Replace the monolithic 500KB <script data-components> block with a 25KB
JSON manifest mapping names to content hashes. Every definition —
components, islands, macros, client libraries, bytecode modules, and
WASM binaries — is now content-addressed and loaded on demand.
Server (sx_server.ml):
- build_hash_index: Merkle DAG over all definitions — topological sort,
hash leaves first, component refs become @h:{hash} in instantiated form
- /sx/h/{hash} endpoint: serves definitions with Cache-Control: immutable
- Per-page manifest in <script data-sx-manifest> with defs + modules + boot
- Client library .sx files hashed as whole units (tw.sx, tw-layout.sx, etc.)
- .sxbc modules and WASM kernel hashed individually
Browser (sx-platform.js):
- Content-addressed boot: inline script loads kernel + platform by hash
- loadDefinitionByHash: recursive dep resolution with @h: rewriting
- resolveHash: 3-tier cache (memory → localStorage → fetch /sx/h/{hash})
- __resolve-symbol extended for manifest-based component + library loading
- Cache API wrapper intercepts .wasm fetches for offline caching
- Eager pre-loading of plain symbol deps for CEK evaluator compatibility
Shell template (shell.sx):
- Monolithic <script data-components> removed
- data-sx-manifest script with full hash manifest
- Inline bootstrap replaces <script src="...?v="> with CID-based loading
Second visit loads zero bytes from network. Changed content gets a new
hash — only that item refetched (Merkle propagation).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root cause: sx_insert_near placed break-lines-greedy, pretext-position-line,
pretext-layout-lines OUTSIDE the define-library begin block. The bytecode
compiler only compiles forms inside begin as STORE_GLOBAL — forms outside
are invisible to the browser VM.
Fix: moved all function definitions inside (begin ...) of (define-library).
Bytecode now includes all 17 functions (11K compiled, was 9K).
Browser load-sxbc: simplified VmSuspended handling — just catch and
continue, since STORE_GLOBAL ops already ran before the import OP_PERFORM.
sync_vm_to_env copies them to global_env.
Island now calls break-lines and pretext-layout-lines from bytecode-compiled
library — runs on VM, not CEK interpreter.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- text-layout.sx added to WASM bytecode pipeline (9K compiled)
- Fix multi-list map calls (map-indexed + nth instead of map fn list1 list2)
- pretext-layout-lines and pretext-position-line moved to library exports
- Browser load-sxbc: handle VmSuspended for import, copy library exports
to global_env after module load (define-library export fix)
- compile-modules.js: text-layout in SOURCE_MAP, FILES, and entry deps
- Island uses library functions (break-lines, pretext-layout-lines)
instead of inlining — runs on bytecode VM when exports resolve
Known issue: define-library exports don't propagate to browser global env
yet. The load-sxbc import suspension handler resumes correctly but
bind_import_set doesn't fire. Needs deeper investigation into how the
WASM kernel's define-library registers exports vs how other libraries
(adapter-html, tw) make their exports available.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The previous K.eval() approach double-escaped backslashes in SX source
strings, breaking the \/ → / unescaping that the server serializer adds
for HTML safety. Using K.callFn() passes strings directly as arguments,
bypassing the escaping problem entirely.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
compile-let now handles dict destructuring patterns:
(let {:key1 var1 :key2 var2} source body). This unblocked core-signals.sx
(deref uses dict destructuring) which was the sole bytecode skip.
Rewrote stripLibraryWrapper from line-based to paren-aware extraction.
The old regex missed (define-library on its own line (no trailing space),
silently passing the full wrapper to the compiler.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The value_to_js resume handler was calling _driveAsync on re-suspension,
but the JS driveAsync caller also processes the returned suspension.
This caused the second wait in each iteration to fire immediately (0ms)
instead of respecting the delay.
Fix: resume handler just returns the suspension object, lets the JS
driveAsync handle scheduling via setTimeout.
Verified: repeat 3 times add/wait 300ms/remove/wait 300ms produces
6 transitions at correct 300ms intervals (1504ms total).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The resume callback in the value_to_js VmSuspended handler now catches
VmSuspended recursively, building a new suspension object and calling
_driveAsync for each iteration. Fixes repeat N times ... wait ... end
which produces N sequential suspensions.
