The grid's ocean_ruins claim at chunk (7,5) matches a fresh vanilla
capture: one cold small brick_2 piece, COUNTERCLOCKWISE_90, template
position (112,50,80), integrity 0.8. Notes record the addPieces draw
order, the template arrays, and the two open mechanics for the port
(piece descent from y=90 to the ocean floor, integrity draw order).
The portal's air pocket is placed for real now: vanilla's fluid ticks
then flood every pocket cell connected to surrounding water while
sealed cells stay air, so the piece ends with a bounded BFS over its
own air cells from water neighbours. BlockAgeProcessor gains the
stairs/slab branches with their property-preserving mossy variants,
and horizontal facings rotate with the piece (mirror first, then
quarter turns).
Portal box match against the fixture jumps from 52.7% to 84.1%;
region parity rises to 98.047% with biomes and heightmaps exact.
Remaining box diffs trace to drip columns reading the shared decoration
stream, which is approximated per-cell.
PlaceRuinedPortalPiece runs the template through the decoded processor
stack — positional Legacy streams per cell drive the gold-to-air,
lava-to-magma, netherrack-to-magma and block-age mossiness rolls, the
features-cannot-replace guard skips protected cells, and template lava
stays when it lands in existing lava. Drip columns below the portal
draw from a documented approximation of the shared decoration stream;
spreadNetherrack is not needed for the underground setups.
The placement chain is now proven end to end: chunk (1,0) reproduces
vanilla's saved start (portal_6, CLOCKWISE_90, NONE, air pocket at
(16,12,0)) and the fixture parity ticks up to 98.033% with biomes and
heightmaps still exact. generatorVersion bumps to 23.
Three fixes close the loop against the captured region file: the giant
roll precedes the template draw, surfaceY is the topmost non-air block
(getHeight already returns one past it), and structure biome refs
resolve through has_structure/ tags to the extracted tag files. The
weighted variant pick walks a fresh per-attempt GenerationContext
stream while failed entries drop from the pick stream without reseeding.
TestRuinedPortalStubMatchesVanillaStart pins chunk (1,0) to vanilla's
own saved start: portal_6, CLOCKWISE_90, mirror NONE, air pocket,
template position (16,12,0) — and asserts the neighbouring fixture
chunks stay portal-free.
generateBaseTerrain now factors the noise+surface stage out of chunk
generation so structure placement can read pre-carve heights exactly
like vanilla's STRUCTURE_STARTS stage. RuinedPortalGenerationPoint
replays the ruined_portals set: the weighted pick across all seven
variants on its own stream, each attempt's setup/template/rotation/
mirror draws on a fresh GenerationContext stream (makeRandom reseeds
per attempt, confirmed from bytecode), findSuitableY with corner-column
settle scan against those heights, and the post-draw 3D-biome filter at
quart-snapped coordinates.
Known gap, documented in STRUCTURE_NOTES.md: no variant is accepted in
the window around the fixture's portal yet while vanilla placed one;
the biome/surface inputs need a vanilla-side probe to bisect. Not wired
into the production replay until that closes.
LoadResolvedTemplate parses the extracted .nbt files with every palette
entry resolved to concrete state IDs through an injected resolver, and
TransformBlockPos reproduces StructureTemplate.transform's mirror-then-
rotate order around the pivot, including all three quarter-turn cases
read off the bytecode. MthGetSeed gives processors their per-block
positional Legacy stream. Verified against the real ruined_portal
templates: every block lands inside the declared size for all eight
mirror/rotation combinations.
Captures the decoded placement grid, the four legacy frequency reducers,
the full ruined-portal findGenerationPoint and findSuitableY algorithms
with their exact draw order, the template name arrays, and the fixture
marker coordinates that pin the grid to vanilla. This is the working map
for the piece-placement ports that follow.
