Address quality review of a9aa248:
- Replace string-prefix discriminator (`prior.starts_with("class ")`)
with a structural `Origin { Class | Fn }` enum; `kind` is now a
`match` on variants, formatting is reserved for error construction.
- Fix fn-fn collision misreport: in the `Def::Fn` arm, only fire
`MethodNameCollision` when the prior origin is a class. Two
`Def::Fn` with the same name are `CheckError::DuplicateDef`'s job
in `ailang-check`, not this pre-pass.
- Tests use `examples_dir()` like the rest of the module; path
construction is no longer ad-hoc.
- Tests assert `first_origin` / `second_origin` so an origin-format
regression cannot silently break the `kind` discriminator.
- Drop forward-looking 22b.4 prelude reservation note from the
variant doc — keep only what's load-bearing for 22b.2.
All six existing agents (implementer, tester, architect, bencher,
docwriter, debugger) restructured into the same superpowers-derived
layout the SKILL.md files use: Iron Law, Carrier contract, Standing
reading list, Status protocol (DONE / DONE_WITH_CONCERNS /
NEEDS_CONTEXT / BLOCKED), Common Rationalisations, Red Flags. Agents
now know about docs/specs/<milestone>.md and docs/plans/<iteration>.md
but do not open them directly — context curation lives at the skill
level (controller hands the agent task_text, hypothesis, etc.).
Implementer carries TDD as an independent discipline layer, mirroring
the superpowers split between subagent-driven-development (outer loop)
and test-driven-development (inner loop). RED-first applies even when
a plan task forgot to script the failing test.
Debugger scope corrected: RED-first only, hands GREEN to implement
mini-mode. Previously the agent self-applied the fix, which
contradicted skills/debug/SKILL.md Phase 4. The skill is the source
of truth; the agent now matches it.
Two new named reviewer agents:
- ailang-spec-reviewer: did the diff match the task text?
- ailang-quality-reviewer: is the diff well-built? (only after spec
is compliant)
Both replace the ad-hoc general-purpose dispatch in skills/implement
Step 2.3 and 2.4. With named agents, AILang quality conventions are
amortised across dispatches instead of being re-stated inline per
prompt.
skills/implement/SKILL.md updated to dispatch the new reviewers.
skills/README.md agent roster expanded; conventions clarified to
state that agents do not open plan/spec files directly.
The skills are functionally complete and standalone — superpowers
references were attribution-only. Removed so a reader without the
superpowers plugin loaded isn't sent chasing dead links.
Plus: .claude/skills/<name> symlinks so Claude Code's Skill tool can
invoke them by name (analogous to the existing .claude/agents/
symlinks). Discovery section in skills/README.md updated to cover
both symlink sets.
agents/ contained only README.md after the 2026-05-09 migration. The
roster information is more naturally located alongside the skills it
points into. skills/README.md is now the single index (skill table,
pipeline diagram, agent roster, discovery, conventions, how-to-add).
CLAUDE.md updated: code-layout entry consolidated, @-reference points
to skills/README.md.
Three symlinks per skill:
.claude/agents/implement -> skills/implement/agents
.claude/agents/audit -> skills/audit/agents
.claude/agents/debug -> skills/debug/agents
Tracked so contributors get correct subagent dispatch on clone.
Resolved correctly via 'ls -L'.
ailang-implementer/-tester -> skills/implement/agents/
ailang-architect/-bencher/-docwriter -> skills/audit/agents/
ailang-debugger -> skills/debug/agents/
agents/ now contains only README.md (rewritten in next commit as
roster of skill-bound agents).
User feedback after first draft:
- family/iter not standard terms — milestone/iteration adopted
- ailang-docwriter folded into audit/ rather than left orphan
- explicit "Core rules adopted from Superpowers" section so each
inherited Iron Law is named at the spec level, not just implied
Five-skill pipeline (brainstorm, plan, implement, audit, debug) under
skills/, emitting durable artefacts to docs/specs/ and docs/plans/.