Bounce works on repeated clicks. 4/4 regression tests pass.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Reverts the source-loading workaround. Bytecode modules go through the
VM which handles IO suspension (perform/wait/fetch) correctly. The
endModuleLoad sync copies VM globals to CEK env, so eval-expr-cek in
hs-handler can find hs-on/hs-toggle-class!/etc.
All three HS examples fully working on live site:
Toggle Color — toggle classes on click
Bounce — add class, wait 1s (IO suspend+resume), remove class
Count Clicks — increment counter, update innerHTML
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- sx-platform.js: add _driveAsync to platform (was sandbox-only) for
driving wait/fetch IO suspension chains in live site
- sx-platform.js: host-callback wrapper calls _driveAsync on callFn result
- sx_browser.ml: value_to_js callable wrapper catches VmSuspended, builds
suspension object, and calls _driveAsync directly
Toggle and count clicks work fully. Bounce adds class but wait/remove
requires IO suspension in CEK context (eval-expr-cek doesn't support
perform — needs VM-path evaluation in hs-handler).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- sx_browser.ml: restore VmSuspended handler in api_call_fn with
make_js_callFn_suspension for IO suspension chains (wait, fetch)
- runtime.sx: delete host-get stub that shadowed platform native —
hs-toggle-class! now uses real FFI host-get for classList access
All three live demo examples work:
Toggle Color — classList.toggle on click
Bounce — add .animate-bounce, wait 1s suspend, remove
Count Clicks — increment @data-count, put into innerHTML
4/4 bytecode regression tests pass (was 0/4 without VmSuspended).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Three bugs fixed:
1. host-get in sx-platform.js: return true for function-valued properties
so dom-get-attr/dom-set-attr guards pass (functions can't cross WASM boundary)
2. hs-runtime.sx: renamed host-get→hs-host-get and dom-query→hs-dom-query to
stop shadowing platform natives when loaded as .sx source
3. compile-modules.js: HS dependency chain (integration→runtime→compiler→parser→tokenizer)
so lazy loading pulls in all deps. Non-library modules load as .sx source
for CEK env visibility.
Result: 8/8 elements activate, hs-on attaches listeners. Click handler needs
IO suspension support (VmSuspended in sx_browser.ml) to fire — next step.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- orchestration.sx: add hs-boot-subtree! call to process-elements
- integration.sx: remove load-library! calls (browser loads via manifest)
- sx_vm.ml: add __resolve-symbol hook to OP_GLOBAL_GET for lazy loading
- compile-modules.js: add HS modules as lazy_deps in manifest
HS compilation works in browser (tokenize→parse→compile verified).
Activation pipeline partially working — hs-activate! needs debugging
(dom-get-data/dom-set-data interaction with WASM host-get on functions).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Reset to last known-good state (908f4f80) where links, stepper, and
islands all work, then recovered all hyperscript implementation,
conformance tests, behavioral tests, Playwright specs, site sandbox,
IO-aware server loading, and upstream test suite from f271c88a.
Excludes runtime changes (VM resolve hook, VmSuspended browser handler,
sx_ref.ml guard recovery) that need careful re-integration.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When cek_call_or_suspend runs a CEK machine for a non-bytecoded Lambda
(e.g. a thunk), _active_vm still pointed to the caller's VM. VmClosure
calls inside the CEK (e.g. hs-wait) would merge their frames with the
caller's VM via call_closure_reuse, causing the VM to skip the CEK's
remaining continuation on resume — producing wrong DOM mutation order
(+active, +active, -active instead of +active, -active, +active).
Fix: swap _active_vm with an empty isolation VM before running the CEK,
restore after. This keeps VmClosure calls on their own frame stack while
preserving js_of_ocaml exception identity (Some path, not None).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Runs test_bytecode_repeat.js as step 6 of build-all.sh.
Currently warns on failure (known bug). Will become a hard
gate once the bytecode when/do/perform fix lands.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- cek_run patched to handle import suspensions via _import_hook.
define-library (import ...) now resolves cleanly on the server.
