SCRE still stores its position through inherited scalar fields mPosX/mPosY/mPosZ at offsets 0x5c/0x60/0x64. At PR #2488 head 3982099, the leaf Pos() helper presents those fields as a Vector3 reference. That is an explicit local overlay, not a reconstructed shared Vector3 position member.
The independent review of head 969d498 measured a stack Vector3 alternative: Behavior grew from 92 to 116 bytes and emitted a four-byte Vector3 destructor. This supports retaining the measured boundary while investigating the shared layout; it does not prove every alternative impossible. The current review will attach fresh scoped probe identities in its handoff.
Follow-through owned by @andrewboudreau through the active humanizer/integrator work:
- Determine the coherent shared position interface/layout from existing dActor_c and Vector3 definitions and their actual consumers.
- Test any narrow improvement under pinned 2004/b56 with exact function bytes, module-correct relocation destinations, metadata and complete affected-consumer proof.
- Preserve ordinary C++ construction and the existing allocator destination.
- Keep reconstruction partial until this shared boundary is resolved or an evidence-bounded alternative is independently accepted.
Related PR #2513 is the open Lava cleanup proposing dActor_c::Pos(); it is not a merged shared API or this issue. PR #2488 remains the SCRE source repair. Current byte matching/promotion does not close this follow-up.
SCRE still stores its position through inherited scalar fields mPosX/mPosY/mPosZ at offsets 0x5c/0x60/0x64. At PR #2488 head 3982099, the leaf Pos() helper presents those fields as a Vector3 reference. That is an explicit local overlay, not a reconstructed shared Vector3 position member.
The independent review of head 969d498 measured a stack Vector3 alternative: Behavior grew from 92 to 116 bytes and emitted a four-byte Vector3 destructor. This supports retaining the measured boundary while investigating the shared layout; it does not prove every alternative impossible. The current review will attach fresh scoped probe identities in its handoff.
Follow-through owned by @andrewboudreau through the active humanizer/integrator work:
Related PR #2513 is the open Lava cleanup proposing dActor_c::Pos(); it is not a merged shared API or this issue. PR #2488 remains the SCRE source repair. Current byte matching/promotion does not close this follow-up.