jakub
6a667d8deb
Include the current addon, terrain integrations, tools, regression tests and recorded development evidence.
53 lines
3.2 KiB
Markdown
53 lines
3.2 KiB
Markdown
# Observer fallback orbit, 13 October 2026
|
|
|
|
The fallback camera now accepts orbit and zoom when no teammate or corpse is
|
|
available locally. Previously both input handling and orbit placement required a
|
|
selected target, leaving the camera fixed when alone or when every surviving
|
|
player was unconscious. Target selection, possession, collision checks and
|
|
recovery behavior are unchanged.
|
|
|
|
The completed local Workbench run passed **38 assertions**, including seven new
|
|
checks for native yaw, pitch and zoom action values, disabled input, unchanged
|
|
casualty/possession, and the actual camera retaining the resulting direction on
|
|
subsequent frames. Existing teammate tracking, player-view interruption and
|
|
resumption, healing, weapon retention and repeated knockout checks also passed.
|
|
See the [a02 report](a02/report.json) and [exact console](a02/console.log.txt).
|
|
|
|
Inputs were supplied programmatically through the native action manager to the
|
|
production camera frame method. This establishes native camera response, not
|
|
physical mouse delivery or dedicated multiplayer behavior. The fixture has one
|
|
local player and a physical stand-in for a teammate. No live server was updated.
|
|
|
|
The first run, [a01](a01/report.json), compiled but failed mission preparation:
|
|
the historical hillside extraction marker lacked room for the newly required
|
|
second US rescue transport. The a02 frozen test copy moved only that marker from
|
|
`7000 0 7500` to previously surveyed flat airport grounds at
|
|
`4776.55 0 11528.9`. [Fixture inputs](a02/fixture-inputs.json) preserve the exact
|
|
arrangement. Launch manifests confirm this is the only addon-source difference
|
|
between the two native runs; neither run is a native pre-fix comparison.
|
|
|
|
The a02 run has zero script/runtime failures. Its strict report retains **21
|
|
engine/resource/platform diagnostics**. Every normalized diagnostic signature
|
|
also appears in a01; a02 has four additional occurrences of the existing
|
|
`worlds/Eden/Eden.ent` resource error as preparation proceeds further. These
|
|
diagnostics are retained, not suppressed or counted as a clean native report.
|
|
The full repository check separately stops at the existing missing saved root
|
|
ID in `ERRM_AtlasLeaf.et`; [exact output](full-check.log.txt) is preserved.
|
|
|
|
Both runners confirm cleanup and unchanged launched addon/external sources.
|
|
The archive verifies manifest/log bindings and copies raw log bytes, launch
|
|
manifests, runner metadata and relevant launched source files. Current production
|
|
observer and native fixture hashes match the completed run; the intentional
|
|
test-marker arrangement does not match the repository's legacy fixture file.
|
|
See [source matches](a02/relevant-source-matches.json) and
|
|
[cross-run review](review.json).
|
|
|
|
The [pre-fix observer](baseline/ERR_Observer.c) and
|
|
[host regression snapshot](regression/test_observer_orbit.py) are retained for
|
|
the separate regression comparison. They are distinguished from native launch
|
|
evidence. The regression fails on the pre-fix missing orbit response and passes
|
|
with the fix; native APIs are doubled in that host test.
|
|
|
|
`archive.py` documents the archive operation and refuses to overwrite evidence.
|
|
The complete runner command is retained in each `launch-source.json`.
|