with
`frozen.py` now, after a freeze voided the first attempt):
| | heading change under full stick | attitude matrix `pos-0x70` |
|---|---|---|
| **before** | `0.00°` in every phase, over 4 s, while travelling 350–735 units | `d 0.0000` — unchanged |
| **after** | `12.72°` on `LX=-1`; `LX=+1` swings the flight direction from `[1,0,0]` to `[0.13,-0.14,-0.98]` | `d 0.4438` / `d 0.3108` under stick, `0.0000` at neutral |
🔴 **So the "stale attitude matrix" reading is refuted**, and with it the earlier
suspicion that the binding was wrong: `pos-0x70` is live and tracks the ship. It
only looked dead because nothing was turning the ship.
## 🟡 Still open — the pilot does not converge
With the pad fixed, a 150 s `pilot.py` run still fired **0** times: over its 86
targeted samples `|aim yaw|` is *still* exactly 90.0° and the range is
36–43 km. Steering now works, so what remains is one of:
* the **sign** of the yaw stick against the pilot's own error convention — a
controller that turns away from the error would hold `ez < 0` forever, which is
exactly the observed signature;
* target selection that keeps committing to something 40 km away, which at
~300 units/s cannot be closed inside the window.
Both are now testable, because the input path finally works. The measurement to
make is the direct one: command the pilot's own `sx` for a known error and see
whether `|yaw|` falls or rises.
## Not fixed, and flagged in place
`findrot_global.py`, `findself.py`, `findspeed.py` and `selfstate.py` still write
to the dead FIFO. Each now carries a comment saying so and pointing here. They
are not repaired because none of them has been re-run since, and quietly changing
the input plumbing under a tool whose results are unverified would only
manufacture new unverified results.

54
tools/re-capture/aim_probe.py
View File

@@ -31,6 +31,7 @@ import time
import numpy as np
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import frozen # noqa: E402
import navigator # noqa: E402
from flight_probe import Pad # noqa: E402
@@ -42,23 +43,56 @@ def main():
W.scan()
pad = Pad()
# There is more than one orthonormal 3x3 block near the player object -- a
# scan finds them at pos-0x70 and pos-0x30, and `entities2.py self` has
# picked each on different runs. Watch BOTH, because "the one the config
# names did not move" is a much weaker statement than "neither of the two
# candidates moved".
deltas = [W.rot_delta]
if "--deltas" in sys.argv:
deltas = [int(x, 0) for x in sys.argv[sys.argv.index("--deltas") + 1].split(",")]
def read_block(off, delta):
b = os.pread(W.fd, W.rot_stride * 3 + 12, off + delta)
rows = []
for r in range(3):
row = np.frombuffer(b[W.rot_stride * r:W.rot_stride * r + 12],
dtype=">f4").astype(np.float64)
rows.append(row)
return np.array(rows)
def fwd_now():
ents = W.sample(time.time())
me = next((e for e in ents if "Player" in e[1]), None)
if me is None:
return None, None
return None, None, None
mats = {}
for d in deltas:
try:
M = read_block(me[0], d)
except Exception:
continue
if np.all(np.isfinite(M)):
mats[d] = M
M = W.rot(me[0])
if M is None:
return None, None
return M[W.fwd_row] * W.fwd_sign, me[2]
return None, None, mats
return M[W.fwd_row] * W.fwd_sign, me[2], mats
def phase(name, lx):
# BRACKET the phase with a liveness check at BOTH ends. The first run of
# this probe reported the forward vector pinned and 0.00 deg/s under
# every stick, which reads as "the attitude matrix is dead" -- and the
# guest had simply FROZEN partway through. A dead world holds every
# matrix still, so a phase that ends frozen proves nothing and must be
# thrown away rather than reported.
pad.axis("LX", lx)
f0, p0 = fwd_now()
f0, p0, m0 = fwd_now()
t0 = time.time()
time.sleep(secs)
f1, p1 = fwd_now()
f1, p1, m1 = fwd_now()
pad.axis("LX", 0.0)
froze = frozen.frozen(3.0)[0]
if f0 is None or f1 is None:
print(f"{name:>12}: NO PLAYER/ORIENTATION")
return
@@ -66,10 +100,18 @@ def main():
turn = math.degrees(math.acos(dot)) / max(time.time() - t0, 1e-3)
cross = np.cross(f0, f1)
moved = float(np.linalg.norm(p1 - p0)) if p0 is not None else float("nan")
blocks = " ".join(
f"pos{d:+#06x} d{float(np.abs(m1[d]-m0[d]).max()):.4f}"
for d in deltas if d in m0 and d in m1)
print(f"{name:>12}: |turn| {turn:6.2f} deg/s fwd {f0.round(3)} -> "
f"{f1.round(3)} cross {cross.round(3)} moved {moved:8.1f}")
f"{f1.round(3)} moved {moved:8.1f} [{blocks}]"
f"{' *** GUEST FROZE - DISCARD ***' if froze else ''}")
if frozen.frozen(3.0)[0]:
print("# GUEST IS ALREADY FROZEN — nothing to measure")
return 3
print("# phases: neutral, then full LEFT stick, then full RIGHT stick")
print("# (each phase is followed by a liveness check; a frozen tail voids it)")
phase("neutral", 0.0)
phase("LX=-1", -1.0)
phase("neutral", 0.0)

