Robert Allan JamesandClaude Sonnet 5
8a2ee0fdad
sk_hermes_publish() now enforces SK_HERMES_CHUNK_MAX_PAYLOAD (1024) on every message's payload_len uniformly, chunked or not -- closing the gap task 3.3 explicitly parked. A chunk carrier is [SkHermesChunkHeader][content slice], slice capped at 1024 - sizeof(header) rather than 1024 itself, so every message on the wire satisfies the same one-block bound vm_interpret()'s own drain limit already requires -- a future chunk-aware drain never has to special-case a carrier that can't be handed to vm_interpret() as-is. Deliberately no chunking-sender API: building one would need kernel-Hermes to own chunk-buffer memory with a real lifetime it has no way to track (kept alive until every subscriber drains it). Sending is a loop pattern a caller writes with sk_hermes_chunk_count() + sk_hermes_publish(), demonstrated by this task's own self-test. sk_hermes_reassemble() is pure and memory-agnostic: validates msg_id agreement, exact seq coverage, and per-chunk slice sizes before a single memcpy, with the total length computed once and checked against the caller's buffer once -- never order-dependent on which chunk happens to overflow. Verified live on all three architectures: a 1024-byte payload as one message, a 1025-byte send refused outright with the ledger untouched, and a 3000-byte payload split into 3 chunks, drained, and reassembled byte-exact against the original. dict_hash unmoved and identical across architectures. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
include/
Public headers for the hosted StarForth VM (src/). Kernel-only headers
live under include/starkernel/.
Core VM
vm.h— theVMstruct and all core types (DictEntry,DictPhysics, stacks, dictionary state); the central header most other files include.vm_api.h— external VM API surface.vm_host.h,vm_debug.h,startup.h,version.h— host integration, debug utilities, startup sequencing, version string.cli.h,repl.h,io.h,log.h— CLI parsing, REPL loop, I/O, logging.word_registry.h— word registration system shared by everysrc/word_source/*.cfile.compudynamics.h— the generic compudynamics module: tuning-word/config lookups (cd_tuning_word(),cd_tuning_vm()) consumed byssm_jacquard.c.
Memory / blocks
memory_management.h,dictionary_management.h— dictionary allocator and search.block_subsystem.h,blkcfg.h,blkio.h,blkio_factory.h— logical→ physical block mapper and pluggable block I/O backends (file/RAM).platform_alloc.h,platform_lock.h,platform_time.h— platform abstraction shims (hosted vs. kernel allocation/locking/timing).
Physics-driven adaptive runtime (7 feedback loops)
physics_runtime.h— main physics coordinator.physics_hotwords_cache.h— Loop #1, execution-heat hot-words cache.physics_metadata.h— per-wordDictPhysicsmetadata tracking.physics_pipelining_metrics.h— Loop #4, word-transition prediction.physics_execution_hooks.h— execution instrumentation hook points.rolling_window_of_truth.h,rolling_window_knobs.h— Loop #2 circular execution-history buffer and its tuning knobs.inference_engine.h— Loops #5/#6, window-width and decay-slope statistical inference.dictionary_heat_optimization.h— Loop #1 heat counters.ssm_jacquard.h— L8 Jacquard steady-state mode selector.doe_metrics.h— Design of Experiments (2^7 factorial) metrics.profiler.h— performance profiling hooks.
Arithmetic / codegen
q48_16.h— Q48.16 deterministic fixed-point arithmetic (used instead of IEEE-754 float specifically to keep cross-architecture behavior bit-identical).math_portable.h— portable math helpers.arch_detect.h,starforth_config.h— architecture detection and the build-flag fallback-default layer (used when a.cfile is compiled by hand withoutmake).vm_asm_opt.h,vm_asm_opt_arm64.h,vm_asm_opt_riscv64.h,vm_inner_interp_asm.h,vm_inner_interp_arm64.h,vm_inner_interp_riscv64.h— per-architecture assembler-optimized inner interpreter (USE_ASM_OPT=1).
See include/starkernel/README.md for the bare-metal kernel headers.