The second get implementation in sx_runtime.ml (used by transpiled code)
was still raising on type mismatches. Now returns nil like sx_primitives.
Remove per-call [vm-call-closure] FAIL logging — the jit-hook already
logs failures at the right level. Reduces 70K log lines to ~5.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace continuation-based scope frames with hashtable stacks for all
scope operations. The CEK evaluator's scope/provide/context/emit!/emitted
now use scope-push!/pop!/peek/emit! primitives (registered in
sx_primitives table) instead of walking continuation frames.
This eliminates the two-world problem where the aser used hashtable
stacks (scope-push!/pop!) but eval-expr used continuation frames
(ScopeFrame/ScopeAccFrame). Now both paths share the same mechanism.
Benefits:
- scope/context works inside eval-expr calls (e.g. (str ... (context x)))
- O(1) scope lookup vs O(n) continuation walking
- Simpler — no ScopeFrame/ScopeAccFrame/ProvideFrame creation/dispatch
- VM-compiled code and CEK code both see the same scope state
Also registers scope-push!/pop!/peek/emit!/collect!/collected/
clear-collected! as real primitives (sx_primitives table) so the
transpiled evaluator can call them directly.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Compiler fixes:
- Upvalue re-lookup returns own position (uv-index), not parent slot
- Spec: cek-call uses (make-env) not (dict) — OCaml Dict≠Env
- Bootstrap post-processes transpiler Dict→Env for cek_call
VM runtime fixes:
- compile_adapter evaluates constant defines (SPECIAL_FORM_NAMES etc.)
via execute_module instead of wrapping as NativeFn closures
- Native primitives: map-indexed, some, every?
- Nil-safe HO forms: map/filter/for-each/some/every? accept nil as empty
- expand-components? set in kernel env (not just VM globals)
- unwrap_env diagnostic: reports actual type received
sx-page-full command:
- Single OCaml call: aser-slot body + render-to-html shell
- Eliminates two pipe round-trips (was: aser-slot→Python→shell render)
- Shell statics (component_defs, CSS, pages_sx) cached in Python,
injected into kernel once, referenced by symbol in per-request command
- Large blobs use placeholder tokens — Python splices post-render,
pipe transfers ~51KB instead of 2MB
Performance (warm):
- Server total: 0.55s (was ~2s)
- aser-slot VM: 0.3s, shell render: 0.01s, pipe: 0.06s
- kwargs computation: 0.000s (cached)
SX_STANDALONE mode for sx_docs dev (skips fragment fetches).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Transpiler detects dict literals with a "type" string field and emits
CekFrame records instead of Dict(Hashtbl). Maps frame-specific fields
to generic record slots:
cf_type, cf_env, cf_name, cf_body, cf_remaining, cf_f,
cf_args (also evaled), cf_results (also raw-args),
cf_extra (ho-type/scheme/indexed/match-val/current-item/...),
cf_extra2 (emitted/effect-list/first-render)
Runtime get_val handles CekFrame with direct field match — O(1)
field access vs Hashtbl.find.
Bootstrapper: skip stdlib.sx entirely (already OCaml primitives).
Result: 29 CekFrame + 2 CekState = 31 record types, only 8
Hashtbl.create remaining (effect-annotations, empty dicts).
Benchmark (200 divs): 2.94s → 1.71s (1.7x speedup from baseline).
Real pages: ~same as CekState-only (frames are <20% of allocations;
states dominate at 199K/page).
Foundation for JIT: record-based value representation enables
typed compilation — JIT can emit direct field access instead of
hash table lookups.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Transpiler (transpiler.sx): detects CEK state dict literals (5 fields:
control/env/kont/phase/value) and emits CekState OCaml record instead
of Dict(Hashtbl). Eliminates 200K Hashtbl allocations per page.
Bootstrapper: skip stdlib.sx (functions already registered as OCaml
primitives). Only transpile evaluator.sx.
Runtime: get_val handles CekState with direct field access. type_of
returns "dict" for CekState (backward compat).
Profiling results (root cause of slowness):
Pure eval: OCaml 1.6x FASTER than Python (expected)
Aser: OCaml 28x SLOWER than Python (unexpected!)
Root cause: Python has a native optimized aser. OCaml runs the SX
adapter-sx.sx through the CEK machine — each aserCall is ~50 CEK
steps with closures, scope operations, string building.
Fix needed: native OCaml aser (like Python's), not SX adapter
through CEK machine.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
SX-to-OCaml transpiler (transpiler.sx) generates sx_ref.ml (~90KB, ~135
mutually recursive functions) from the spec evaluator. Foundation tests
all pass: parser, primitives, env operations, type system.
Key design decisions:
- Env variant added to value type for CEK state dict storage
- Continuation carries optional data dict for captured frames
- Dynamic var tracking distinguishes OCaml fn calls from SX value dispatch
- Single let rec...and block for forward references between all defines
- Unused ref pre-declarations eliminated via let-bound name detection
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