Codifies discipline currently spread across CLAUDE.md (TDD-for-bugs,
mandatory tidy-iter, bench-regression rules, feature acceptance) into
trigger-bound skills. Existing per-iter workflow preserved; the layer
formalises the pre-iter design step and the iter handoff contract.
Bootstrap spec — defines the system that future specs will pass through.
Records the iter close: AST schema floor (Def::Class/Instance) +
workspace registry with three coherence checks (Orphan, Duplicate,
MissingMethod), seven new tests (288 → 295), all three bench gates
green at close. JOURNAL queue updated: 22b.1 closed, 22b.2 next
(typecheck arms + FnDef.type.constraints + class-schema validation).
Adds seven fixtures under examples/test_22b1_* exercising the
workspace-load coherence checks plus four positive/negative tests in
ailang-core/src/workspace.rs:
- iter22b1_instance_in_class_module_loads_clean: positive case;
test_22b1_orphan_class.ail.json declares Show + instance Show Int
in the same module (the class's). Registry has one entry.
- iter22b1_orphan_instance_fires_diagnostic: declares Show in
test_22b1_orphan_third_classmod, declares instance Show Int in
test_22b1_orphan_third (a third module). Fires OrphanInstance.
- iter22b1_duplicate_instance_fires_diagnostic: per the JOURNAL hint,
module A defines class Show + instance Show MyInt (legal, A is
class's module), module B defines type MyInt + instance Show MyInt
(legal, B is type's module). Entry imports both. Fires
DuplicateInstance with first/second_module distinct.
- iter22b1_missing_method_fires_diagnostic: class Eq with two
non-default methods (eq, ne), instance Eq Int specifies only ne.
Fires MissingMethod { method: "eq" }.
The surface round-trip gate (ailang-surface/tests/round_trip.rs)
filters out test_22b1_* fixtures because the Form-B parser arms for
ClassDef/InstanceDef are deferred to 22b.4. Without the filter, every
fixture's printed text would re-parse-fail on the unknown `class` /
`instance` head.
Test count: 291 → 295 (4 new workspace tests, all green).
Adds the workspace-global typeclass instance registry per Decision 11
"Resolution and monomorphisation". Built at the end of load_workspace
after the DFS over imports completes; keyed by (class-name, type-hash)
where type-hash uses the new canonical::type_hash (16-hex prefix,
parallel to def_hash and module_hash).
Three coherence checks fire from build_registry, each surfacing as a
distinct WorkspaceLoadError variant:
- OrphanInstance: `instance C T` not in C's or T's defining module.
- DuplicateInstance: two entries share the same (class, type-hash) key.
- MissingMethod: instance omits a required (non-default) method.
The CLI's workspace_error_to_diagnostic is extended with the three new
codes (orphan-instance / duplicate-instance / missing-method).
All existing Workspace { ... } construction sites get the new
`registry` field, defaulting to Registry::default() at synthetic /
test-only paths and threading through ws.registry.clone() at codepath
sites that already hold a real workspace.
Includes hash-stability regression tests (iter22b1_schema_extension_*
and iter22b1_classdef_empty_optionals_hash_stable) and the empty-
registry positive test against examples/sum.ail.json. Test count:
288 → 291 (3 new tests, all pass).
Adds Def::Class(ClassDef) and Def::Instance(InstanceDef) variants per
Decision 11 §"Form-A schema". All optional fields (superclass, doc,
default body) carry skip_serializing_if so canonical-JSON bytes of
pre-22b fixtures stay bit-identical.
Downstream match-on-Def sites are extended with explicit Class/Instance
arms. Behaviour for 22b.1 is placeholder (skip in typecheck/codegen,
one-line summary in pretty/manifest, placeholder marker in prose/print)
with TODO comments naming the deferred sub-iters (22b.2 typecheck,
22b.3 codegen, 22b.4 prose-projection arms).
Adds DESIGN.md "Feature-acceptance criterion" as a top-level section:
a feature ships only if (1) an LLM author naturally produces code
that uses it, and (2) it measurably improves correctness or removes
redundancy. Aesthetic appeal and human ergonomics do not count.
