Adds an "Iter cycle" top-level section: work clusters into named
iter families (18a–f, 19a–c, …); after each family closes, the
next iter is a non-optional tidy-iter that runs ailang-architect
over the surface and resolves every drift item by fixing,
ratifying in DESIGN.md, or recording acceptable drift in
JOURNAL.md. Skipping requires an explicit JOURNAL entry naming
the reason.
Also writes down the previously-implicit roles of DESIGN.md
(canonical spec; what AILang IS) and JOURNAL.md (decisions log;
HOW we got here, what's queued, rationale not in DESIGN.md). The
rule was practice for a stretch, then quietly evaporated when no
written rule kept it alive across sessions; codifying it here so
the next compaction doesn't lose it again.
User correction during the post-RC-overnight check-in: the 18a
"Type::Fn metadata vs. new Type variant" call was justified
across DESIGN.md, JOURNAL, and the commit message primarily by
"avoids ~250 match-arm sites". That is an observation about the
current state of the code, not a design rationale.
CLAUDE.md gains two binding rules:
- Design rationale ≠ implementation effort. Effort is at most a
tiebreaker; a choice whose only stated reason is effort is
suspect. The rule names the 18a misstep as the canonical
anti-example so future sessions catch it earlier.
- Direction freedom + bounce-back conditions inlined (was
previously a cross-reference to a private auto-memory file
outside the repo, which the user couldn't see).
DESIGN.md Decision 10's Schema-additions block now leads with
the substantive reasons for per-position metadata: semantic
locality (modes belong to fn-parameter positions, not to types
in general — Decision 1 line); compositional clarity (type
identity vs. calling convention factor apart); future-proofing
(per-position metadata generalises; Type-variant approach
combinatoric blows up). The match-arm count remains parenthetical,
explicitly named a tiebreaker.
JOURNAL records the correction itself — the mistake stays as a
data point because how design discipline corrupts is informative.
Records that 18c.1 ran zero-surprise (mechanical recursion at
~10 sites across 6 files), the future-work seam in codegen for
18c.3 is a single line (emit one rc_inc call before returning
the inner SSA reg), and 18c.2's linearity check should start
as 'green for borrow_own_demo' since that fixture's shape is
exactly what the check needs to accept.
Adds (clone X) as a new Term variant. Form-A:
(let p (expensive_fn x)
(do consume_a (clone p)) ; explicit RC inc — needs 18c.3
(do consume_b p))
Schema floor for the LLM-aware RC design (Decision 10): explicit
clone is the author-visible alternative to implicit sharing. In
18c.1 the variant is purely additive — typechecker treats it as
identity (same type as inner), codegen lowers it to the inner
term's IR with no extra calls. Iter 18c.3 will replace the
codegen identity-arm with a single call:
Term::Clone { value } =>
let v = self.lower_term(value, ...)?;
self.emit("call void @ailang_rc_inc(ptr {v})");
Ok(v)
Match-arm count: ~10 sites across 6 files, all mechanical
structural recursion through `value`. The codegen `lower_term`
arm carries an explicit comment marking the 18c.3 emission seam.
Hash invariant: `(clone X)` uses a new serde tag "clone" that no
pre-18c.1 fixture contains. Every existing fixture's canonical
JSON is bit-identical (git diff examples/ empty).
New fixture examples/clone_demo.{ailx,ail.json}:
(let x 42 (do io/print_int (clone x)))
Stdout 42 under both --alloc=gc and --alloc=rc.
Tail-position propagates through clone (verify_tail_positions
treats `(clone tail-call)` as itself a tail call), per the
identity-of-clone semantics in 18c.1.
Verified:
- cargo build --workspace clean
- cargo test --workspace green: 52 e2e (+1), 14 surface parse
(+2 for clone tests), all other buckets unchanged
- ail render round-trip exact
- clone_demo --alloc=gc → 42; --alloc=rc → 42
Records that the codegen change was a one-line extension because
the 18a bump path had centralised the allocator-symbol decision
on fn_name(). Documents the runtime ABI (8-byte refcount header,
ailang_rc_alloc/inc/dec) and the deliberate no-inc/dec posture:
programs leak intentionally under --alloc=rc; 18c lights up the
actual reference counting.
Next-step plan for 18c is split into three sub-iters
(clone schema / linearity check / inference + codegen) since
each wants its own verification cycle.
Wires up reference-counting allocator end-to-end without any
inc/dec emission. Programs run under --alloc=rc and produce
correct stdout (validated against --alloc=gc); they leak every
allocation, exactly like the pre-Boehm era. The point of 18b is
to establish the runtime contract before 18c adds the
inc/dec-emission codegen pass.
runtime/rc.c — new file. 8-byte uint64 refcount header
prepended to every payload; ailang_rc_alloc(size) returns a
ptr to the payload (header at ptr-8). ailang_rc_inc / dec are
declared but never called by codegen yet; they exist so 18c
can wire codegen against a stable runtime ABI. dec frees on
zero refcount but does NOT recursively dec child references —
that's 18c's job once it has per-ctor type info.
crates/ailang-codegen/src/lib.rs — AllocStrategy::Rc variant
added; fn_name() returns "ailang_rc_alloc". Single-line
extension because Iter 18a's bump path had already centralised
the allocator-symbol decision on fn_name() for all four
allocation sites.
crates/ail/src/main.rs — --alloc=rc accepted by Build/Run;
parse_alloc_strategy extended; locate_rc_runtime() helper
mirrors locate_bump_runtime; new Rc arm in build_to compiles
runtime/rc.c with clang -O2 -c and links the resulting .o
into the final binary (no -lgc).