IO suspension errors: 190 → 0. JIT failures: ~50 → 0.
- _import_hook wired in sx_server.ml to load .sx files on demand.
- compile-modules.js syncs source .sx files to dist/sx/ before
compiling — eliminates stale bytecode from out-of-date copies.
- WASM binary rebuilt with all fixes.
- 2658/2658 tests pass (8 new — previously failing import tests).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When the VM or CEK hits an undefined symbol, it checks a symbol→library
index (built from manifest exports at boot), loads the library that
exports it, and returns the value. Execution continues as if the module
was always loaded. No import statements, no load-library! calls, no
Suspense boundaries — just call the function.
This is the same mechanism as IO suspension for data fetching. The
programmer doesn't distinguish between calling a local function and
calling one that needs its module fetched first. The runtime treats
code as just another resource.
Implementation:
- _symbol_resolve_hook in sx_types.ml — called by env_get_id (CEK path)
and vm_global_get (VM path) when a symbol isn't found
- Symbol→library index built from manifest exports in sx-platform.js
- __resolve-symbol native calls __sxLoadLibrary, module loads, symbol
appears in globals, execution resumes
- compile-modules.js extracts export lists into module-manifest.json
- Playground page demonstrates: (freeze-scope) triggers freeze.sxbc
download transparently on first use
2650/2650 tests pass.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- load-library! native: islands can declare module dependencies at
hydration time, triggering on-demand .sxbc loading
- JIT compiler lazy-load: compiler.sxbc loads via setTimeout after boot,
eliminating "JIT: compiler not loaded" errors
- _import_hook on sx_types: infrastructure for hosts to resolve import
suspensions inside eval_expr (server wiring deferred to Step 8)
- Playground page (/sx/(tools.(playground))): REPL island that lazy-loads
the compiler module when navigated to — demonstrates the full
lazy loading pipeline
Known remaining issues:
- SPA navigation broken for pages using let-match (orchestration.sx,
router.sx) — bytecode compiler doesn't handle let-match special form
- Server-side "IO suspension in non-IO context" during http_load_files —
needs cek_run import handling (deferred to Step 8)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Bytecode compiler now emits OP_PERFORM for (import ...) and compiles
(define-library ...) bodies. The VM stores the import request in
globals["__io_request"] and stops the run loop — no exceptions needed.
vm-execute-module returns a suspension dict, vm-resume-module continues.
Browser: sx_browser.ml detects suspension dicts from execute_module and
returns JS {suspended, op, request, resume} objects. The sx-platform.js
while loop handles cascading suspensions via handleImportSuspension.
13 modules load via .sxbc bytecode in 226ms (manifest-driven), both
islands hydrate, all handlers wired. 2650/2650 tests pass including
6 new vm-import-suspension tests.
Also: consolidated sx-platform-2.js → sx-platform.js, fixed
vm-execute-module missing code-from-value call, fixed bootstrap.py
protocol registry transpiler issues.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The transpiled VM (sx_vm_ref.ml, from lib/vm.sx) is now the ACTIVE
bytecode execution engine. sx_server.ml and sx_browser.ml call
Sx_vm_ref.execute_module instead of Sx_vm.execute_module.
Results:
- OCaml tests: 2644 passed, 0 failed
- WASM tests: 32 passed, 0 failed
- Browser: zero errors, zero warnings, islands hydrate
- Server: pages render, JIT compiles, all routes work
The VM logic now lives in ONE place: lib/vm.sx (SX).
OCaml gets it via transpilation (bootstrap_vm.py).
JS/browser gets it via bytecode compilation (compile-modules.js).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
All 26 browser modules recompiled with define-library/import forms.
Compilation works without vm-compile-adapter (JIT pre-compilation
hangs with library wrappers in some JIT paths — skipped for now,
CEK compilation is ~34s total).
Key fixes:
- eval command: import-aware loop that handles define-library/import
locally without touching the Python bridge pipe (avoids deadlock)
- compile-modules.js: skip vm-compile-adapter, bump timeout
2621/2621 OCaml tests passing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
sx_server.ml: the "eval" command now uses cek_run_with_io instead of
raw eval_expr. This handles import suspensions during eval-blob
(needed for .sx files with define-library/import wrappers).
compile-modules.js: timeout bumped 5min → 10min for sxbc compilation
with define-library overhead.