Mirrored in CLAUDE.md as "Feature acceptance: LLM utility", paired
with the existing "Design rationale != implementation effort". The
two together narrow valid feature rationales to one thing: what the
LLM author gets out of the feature.
Trigger: the typeclass-design conversation around 22a. Rule was
implicit in many past decisions (Decision 1's JSON-over-text choice,
Decision 10's "what LLMs are good at" reasoning) but never stated
as a feature-gate. Codifying it now means future feature proposals
get evaluated against an articulated criterion instead of being
re-derived each time.
Documentation-only; no Rust, schema, or bench changes. Test state
288/0/3 unchanged.
Three drift items from ailang-architect, plus one false-positive
surfaced during verification:
1. DESIGN.md silent on closure-pair 4.14x finding (21'b).
Decision-10 ratified: "Workload scope of the 1.3x target"
paragraph scopes the retirement gate to linear/tree/poly-ADT
workloads; closure-pair carve-out documented as
representational cost (closure cell + env struct = 2 allocs
per step) until a slab/pool answer ships.
2. bench/compile_check.py corpus drift. Three fixtures added
(bench_compute_intsum, bench_compute_collatz,
bench_list_sum_explicit), re-baselined. Now 12 fixtures x
2 ops = 24 compile-time metrics.
3. baseline.json convention not codified. Note field gains
"max-of-distribution gets wider band than percentile"
convention discovered in 21'd.
4. (verification finding) bench_compute_intsum cross_lang
tolerances at 15%/12% fire on subprocess-spawn jitter for
sub-millisecond fixtures. Widened to 35% across all five
intsum metrics; convention recorded in baseline_cross_lang
note field (sub-ms fixtures need looser bands).
All three bench gates re-run sequentially after edits:
bench/check.py — 63 metrics; 63 stable
bench/compile_check.py — 24 metrics; 24 stable
bench/cross_lang.py — 25 metrics; 25 stable
Total: 112 metrics under regression coverage, all green.
288 tests passing, 3 ignored. No Rust changes.
Closes the apples-to-apples gap from 21'e. Adds:
- examples/bench_list_sum_explicit.ailx — same algorithm and sizes
as bench_list_sum, fully (borrow)/(own)/(drop-iterative)
annotated so codegen emits proper inc/dec instrumentation.
- bench/reference/list_sum_explicit_free.c — same algorithm
with explicit free() walking the chain after sum.
The full alloc+dec vs malloc+free comparison reveals two non-
trivial conclusions:
1. AILang's full RC pipeline is only 26% slower than glibc
malloc+free on this workload (rc/c = 1.26x). The implicit-
mode comparison's 1.42x was misleading — it counted neither
pipeline's free path. The fair ratio is 1.26x, materially
better than the previous read.
2. RC's dec is cheaper per cell than glibc free(). AILang
dec-tax: ~3 ns/cell. C free-tax: ~5.5 ns/cell. Plausible
cause: ailang_rc_dec operates on a known-shape cell with a
fixed-offset refcount and a static per-type drop fn — no
free-list bucketing, no header introspection, no global lock.
bump's advantage expresses fully: bench_list_sum_explicit.bump/c
= 0.42x means AILang at bump is 2.4x faster than C malloc+free.
Sets a useful upper bound on a slab/pool RC allocator's potential.
The 21'-family arc — bench-regression infrastructure — is now
substantively complete: 21'a (bench/check.py), 21'b (corpus
widening), 21'c (compile_check.py), 21'd (pure-compute fixtures
+ harness hardening), 21'e (cross-language hand-C), 21'f (explicit
apples-to-apples). 63 runtime metrics + 18 compile metrics + 25
cross-lang metrics under regression coverage. Any future iter
that regresses any axis beyond tolerance gets caught at the next
family close.
Remaining queue is back to substantive language work — Family 21
(typeclasses / polymorphic ADTs at runtime / pattern-binding
generalisation) is now an orchestrator-level fork that needs
direct user input.