Loads the 26.1.2 structure datapack: 20 structure sets, 34 structure
definitions, and the worldgen biome tags they reference. Random-spread
placement follows RandomSpreadStructurePlacement byte for byte — region
floorDiv by spacing, one Legacy stream salted per region, two
spread-type draws for the offset — and the four legacy frequency
reducers keep their exact draws, including the hardcoded buried-treasure
salt and the outpost's discarded full-range int.
Verified against the fixture: on seed 12345 the ruined_portals set
claims exactly chunk (1,0), the chunk whose captured blocks carry the
ruined portal's obsidian, gold, and crying obsidian. TestStructure-
PlacementGrid pins that plus village separation invariants.
Ports MonsterRoomFeature from bytecode with vanilla draw order and loop
structure: shell validation (solid floor and ceiling, one to five open
side columns), interior cave_air carving, cobblestone walls with the
mossy three-in-four floor roll drawn only where vanilla draws it, chest
placement faced away from its single opaque neighbour with the loot-seed
long consumed even though block entities are not modelled yet, and the
spawner's one-draw mob pick. Writes go through Feature.safeSetBlock
semantics against #minecraft:features_cannot_replace, captured verbatim
from the jar into the embedded datapack data.
The pass slots between geodes and the ores so the rooms' cave_air
pockets exist before ore ellipsoids roll their air-exposure discards.
The committed fixture's four chunks happen to contain no dungeons, so
measured parity holds at 98.028%, but the generator's output can differ
elsewhere: generatorVersion bumps to 22.
All three jobs move to [self-hosted, windows]. The race job now adds the
machine-wide MSYS2 mingw64 toolchain to GITHUB_PATH: the runner service
account only sees the machine PATH, which has no C compiler, so -race
could not build its cgo runtime. The dead verify.yml.selfrun file goes
away — GitHub never reads that suffix and its content already lives
here.
The featureless capture settles what the 98.028% gap is made of: base
terrain is bit-exact, so the drift comes from air-writing features
(monster rooms, geode voids, lava lakes) that vanilla runs before the
ores and the region replay does not cover yet. Both files now point at
monster rooms as the next step instead of at carver and aquifer edges.
TestVanillaBaseTerrainParity compares our noise+surface+carver+aquifer+
vein pipeline against a vanilla capture whose biomes carry no features
and no structure sets: all 393,216 cells match. The plain-state fixture
mismatches the base-terrain diagnostic used to attribute to carver or
aquifer defects are therefore feature outputs wearing plain states —
monster-room cave_air, geode voids, lava-lake lava — and its doc comment
now says so.
-featureless derives a world datapack from the server jar (inner server
or bundler layout): every biome keeps its carvers but loses its feature
stages and every structure set loses its structures, so the capture is
pure noise terrain with surface rules, carvers, aquifers, and veins.
-blocks-only writes a blocks-only RIOBASE1 fixture for it. -port avoids
colliding with something else already listening on 25565.
CLAUDE.md still claimed there was no placed/configured feature system and
that ores never generate below y=0; both stopped being true when the
stage-6 region replay landed. Document what the replay covers, which
decoration stays hand-written, the measured parity baseline, and the new
env-gated base-terrain diagnostic. README gains the same correction plus
the 966-cell base-terrain measurement.
Splits fixture mismatches into classes that name the responsible
subsystem: plain-vs-plain cells are direct base-terrain defects and come
with sample coordinates, single-side feature states count as flips per
family, and disk-shaped outputs land in an ambiguous bucket so they
cannot pollute the plain signal.
First run on seed 12345: 966 base cells (0.246%), dominated by vanilla-
carved cave segments missing from our carver output and deepslate-to-
water aquifer edges.
The combined stage-6 pass in placeScheduledUndergroundOresStage already
replays ores, underwater magma, and disks in datapack order; the separate
placeScheduledDisks and placeScheduledUnderwaterMagma passes have no
callers left. Keep the per-position helpers the combined pass and tests
use.