2608/2608 tests passing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds IO suspension support to the browser WASM kernel. When the VM
hits OP_PERFORM or the CEK suspends (e.g. from import), the kernel
now handles it instead of crashing.
sx_browser.ml additions:
- make_js_suspension: creates JS object {suspended, op, request, resume}
for the platform to handle asynchronously
- handle_import_suspension: checks library registry, returns Some result
if already loaded (immediate resume), None if fetch needed
- api_load_module: catches VmSuspended, resolves imports locally if
library is loaded, returns suspension marker to JS if not
- api_load: IO-aware CEK loop using cek_step_loop/cek_suspended?/
cek_resume — handles import suspensions during .sx file loading
This enables the lazy loading pattern: when a module's (import ...)
encounters an unloaded library, the suspension propagates to JS which
can async-fetch the .sxbc and call resume(). Currently all libraries
are pre-loaded so suspensions resolve immediately.
2608/2608 OCaml tests passing. WASM kernel builds clean.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Phase 1: Absorb sx_scope.ml (180 lines) into sx_primitives.ml. Scope stacks,
cookies, and trace infrastructure now live alongside other primitives. All 20
scope primitive registrations moved. References updated in sx_server.ml and
sx_browser.ml. sx_scope.ml deleted.
Phase 2: Transpiler handles (define name :effects (...) (fn ...)) forms.
ml-emit-define and ml-emit-define-body detect keyword at position 2 and use
(last expr) instead. Unblocks transpilation of spec/render.sx and
web/adapter-html.sx which use 4-child defines extensively.
2598/2598 tests passing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
bootstrap.py produces correct output with no post-processing.
Browser sx_browser.ml updated to use Sx_runtime._jit_try_call_fn.
2566/2568 tests pass (2 pre-existing scope).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Phase 1 engine step 4 — R7RS compatibility primitives for the CEK evaluator.
call/cc: undelimited continuation capture with separate CallccContinuation
type (distinct from delimited shift/reset continuations). Escape semantics —
invoking k replaces the current continuation entirely.
raise/raise-continuable: proper CEK arg evaluation via raise-eval frame.
Non-continuable raise uses raise-guard frame that errors on handler return.
host-error primitive for safe unhandled exception fallback.
Multi-arity map: (map fn list1 list2 ...) zips multiple lists. Single-list
path unchanged for performance. New multi-map frame type.
cond =>: arrow clause syntax (cond (test => fn)) calls fn with test value.
New cond-arrow frame type.
R7RS do: shape-detecting dispatch — (do ((var init step) ...) (test result) body)
desugars to named let. Existing (do expr1 expr2) sequential form preserved.
integer? primitive, host-error alias. Transpiler fixes: match/case routing,
wildcard _ support, nested match arm handling.
2522/2524 OCaml tests pass (2 pre-existing scope failures from transpiler
match codegen, not related to these changes).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Core SX has zero IO — platforms extend __io-registry via (defio name
:category :data/:code/:effect ...). The server web platform declares 44
operations in web/io.sx. batchable_helpers now derived from registry
(:batchable true) instead of hardcoded list. Startup validation warns if
bound IO ops lack registry entries. Browser gets empty registry, ready
for step 5 (IO suspension).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
cssx.sx was deleted in the CSSX → ~tw migration. The test scripts
(test_wasm.sh, test_wasm_native.js, bisect_sxbc.sh) still referenced
it, causing ENOENT during WASM build step 5.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Phase 1 Step 1 of the architecture roadmap. The old cssx.sx
(cssx-resolve, cssx-process-token, cssx-template, old tw function)
is superseded by the ~tw component system in tw.sx.