Closes the question CLAUDE.md has carried since day one ("LLVM-
linkable, performance is extremely important") with data. Hand-C
variants of the four bench fixtures, compiled with clang -O2,
each carefully matching the AILang algorithm and explicitly
documenting representation differences (cell width, leak policy)
that affect the ratio.
Three substantive findings:
1. Pure-compute parity with C: bench_compute_collatz runs at
AILang/C = 0.99x across both allocators. AILang's IR composes
with LLVM's optimizer at the same level a hand-C source does.
This is the LLVM-linkable performance claim, backed by data
for the first time. bench_compute_intsum (1.05-1.18x) confirms.
2. AILang bump beats glibc malloc 2x on linear allocation:
bench_list_sum.bump/c = 0.50x. Bump's two-instruction inline
fastpath outperforms glibc's free-list-managed malloc on
no-free workloads. Quantitatively measured for the first time.
3. RC overhead vs C malloc quantified: bench_list_sum.rc/c =
1.49x, bench_tree_walk.rc/c = 2.61x. The 8-byte refcount
header + zero-init + libc backing add 50-160% over glibc
malloc on these implicit-mode workloads. Explicit-mode + a
free()-adding C variant (21'f, queued) will close the
apples-to-apples gap on dec-cost.
CLAUDE.md updated to list bench/cross_lang.py as the third
tidy-iter gate alongside bench/check.py and bench/compile_check.py.
20 new metrics in bench/baseline_cross_lang.json with 12-15%
tolerances (cross-language ratios are inherently noisier than
within-AILang ratios — two compiler stacks contribute variance).
Closes the third corpus blind spot (heap-allocation-only) by
adding two fixtures with no allocation pressure: bench_compute_
intsum (tail-recursive integer accumulator) and bench_compute_
collatz (Collatz step-counter, branchy).
Surprise on intsum: 50M-iteration loop runs in 1ms wall under
all three allocators. LLVM's induction-variable analysis applies
the closed-form triangular-sum reduction to AILang's IR — a
positive codegen finding (the IR composes with LLVM's optimizer
at the same level a hand-C loop would) but it makes intsum
useless as a runtime regression bench. Excluded from run.sh's
fixtures array; kept in examples/ as reference and as a future
cross-language comparison anchor.
Collatz survives optimization (data-dependent control flow). At
56ms wall, gc/bump/rc all within 2% — the canonical "pure-compute
is allocator-invariant" data point this fixture is meant to
prove. If a future codegen change leaks an allocation into the
inner loop, the 1.00x / 1.02x ratios diverge visibly.
Two infrastructure fixes the new fixtures forced:
- 6-decimal precision in run.sh's Python timing helper and median
averager (was 3-decimal; sub-ms times rounded to 0.000 and
crashed the ratio awk with Division durch Null).
- Zero-guard in the ratio awk (defensive even with the precision
bump, since LLVM-eliminated workloads can still hit zero).
Latency baseline: implicit_at_rc.max_us tolerance 25% -> 30%.
Three captures today (477 / 456 / 609 µs) show natural run-to-run
dispersion wider than the original tolerance accounts for. Not a
softening to dodge regression — the original baseline was the
first capture; a fairer tolerance across natural max-of-1000-
samples width is what the harness needed from the start.
Baseline file: 47 -> 55 metrics. 21'e (cross-language reference,
clang -O2 hand-C ratios) is the natural next dispatch.
Closes the second axis the user named: every typechecker / codegen
perf change was previously invisible to the tidy-iter gate. With
Family 21 typeclasses (queued) and 21'b's poly-ADT additions both
pushing on the typechecker, naive substitution loops would have
landed silently and decayed the compile path.
bench/compile_check.py is a separate script from bench/check.py
because the methodology differs: sub-process spawn timing on small
workloads (1ms scale for `ail check`, 65ms for `ail build`) vs.
allocator-stress on large ones (multi-second). Tolerances differ
by an order of magnitude (25% / 20% here vs. 5-15% there).
