EvalError moved to types.py. All 27 files updated to import eval_expr,
trampoline, call_lambda, etc. directly from shared.sx.ref.sx_ref instead
of through the evaluator.py indirection layer. 320/320 spec tests pass.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Evaluator: add _Thunk + _trampoline for tail-call optimization in
lambdas, components, if/when/cond/case/let/begin. All callers in
html.py, resolver.py, handlers.py, pages.py, jinja_bridge.py, and
query_registry.py unwrap thunks at non-tail positions.
SX docs: update tagline to "s-expressions for the web", rewrite intro
to reflect that SX replaces most JavaScript need, fix "What sx is not"
to acknowledge macros and TCO exist.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The inter-service data layer (fetch_data/call_action) was the least
structured part of the codebase — Python _handlers dicts with ad-hoc
param extraction scattered across 16 route files. This replaces them
with declarative .sx query/action definitions that make the entire
inter-service protocol self-describing and greppable.
Infrastructure:
- defquery/defaction special forms in the sx evaluator
- Query/action registry with load, lookup, and schema introspection
- Query executor using async_eval with I/O primitives
- Blueprint factories (create_data_blueprint/create_action_blueprint)
with sx-first dispatch and Python fallback
- /internal/schema endpoint on every service
- parse-datetime and split-ids primitives for type coercion
Service extractions:
- LikesService (toggle, is_liked, liked_slugs, liked_ids)
- PageConfigService (ensure, get_by_container, get_by_id, get_batch, update)
- RelationsService (wraps module-level functions)
- AccountDataService (user_by_email, newsletters)
- CartItemsService, MarketDataService (raw SQLAlchemy lookups)
50 of 54 handlers converted to sx, 4 Python fallbacks remain
(ghost-sync/push-member, clear-cart-for-order, create-order).
Net: -1,383 lines Python, +251 lines modified.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