7
tools/re-capture/findrot_global.py
View File

@@ -20,6 +20,13 @@ import numpy as np
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import gmem # noqa: E402
# 🔴 DEAD FIFO. The vgamepad server this names does not exist any more - the
# uinput pad was replaced by Canary's `--hid=file` driver - so every write here
# goes into an ordinary file that nothing reads, silently. See the rewritten
# `Pad` in flight_probe.py for the format the emulator actually polls, and
# docs/re/pilot-never-fires.md for what this cost. NOT fixed here: these scripts
# have not been re-run since, and changing input plumbing under a tool whose
# results are unverified would only produce new unverified results.
FIFO = "/tmp/sylph-vgamepad.fifo"
TOL = 2e-3

7
tools/re-capture/findself.py
View File

@@ -23,6 +23,13 @@ import numpy as np
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import gmem # noqa: E402
# 🔴 DEAD FIFO. The vgamepad server this names does not exist any more - the
# uinput pad was replaced by Canary's `--hid=file` driver - so every write here
# goes into an ordinary file that nothing reads, silently. See the rewritten
# `Pad` in flight_probe.py for the format the emulator actually polls, and
# docs/re/pilot-never-fires.md for what this cost. NOT fixed here: these scripts
# have not been re-run since, and changing input plumbing under a tool whose
# results are unverified would only produce new unverified results.
FIFO = "/tmp/sylph-vgamepad.fifo"

7
tools/re-capture/findspeed.py
View File

@@ -25,6 +25,13 @@ import numpy as np
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import gmem # noqa: E402
# 🔴 DEAD FIFO. The vgamepad server this names does not exist any more - the
# uinput pad was replaced by Canary's `--hid=file` driver - so every write here
# goes into an ordinary file that nothing reads, silently. See the rewritten
# `Pad` in flight_probe.py for the format the emulator actually polls, and
# docs/re/pilot-never-fires.md for what this cost. NOT fixed here: these scripts
# have not been re-run since, and changing input plumbing under a tool whose
# results are unverified would only produce new unverified results.
FIFO = "/tmp/sylph-vgamepad.fifo"