Empirically: ail check is sub-ms across the corpus, dominated by
subprocess spawn (~5-10ms on Linux); ail build is 63-69ms,
dominated by clang's link step. The bench is a catastrophe
detector (10x slowdowns visible) — finer regressions need a
profiler. CLAUDE.md updated to list both scripts as co-equal
tidy-iter gates alongside the architect drift report; exit 0/1/2
semantics are uniform across both.
JOURNAL queue: 21'd (pure-compute fixtures) and 21'e (cross-
language reference / hand-C ratio) remain to land the LLVM-
linkable performance claim.
Two new throughput fixtures targeting blind spots in the 21'a
corpus:
- bench_closure_chain exercises the build_pair_drop_fn codegen
path (the 18c.4 doubled-braces trigger). Each iteration of
run_loop allocates a {thunk, env} closure pair via the
let-rec-name-as-value escape route. Sizes 10k / 100k / 500k.
rc/bump = 4.14x — materially worse than the 2.91x / 2.59x of
the linear/tree fixtures, exposing that closure work pays the
RC alloc tax twice (pair + env-struct).
- bench_hof_pipeline exercises poly-ADT instantiation and
indirect dispatch via fold_with_fn over List<a>. Sizes 100k /
1M / 3M elements. Ratios essentially match bench_list_sum,
confirming the 13b static-template-plus-ctor-inline design
has zero measurable overhead at this scale.
Baseline file extends from 31 to 47 metrics. The two new fixtures
build clean under all three allocators; the rc-arm build exercises
the per-type drop fn for the closure-pair, providing a tripwire
for any future 18c.4-class IR malformedness.
JOURNAL records both surprises (4.14x closure tax, ~zero HOF/poly
overhead) and explicitly notes the dispersion observation on
explicit_at_rc.p99 — three captures today (357.5 / 294.6 / 251.5)
confirm wide run-to-run variance on that fixture. Methodology
upgrade (n>=10 captures or tighter fixture) deferred to 21'c.
bench/run.sh fixtures array updated. bench/check.py needed no
changes — its parser handles the wider table by metric name.
Adds a "Performance regressions" subsection to the iter-cycle
discipline. bench/check.py runs at every family close alongside
the architect drift report; its exit code is the gate. Skipping
it requires the same explicit JOURNAL entry as skipping the
drift review.
Codifies the discipline that 21'a's JOURNAL entry recommended
and explicitly punted to the orchestrator. The tooling shipped
yesterday; the rule that makes it load-bearing ships now.
Also clarifies two non-goals: per-metric tolerances are not the
language correctness bar (Decision-10 thresholds remain in
DESIGN.md, evaluated against absolute numbers); and improvements
do not auto-update the baseline — they surface for ratification.
Closes the structural gap between bench/run.sh (one-shot capture
into JOURNAL) and a continuous tripwire. baseline.json records 31
metrics (16 throughput, 15 latency) with per-metric one-sided
tolerances tuned to absorb run-to-run noise on a quiet developer
machine; check.py spawns run.sh, parses both the throughput
pipe-table and the per-arm latency stanzas, diffs against the
baseline, prints a per-metric report, and exits non-zero on any
regression beyond tolerance.
User-facing flags: --from-file, --stdin, --baseline, and
--update-baseline (re-run + overwrite baseline.json after
intentional improvements).
Validation: captured the baseline, then re-ran end-to-end. All 31
metrics within tolerance. The harness also caught a single-capture
explicit-rc p99 spike (357.5 us) that was first read as drift vs.
yesterday's 18g.tidy.fu2 numbers but came in at 294.6 us on the
follow-up run — exactly the kind of noise the tolerance band is
there to absorb. Without 21'a we would have either chased a
phantom or buried the signal; with it, single noisy runs are data
points, not verdicts.
Tidy-iter discipline addition (run check.py at every family close
alongside the architect drift report) is recommended in JOURNAL
but not enacted in this iter — that's an orchestrator-level
update to CLAUDE.md.
Brummel