More issues with nodes

[mellox]

Node issues (running list):

1. Snap point on node not centered. ✅ Fixed (1.2.0) — rock now centered on the node.
2. Collision box on nodes are rising up. Yup, learn to jump at times. ✅ Fixed (1.2.0) — nodes are solid like vanilla; the rock no longer protrudes oddly and miner placement was reworked (Mk1 sits naturally, portable miners sit on the rock surface).
3. Shuffled node coloring a flat color. ✅ Fixed (1.2.0) — proper two-tone (outer + core) coloring restored.

---

Known issues (to look at later):

4. Third-party node-manager mods list/ping original node locations after a shuffle or re-roll. Example: the ResourceNodesManager mod still shows the old vanilla locations (e.g. the 3 vanilla oil spots ping). NodeShuffle hides original nodes rather than destroying them (so the world stays save-safe and restorable if the mod is disabled), and such mods scan every resource-node actor — including hidden ones — so they still see the originals. The vanilla map / compass / handheld scanner are clean. Possible future options: a "fully remove originals" toggle (destroys instead of hides; loses restore-on-disable), or documenting it as a compatibility note.

[mellox]

Plan & technical context (captured for later resume)

Agreed working model (from discussion):

• Phase A — parallel diagnosers: 3 read-only investigator agents, one per issue. No builds, no edits — each locates the exact code, root-causes, and proposes a fix. Read-only = no conflicts, truly parallel.
• Phase B — coordinated fixes: lead applies fixes on a single shared branch fix/node-issues (issues 1 & 2 edit the same UseBox code, so one branch avoids merge conflicts). Build cadence situational (batch where it amortizes the serial build+test; split when isolating a failure).
• Every code change gets a fresh COLD code-review pass before any in-game test by the user.
• All fixes gated behind an EXPERIMENTAL toggle so stable 1.1.1 behavior stays available — user can flip it off and play stable if taking a break. (Likely a new config bool, e.g. EnableExperimentalNodeFixes, separate from the existing oil-shuffle EnableExperimentalFeatures.)
• Agents persistent (cloud AI-resource limits were the prior failure mode; keep them resumable).
• Reality: build+test is a single serial resource (one SatisfactoryModLoader\Mods\NodeShuffle build copy, one deployed DLL, human-in-the-loop in-game testing). Worktrees isolate source only — they do not parallelize builds or testing.

The three issues + initial diagnosis pointers (hypotheses to confirm)

1. Snap point on node not centered — the Miner snaps but off-center. Likely the node actor origin vs the visual rock center: the rock is offset via DressRock(... RelativeOffset) while the snap uses the node origin / IFGExtractableResourceInterface::GetPlacementLocation. Look at ANodeShuffleResourceNode (UseBox/origin) + the placement-location path.

2. Collision box on nodes rising up (player bumps it / has to jump) — the interaction box UseBox = InitBoxExtent(650,650,180) with the \"Resource\" collision profile in NodeShuffleResourceNode.cpp ctor (re-asserted by the subsystem's EnsureNodeUseBox). 180 half-height ≈ 3.6 m tall box that likely blocks the pawn. Fix direction: lower/sink the box Z, or make it pawn-passthrough (the rock mesh already uses ResourceNoCollision). Issues 1 & 2 both touch UseBox → same files → do together.

3. Shuffled node coloring is a flat color — ONLY on shuffled nodes (confirmed by user screenshot: a relocated Iron node renders a flat-ish color vs the richer vanilla look). This is the known per-slot-materials gap: node rocks are MULTI-SLOT (outer slabs + Middle/core); correct color = per-slot node materials, NOT the deposit material. See the prior consolidated analysis in memory `nodeshuffle-descriptor-visuals`. Look at DressRock + the subsystem's ResolveNodeMaterials / SpawnVisualRockForNode. Separate from 1 & 2.

Build / infra notes to resume

• Two source trees: edit C:\Claude\Projects\NodeShuffle\Source (the repo) then sync → C:\Claude\Projects\SatisfactoryModLoader\Mods\NodeShuffle\ before building, or UBT reports "Target is up to date" and ships stale. Verify the embedded build marker (NodeShuffle.cpp) in the deployed DLL after every build.
• Dev build: RunUAT PackagePlugin -DLCName=NodeShuffle (single target, fast). Release: skills\satisfactory-modding\scripts\package_release.ps1 -ModName NodeShuffle (merged multi-target).
• Diagnostics: flip Enable Diagnostic Logging (Mods → Node Shuffle, default OFF) for verbose HOLOGRAMHOOK/placement logs; game log at %LOCALAPPDATA%\FactoryGame\Saved\Logs\FactoryGame.log.
• Current baseline: 1.1.1 on master (commit pushed). Branch off master for this work.

Parked while we look at something else the user found first; resume from Phase A (spawn the 3 diagnosers).

[mellox]

Items 1–3 are resolved in the upcoming 1.2.0 release (rock centering, collision/solidity rework, and two-tone coloring). Added item 4 as a known third-party-mod compatibility issue to revisit later.

[mellox]

Known issue #4 (third-party node-manager mods listing original locations) also affects modded/"dirty" (esc_) nodes and culled vanilla nodes — not just oil. Confirmed via log: the shuffled-out original is hidden + deregistered from the scanner on our side (vanilla map/scanner are clean), but ResourceNodesManager reads node actors directly and still shows the original spot. Same root cause as the oil case; we hide (not destroy) originals for save-safety and because re-roll un-hides them to rebuild the pool, so a clean fix isn't a simple toggle.

[mellox]

Fixed in 1.2.1:

• Nodes no longer spawn on top of machinery (they avoid machines/walls/belts within ~6 m; spawning on foundations/ramps is allowed).
• Nodes no longer spawn sealed inside a rock column/slot with no way to reach them (real caves, ravines, and cliff-bases are unaffected).
• Crash fix: placing a fracking-type well extractor (Resource Well Pressurizer / satellite Resource Well Extractor, including modded wells that derive from them) on a shuffled node no longer crashes — it now shows invalid placement, like vanilla on a non-fracking node.

(Item 4 — third-party node-manager mods listing original locations — remains a known issue.)