[TomCC7]

OdometryHist now consumes generated Odometry and publishes generated Path values. It reads the standard nested pose, retains each captured PoseStamped Header, and compares source nanoseconds for the existing publication interval/replay-reset logic. Generated snapshots own their pose data. Rerun renders generated Path geometry and its full source fields outside the value class; the existing true-height override also continues to accept legacy Path callers.

Parent: #4671. Part of the message encoding refactor.

Area	Change
OdometryHist	Generated input/output, nested pose access, integer source-time comparisons and independent Path snapshots; existing handler/config names retained
Existing Rerun CDR adapter	Path line strips with existing color/lift/radius defaults, parent frame and full field text
One adjacent test file	Real memory-LCM RPC/ports, interval queue, stationary updates, zero/epoch time, replay backwards, max_poses, snapshot ownership, frame override and rendering
Workflow	Installed-production provenance and the focused functional/typing checks

Check the generated value/renderer from the proposal's installed main package:

import numpy as np
from dimos.messages.geometry_msgs.msg import PoseStamped
from dimos.messages.nav_msgs.msg import Path
from dimos.messages.std_msgs.msg import Header
from dimos.msgs.geometry import pose_from_matrix
from dimos.msgs.time import time_from_nanoseconds
from dimos.visualization.rerun.cdr import generated_archetypes

header = Header(frame_id="map", stamp=time_from_nanoseconds(1700000000123456789))
matrix = np.eye(4)
matrix[:3, 3] = [1, 2, 3]
path = Path(header=header, poses=[PoseStamped(header=header, pose=pose_from_matrix(matrix))])
entries = generated_archetypes(path, "world/path")
assert entries[0][1].strips.as_arrow_array().to_pylist() == [[[1.0, 2.0, 3.5]]]
assert entries[1][1].parent_frame.as_arrow_array().to_pylist() == ["tf#/map"]

From a prepared checkout, python -m pytest --noconftest -o addopts='' -q dimos/mapping/test_generated_odometry_hist.py dimos/visualization/rerun/test_cdr.py runs 21 checks, including actual CDR input dispatch and output. It needs normal local process IPC, with RPC/CDR explicitly using memq://; no external network or assets.

Validation: 21 source and 21 installed-main-wheel checks pass, with actual production modules asserted to come from site-packages. Two production files pass strict typing; applicable hooks, normal main wheel and default 198-package pip check pass. The exact renderer block also ran in the default-dependency installed environment. Wheel SHA256 22d5b3723e9c752c2543eb8a5a960b9d8064b45dd5871a12679c4071898db8cc.

The restricted local executor blocked socketpair.send with EPERM and could not wake asyncio; the same bounded checks passed when local IPC was permitted. No core scheduler/API workaround was added. No hardware, simulator, benchmark investigation, human visual acceptance or CI polling. Remaining odometry producers/default encoding and legacy retirement stay in subsequent cutover work.