- Delete shared/sx/templates/cssx.sx
- Remove cssx.sx from all load lists (sx_server.ml, run_tests.ml,
mcp_tree.ml, compile-modules.js, bundle.sh, sx-build-all.sh)
- Replace (tw "tokens") inline style calls with (~tw :tokens "tokens")
in layouts.sx and not-found.sx
- Remove _css-hash / init-css-tracking / SX-Css header plumbing
(dead code — ~tw/flush + flush-collected-styles handle CSS now)
- Remove sx-css-classes param and meta tag from shell template
- Update stale data-cssx references to data-sx-css in tests
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The conditional seeding left gaps — sync_env_to_vm() would wipe entries
from vm_globals that weren't also in global_env. Unconditional seeding
into both tables ensures CALL_PRIM always finds native primitives.
Also prevents SX definitions (stdlib.sx has-key?) from overwriting native
primitives via the env_bind hook on the server.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Seed all primitives into vm_globals as NativeFn values at init.
CALL_PRIM now looks up vm.globals only (not the separate primitives
table). This means OP_DEFINE and registerNative naturally override
primitives — browser.sx's (define set-cookie ...) now takes effect.
The primitives Hashtbl remains for the compiler's primitive? predicate
but has no runtime dispatch role.
Tests: 2435 pass / 64 fail (pre-existing), vs 1718/771 baseline.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Parse Cookie header in OCaml HTTP server for get-cookie primitive
- Stepper saves step-idx to cookie via host-set! FFI on click
- Stepper restores from cookie: get-cookie on server, host-get FFI on client
- Cache key includes stepper cookie value to avoid stale SSR
- registerNative: also update Sx_primitives table for CALL_PRIM dispatch
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root cause: process-elements during WASM boot-init marks elements as
processed but process-one silently fails (effect functions don't execute
in WASM boot context). Deferred process-elements then skips them.
Fixes:
- boot-init: defer process-elements via set-timeout 0
- hydrate-island: defer process-elements via set-timeout 0
- process-elements: move mark-processed! after process-one so failed
boot-context calls don't poison the flag
- observe-intersection: native JS platform function (K.registerNative)
to avoid bytecode callback issue with IntersectionObserver
- Remove SX observe-intersection from boot-helpers.sx (was overriding
the working native version)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
highlight.sx now returns a list of (span :class "..." "text") elements
instead of a string. The rendering pipeline handles the rest:
- Server render-to-html: produces <span class="...">text</span>
- Client render-to-dom: produces DOM span elements
- Aser: serializes spans as SX for client rendering
Key fixes:
- hl-span uses (make-keyword "class") not :class (keywords evaluate
to strings in list context)
- render-sx-tokens returns flat list of spans (no wrapper)
- hl-escape is identity (no escaping needed for tree values)
- highlight.sx added to browser bundle + platform loader
- ~docs/code renders src directly as child of pre
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add missing ./sx/sxc:/app/sxc:ro volume mount to dev-sx compose —
container was using stale image copy of docs.sx where ~docs/code had
(&key code) instead of (&key src), so highlight output was silently
discarded
- Register HTML tags as special forms in WASM browser kernel so keyword
attrs are preserved during render-to-dom
- Add trace-boot, hydrate-debug, eval-at modes to sx-inspect.js for
debugging boot phases and island hydration
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Tests moved from inline JS assertions to web/tests/test-wasm-browser.sx
using the standard deftest/defsuite/assert-equal framework. The JS driver
(test_wasm_native.js) now just boots the kernel, loads modules, and runs
the SX test file.
15/15 source, 15/15 bytecode.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
build-all.sh was missing the sx-platform.js → sx-platform-2.js copy,
so the served file was stale (still requesting .sxbc.json instead of
.sxbc).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The bytecode compiler generates .sxbc (SX text format), not .sxbc.json.
Updated loadBytecodeFile to fetch .sxbc and use load-sxbc for parsing.
Eliminates 404s for non-existent .sxbc.json files.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The reactive-spread test used host-get "className" which doesn't exist
on Node.js DOM stubs. Changed to dom-get-attr "class". Also added
render-to-dom CSSX-in-island-scope test verifying the full hydration
pattern produces correct class attributes.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
reactive-spread (module-loaded via K.load) doesn't apply CSSX classes
to DOM elements. Returns null instead of "sx-text-center". This is the
root cause of hydration stripping all styling — every rendering function
is define-bound from module load.
12 pass, 1 fail — the reactive-spread test is the blocker.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
