- Server stop defers teardown only while the plugin class loader close is held
- Shutdown with no Iris world loaded tears down during plugin disable
- Pre-unload hook drains generators inline when the class loader close cannot be held
- INMSBinding reports whether the class loader close is held
- Shutdown tests cover the held and unheld pre-unload paths
- Shutdown waits for an in-progress world replacement to finish staging
- Replacements started after shutdown begins are refused
- Admission tests cover the shutdown wait and the refusal
- Merge the unified Studio interface and command entry points.
- Adapt water views and provisional sampling to basin hydrology.
- Restore diagnostic estimates and the biome water overlay.
- Fix render cancellation, preview closure, and platform position callbacks.
- Fall back to software rendering when native bindings are unavailable.
- Cover renderer, lifecycle, and scheduler regressions.
- The route-and-reject hydrology system and its probes are removed
- Hydrology is configured under dimension hydrology and per biome, region and dimension hydrologyPolicy
- riverPolicy and the old hydrology config types are gone
- Grid kernels cover priority flood, D8 routing, flow accumulation, basin extraction, breaching and stream burning
- A basin spills at its pour point and an interior bump can never be the spill
- Region plans cover 4096-block regions on a 64-block grid for major rivers and lakes
- Window plans cover 512-block windows on an 8-block grid for reaches and ponds
- Plans are cached in memory, persisted per world under iris/hydrology and keyed by pack, seed, world height and settings
- Only the main runtime of an engine persists plans and stores swap on hotload and default runtime change
- Demand rasters outrank prefetch and stopping pregen withdraws its queued prefetch
- Ponds come from natural basins, seeded cups and chained cups with strictly falling levels
- A pond level never exceeds its rim plus maxFill or the end grade of any inflow
- Pond outlet loops are opened at their lowest pond
- Reaches split at junctions, pond outlets and 256-block grid lines
- Reaches meander, fit the valley, grade to a running minimum and smooth with Chaikin passes
- Drops of two blocks or more get a lip and one-block corner steps become dry rim ledges
- Cascades and waterfalls write falling fluid from the mantle at level eight
- Inflows into region lakes end at the lake level
- River mouths and water beside the sea sit at the sea surface
- The per-chunk raster gives terrain height, water head, role, tier, fluid, cave ceiling and shore distance
- WaterRole codes are stable and exposed to packs through the WATER_ROLE engine stream
- WATER_HEAD replaces the old river engine streams and an unknown stream name fails the pack load
- Banks take SMOOTH, LINEAR, CONCAVE, TERRACED or CLIFF profiles
- Containment berms grow as tall as the water needs
- Columns within two blocks of water stay solid and keep no Terrain3D overhang
- Seam dams hold water where new chunks meet chunks generated earlier
- The transition blend skips hydrology-owned columns
- Caves stay at least two blocks below river and pond beds
- The painter is the only writer of still water
- Objects cannot open water, falls or bank ground and solid objects may stand in water
- On-water objects need fluid at their anchor column
- Floating child islands stay out of channels
- Sea inlets run inland from cliff coasts at sea level and end in a chamber pool
- Inlet passages are sealed against carvers, subterrain features and the post pass
- Cave pools settle into carved lows, keep one fluid and stay out of inlet space
- Pool and inlet cells use their own cave actions in mantle data
- Water vegetation lists sit on tiers, ponds and hydrologyPolicy with SEA_FLOOR, SEA_SURFACE and SHORE_LINE parts
- Pack validation compiles hydrology settings and rejects empty or water-holding palettes
- Chunk generation fails closed when its 3x3 water rasters are not planned
- /iris find covers stream, river, major river, cascade, waterfall, mouth, pond, lake, dam, sea inlet and cave pool
- Generation history records the new feature types and drops old hydrology entries
- The studio map draws water from rasters up close and from plans when zoomed out
- Pregen prepares the current region and one region ahead
- hydrologyProbe, riverTransectProbe, hydrologyBlockProbe, seaInletProbe and cavePoolProbe report containment, rises, seams, tick conversion and missing surface
- Test forks run with a 2 GB heap, a 60-minute timeout and the Mockito agent loaded explicitly
- Generation with hydrology off is unchanged
- Compact connected roots extend bounded gaps into supporting terrain
- Unsupported cliff placements leave no partial trees
- Root and canopy roles follow object scaling and rotation
- Canopies preserve solid trunk cells across chunk replay and saved continuations
- Branch wood replaces leaf clusters within each tree
- Regression tests cover slopes and overlapping tree placements
- Gameplay checks compare grounded spindle geometry across both packs
- Chest activation and diamond withdrawal stop after ten seconds
- Partial chest evidence records the active phase and window layout
- Folia reports its unavailable reload command without changing pack files
- Gameplay opens normal and copper chests before inspecting loot
- Map markers and saved centers match verified structure targets
- Missing and disabled destinations retain unrelated chest loot
- Storage variants, insertion rollback, region ownership, and resource reloads have runtime assertions
- Separate data packs cover Minecraft 26.2 and 26.3 loot formats
- Pending biome reads no longer cancel block breaking
- Custom loot resolves saved biome history after accepted breaks
- Native drops survive failed reads and runtime teardown without duplication
- Tree felling captures the original tool for fallback drops
- Modded block loot resolves one historical biome context
- Timeout settings persist atomically through the actual server configuration paths
- Tests cover deferred loot, teardown and timeout persistence
- Scope cached generation fields to their seed and context.