71
tools/re-capture/flight_probe.py
View File

@@ -19,44 +19,91 @@ import time
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import gworld # noqa: E402
FIFO = "/tmp/sylph-vgamepad.fifo"
# (the old vgamepad FIFO constant is gone with the server it named)
class Pad:
"""Talk to the vgamepad server directly over its FIFO.
"""Drive Canary's `--hid=file` pad by rewriting its state file.
The `vgamepad` CLI spawns a process per command (~20 ms); a control loop
cannot afford that, and `tap`/`hold` additionally sleep *inside* the server.
Writing lines to the FIFO ourselves keeps a tick under a millisecond.
🔴 REWRITTEN 2026-08-23, and this is the whole reason `pilot.py` never fired.
This class used to write lines to `/tmp/sylph-vgamepad.fifo`, "the vgamepad
server's FIFO". That server does not exist any more — the uinput pad was
replaced by the `--hid=file` driver, because a uinput device is not
namespaced and every scripted press leaked to the HOST's desktop. The FIFO is
now an ordinary 91-byte file that **nothing reads**, so every axis, trigger
and button this class emitted went nowhere, silently, and the craft flew on
its own for the whole of every "autopilot" run.
The corpus already carries this exact trap for the MENU path
(`canary-scripted-input-traps.md`: "pad.py, NOT vgamepad … every call here
failed silently"). The shell scripts were fixed then; this class was not, and
every flight tool imports it — `pilot.py`, `autopilot3.py`, `aim_probe.py`,
`flight_probe.py` itself.
Measured symptom, on a run confirmed to be animating throughout: full left
and full right stick for four seconds each, and the ship's **velocity
direction did not move by 0.01°** — heading change 0.00° in every phase while
it travelled 350–735 units. With this rewrite the same probe is what checks
the fix.
Format is the driver's own (`file_input_driver.h`): `key=value` pairs, with
`press=A,START`, `lt`/`rt` 0..255 and `lx`/`ly`/`rx`/`ry` -32768..32767.
Written through a temp file and renamed, so a poll can never see a partial
state; the driver re-parses on any nanosecond-mtime or size change.
"""
PATH = os.environ.get("XENIA_PAD_FILE", "/tmp/xenia_pad.txt")
def __init__(self):
self.f = open(FIFO, "w", buffering=1)
self.state = {"LX": 0.0, "LY": 0.0, "RX": 0.0, "RY": 0.0, "LT": 0.0, "RT": 0.0}
self.state = {"LX": 0.0, "LY": 0.0, "RX": 0.0, "RY": 0.0,
"LT": 0.0, "RT": 0.0}
self.buttons = set()
self._write()
def _write(self):
parts = []
if self.buttons:
parts.append("press=" + ",".join(sorted(self.buttons)))
for k, key in (("LT", "lt"), ("RT", "rt")):
v = int(round(max(0.0, min(1.0, self.state[k])) * 255))
if v:
parts.append(f"{key}={v}")
for k, key in (("LX", "lx"), ("LY", "ly"), ("RX", "rx"), ("RY", "ry")):
v = int(round(max(-1.0, min(1.0, self.state[k])) * 32767))
if v:
parts.append(f"{key}={v}")
tmp = self.PATH + ".tmp"
with open(tmp, "w") as f:
f.write(" ".join(parts))
os.replace(tmp, self.PATH)
def axis(self, name, v):
self.state[name] = v
self.f.write(f"axis {name} {v:.3f}\n")
self._write()
def trig(self, name, v):
self.state[name] = v
self.f.write(f"trig {name} {v:.3f}\n")
self._write()
def press(self, b):
self.f.write(f"press {b}\n")
self.buttons.add(b.upper())
self._write()
def release(self, b):
self.f.write(f"release {b}\n")
self.buttons.discard(b.upper())
self._write()
def reset(self):
self.f.write("reset\n")
for k in self.state:
self.state[k] = 0.0
self.buttons.clear()
self._write()
def vector(self):
return [self.state[k] for k in ("LX", "LY", "RX", "RY", "LT", "RT")]
# (duration_s, description, action)
SCRIPT = [
(3.0, "idle", lambda p: p.reset()),

7
tools/re-capture/selfstate.py
View File

@@ -27,6 +27,13 @@ import gmem # noqa: E402
import gworld # noqa: E402
from findrot_global import scan as scan_rot # noqa: E402
# 🔴 DEAD FIFO. The vgamepad server this names does not exist any more - the
# uinput pad was replaced by Canary's `--hid=file` driver - so every write here
# goes into an ordinary file that nothing reads, silently. See the rewritten
# `Pad` in flight_probe.py for the format the emulator actually polls, and
# docs/re/pilot-never-fires.md for what this cost. NOT fixed here: these scripts
# have not been re-run since, and changing input plumbing under a tool whose
# results are unverified would only produce new unverified results.
FIFO = "/tmp/sylph-vgamepad.fifo"