- Warm decorators and ores with their canonical generation seeds.
- Stabilize object traversal and procedural placement choices.
- Isolate smart-bore variants from shared source objects.
- Pin noise and full-engine output across generation orders.
- Verify same-jar generation through cold server restarts.
- Test that capture drops generation timings once generation is idle
- Test that capture carries the engine timing averages while fresh
- Remove the hand-built idle snapshot test that passed without the gate
- Publish in-flight pregen chunk requests as iris.pregen-queue
- Publish chunks left in the job as iris.pregen-remaining
- Mark generation timings unavailable after 10 seconds without generation
- Bootstrap audits vanilla slots that hold an Iris world
- Startup stops when such a slot has unusable history
- Startup also stops when bukkit.yml no longer binds such a slot to Iris
- The audit detail carries the history read cause
- File system failures are named by type
- Bootstrap failures print the refusal banner instead of a bare stack trace
- A vanilla slot holding an Iris world is told to restore its history and keep its binding
- The vanilla slot remedy adds deleting level.dat when the level was never created
- A failed enable locks only for bukkit.yml worlds the server loads
- Modded logins are refused only when this save's dimension registry lists the primary world
- Otherwise Iris warns once and leaves routing idle
- An unreadable dimension registry counts as expecting the primary world
- A vanilla slot bound to Iris without Iris storage points at the bukkit.yml entry and /iris replace
- The refusal consequence comes from whether startup worlds exist instead of the first tick
- The first WorldLoadEvent or a hotload marks startup worlds as existing
- A failed enable on a hotload says the server keeps running
- Reconciler resolution failures report the refusal banner
- Operators get a join notice for such failures until the world loads
- The fail-closed generator is built even when the cause message cannot be read
- Tests drive Iris.onEnable for the lock, the unsupported-server lock and the rethrow
- The failed-enable teardown decision lives in FailedEnableLock with a test
- Every Iris world the Bukkit resolver cannot serve gets a fail-closed generator
- CraftServer no longer receives a throw that it turns into vanilla
- A refusal banner logs the world, key, cause chain and what happens next
- The reconciler uses a throwing resolution so a broken world fails without stopping the server
- A failed enable on a server with Iris worlds keeps Iris enabled with a locked runtime
- bukkit.yml bindings in vanilla slots count as Iris worlds for that lock
- The shutdown-based vanilla fallback guard is gone
- A recorded modded bind failure rethrows instead of rebinding on every chunk call
- The first modded bind failure logs a banner
- Persistent-dimension reinject failures log a banner and repeat to operators on join
- Modded logins are refused while the configured primary Iris world is missing
- Startup holds generation admission through recovery and runtime attachment
- Prepared startup state removes the second recovery scan before generator readiness
- Recovery batches allocated chunks by region with reusable snapshot buffers
- Region decoding uses up to four workers within processor and heap limits
- Oversized regions and external chunks use bounded direct reads
- Status parsing validates chunk coordinates without constructing the full NBT tree
- Region snapshots reject changed files, allocation headers and truncated payloads
- Native zlib decompression uses jlibdeflate with a Java inflater fallback
- Chunk readers reuse region handles and decompression buffers during recovery and boundary capture
- Semantic shards reuse validated summaries with invalidation on mutation
- Semantic key validation avoids regular expression allocation
- Saved biome indexes use buffered sequential reads with reusable CRC state
- Pack inspection reads validated manifest and pack state without opening terrain indexes
- Datapack compilation reuses pack inspection within each operation
- Kernel staging avoids duplicate pack hashing when no activation is pending
- Pack copying and directory synchronization use up to four available processors
- Boundary publication batches directory synchronization around durable catalog publication
- Configured world reconciliation reuses ready runtimes after identity and saved-state checks
- Explicit world loads retain the full validation path
- Regression tests cover startup admission, recovery, malformed data, resource limits, publication and reconciliation
- Packages regrouped by domain: generation, world, pack, studio, structure, platform, command, localization, diagnostics
- Forked VolmLib utilities dropped for the shared library: noise, stream, interpolation, hunk, matter
- Engine-coupled stream decorators moved to generation.stream.GenerationStreams
- Hunk section splits now run on the Iris burst pool through HunkBurst
- Artifact entry lists, purity allowlists and shrink rules follow the new paths
- Debug dump replaces the root debugdump command
- Debug toggle controls diagnostic logging
- Debug version uses the shared version display
- Root version remains executable but hidden from help
- Diagnostic dump routing retains its separate permission path
- Translations and tests cover the revised command structure
- Move each jigsaw subgroup executor into its own top-level command class
- Move the adoption plan store, handlers and helpers into CommandJigsawAdopt
- Move the vanilla export lease and reporting helpers into CommandJigsawExports
- Move the studio open lifecycle into CommandJigsawOpen
- Point the jigsaw contract test at the relocated command types
- Exclude the new jigsaw command sources from the generation revision scope
- Move art.arcane.iris.testsupport into core/src/testFixtures and apply java-test-fixtures to core
- Consume core test fixtures from the bukkit plugin, nms, probe and the three loader adapters
- Delete the duplicate ProjectPaths copies in the bukkit-plugin and modded-common test trees
- Point ModdedMessageContributorTest at the shared ProjectPaths
- Guard CommandPackCompatTest, NativeTransitionColumnTest and StubPlatformStateTest with PlatformLeakGuard
- Release the probe runtime state after DimensionStackSuperflatProbeTest opens a workspace
- Keep core main classes ahead of the core jar on the test classpath so the bytecode gates still scan a directory
- Require the core source tree alongside the settings file when walking up for the repository root, so modded-common tests built inside a loader adapter no longer stop at that adapter's own settings file
- Add the ProjectPaths test helper that derives the module root from the test class code source and the repository root from the settings file above it
- Replace the working-directory-relative source and resource literals in thirty-five test classes with module and repository lookups
- Drop the bespoke source-root walk in PregenMantleCleanupContractTest
- Delete 22 test classes whose assertions only read production .java text and check call ordering or literal presence
- Remove 77 source-text assertion methods from 34 test classes that keep real behavioural coverage
- Replace the core process-stream source grep with an ASM gate over compiled core classes in CoreProcessStreamPurityGateTest
- Replace the packs-root source grep in PacksFolderContractTest with an ASM call-site gate over compiled core classes
- Replace the IrisData registered-root regex in PackTypeBukkitPurityGateTest with an ASM scan of the registerLoader class constants
- Trim six mixed tests down to their behavioural assertions and drop their source-text tails
- Drop the eight iris.*Source and three iris.moddedCommonSources test system properties now that nothing reads them
- Removed the core processResources skip that fired on test-only invocations, which packaged a core jar with no generation kernel catalog and made all 10 probe and bukkit-plugin tests that reach GenerationKernelRegistry fail on every ./gradlew test
- Set maxParallelForks on :core:test to half the available processors, overridable with -PtestForks, and left the small modules single-forked so they do not re-pay the Minecraft bootstrap per fork or steal worker leases from core
- Resolved B.registries() per bound platform instead of latching the first one for the JVM lifetime, keeping the previous unbound behaviour, so a test that binds its own platform no longer reads a foreign platform's block states
- Resolved IrisObject.States per bound platform through accessors instead of static finals fixed at class-init, removing the NoClassDefFoundError and null material-key failures that appeared whenever class partitioning changed
- Bound an explicit platform in EngineMantleCoverageTest and IrisDimensionStackActuatorMetadataTest so they no longer depend on another class leaking a binding first
- Gated BukkitSpiConformanceTest registry probes on a live RegistryAccess and restored the IrisPlatforms binding it leaked, which was silently swallowing the stderr two SettingsHotloadWatchTest assertions read
- Initialised org.bukkit.Registry and Biome inside the RegistryAccess mock scope in ObjectStudioGeometryTest so Mockito no longer instruments Registry mid-clinit
- Replaced the SettingsHotloadWatchTest fixed 1200ms diagnostic window with a 10ms condition poll bounded at 4s
- Replaced the IrisCaveCarver3DNearParityTest wall-clock speedup benchmark with a deterministic density-sample count, cutting 28 carves to 2
- Cut HydrologyPlannerTest cold-owner parity from 12 tile plans to 4 and shared the immutable baseline plan across the early-owner tests
- Built one prototype project or staged history per class in the jigsaw studio and world replacement tests and copied it per test, replacing repeated fsync-bound fixture construction
- Replaced Mockito block-state fixtures with concrete implementations in the object IO, deposit modifier and nearest-block-index tests, removing about 2.5 million interceptions
- Replaced Thread.sleep with bounded condition polling in HydrologyTileCacheTest
- Deleted 19 test classes and 60 further test methods that asserted on production .java source text rather than behaviour, and removed the 19 now-unread iris.*Source system properties that fed them
- Server shutdown advances native chunk tasks while waiting for pregeneration
- Native world close owns chunk saves during server shutdown
- Pending plugin flushes stop waiting when native shutdown begins
- Late eviction completion cannot schedule another shutdown flush
- Regression tests cover owner waits and shutdown save ownership