E2E coverage: alloc_rc_produces_same_stdout_as_gc on
list.ail.json (42), alloc_rc_matches_gc_on_std_list_demo for
broader allocation-site coverage. Total e2e bundle: 51 tests
(was 49).
Hand-verified on:
sum.ail.json --alloc=rc → 55
list.ail.json --alloc=rc → 42
borrow_own_demo.ail.json --alloc=rc → 3 then 6
std_list_demo.ail.json --alloc=rc → matches --alloc=gc
cargo build/test --workspace green; git diff examples/ empty.
Records the schema-choice rationale (per-position metadata on
Type::Fn vs new Type variants — the latter would be a 250-site
mechanical migration, the former is one iter), the
padding-on-read trick that kept construction sites unchanged,
and the forward-compat note on the new fixture's borrow+own
sharing pattern.
Adds form-A surface (borrow T) / (own T) wrappers in fn-type
params and ret slots. Internally these are not new Type variants
but per-position metadata on Type::Fn:
Type::Fn { params, param_modes, ret, ret_mode, effects }
enum ParamMode { Implicit, Own, Borrow }
This keeps Type itself unchanged, so unification, occurs, apply,
and ~190 other Type match-arms in the typechecker need no new
branch. Fields use serde skip-if-default predicates so canonical
JSON hashes for every pre-18a fixture stay bit-identical (sum,
list, hof, closure, list_map, etc.: zero diff under git).
Iter 18a is purely additive: typechecker treats modes as
transparent (mode-compat check deferred to 18c), no codegen
change (--alloc=gc / Boehm path unchanged), no linearity
enforcement. The annotation info flows into the JSON side-table
that 18c will consume.
Equality of mode slices is length-tolerant: a vec![] (the form
written by mechanically-updated construction sites that elide
modes) compares equal to vec![Implicit; n]. This kept the patch
from being a 100-site mechanical migration through the
typechecker. 18c will choose: keep the slack, or normalise to
full-length on construction.
New fixture examples/borrow_own_demo.{ailx,ail.json} exercises
both modes. list_length declares (borrow (con List)); sum_list
declares (own (con List)). Stdout: 3 then 6.
Documentation: DESIGN.md schema-additions block clarified that
modes are Type::Fn metadata (not Type variants); migration plan
flag rename --memory=rc → --alloc=rc to match the existing CLI
flag.
Verification:
- cargo build --workspace green
- cargo test --workspace green (154 tests, +1 vs baseline 153)
- git diff examples/{sum,list,hof,closure,list_map,...}.ail.json: empty
- borrow_own_demo runtime stdout: "3\n6\n"
- canonical JSON contains "param_modes":["borrow"] and ["own"]
Avoids the ~190 Type match-arms in the typechecker that a new
Type variant would require. ParamMode enum (Implicit | Own |
Borrow) attaches per-param to Type::Fn, with serde defaults so
existing fixture hashes stay bit-identical.
Replaces the earlier "RC + inference, no annotations" position
with a two-layer architecture: inference (intra-fn) + mandatory
LLM-author mode annotations on fn signatures (inter-fn). Adds
structured rationale for rejecting region inference (regions
assume stack-shaped lifetimes; real programs need
non-stack-shaped lifetimes for caches / memo tables) and linear
types as primary mechanism.
Schema additions enumerated for the 18-series:
Type::Borrow / Type::Own (Iter 18a)
Term::Clone (Iter 18c)
Term::ReuseAs (Iter 18d)
TypeDef.drop_iterative (Iter 18e)
Migration plan extended from 18a-d (4 iters) to 18a-f (6 iters).
JOURNAL records the regions excursion, the scope-shape reduction,
and the LLM-author lever as the structural advantage Decision 10
cashes in on.
Adds --alloc=<gc|bump> to ail build/run. Bump path links a 256MB
no-free arena C stub instead of libgc; IR is byte-identical except
for the @GC_malloc → @bump_malloc symbol swap. Bench harness times
two allocation-heavy workloads (list cons/sum and balanced tree
build/walk) under both modes.
Numbers (RUNS=5, median of 4):
bench_list_sum gc 0.141s bump 0.048s +194%
bench_tree_walk gc 0.103s bump 0.041s +151%
Bucket: large. ~60% of runtime is Boehm on these workloads —
upper bound for any realistic program. Both fixtures hold the
heap fully live, so the cost we're seeing is Boehm's allocate
path itself, not collection work; that fact narrows the design
space for the GC discussion.
- crates/ailang-codegen: AllocStrategy enum, three callsites and
the IR header parameterised.
- crates/ail/src/main.rs: --alloc flag plumbed; bump runtime
located + compiled on demand.
- runtime/bump.c: 256MB static arena, abort-on-overflow.
- examples/bench_list_sum, bench_tree_walk: accumulator-form
fixtures (textbook recursive sum was constructor-blocked).
- bench/run.sh: harness with Python timing helper (Arch's
/usr/bin/time isn't part of the base install).
No language-level changes; default --alloc=gc, all 141 workspace
tests green, all 5 IR snapshots unchanged, 11 prior fixtures
produce identical stdout.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a conservative escape analysis that flags Term::Ctor and
Term::Lam allocations as non-escaping when they are bound by a
let, never flow to a fn arg / ctor field / tail-return / lambda
capture, and remain inside the allocating fn's frame. Codegen
lowers flagged sites to LLVM `alloca` instead of `@GC_malloc`;
escaping sites continue to use Boehm.
- escape.rs (new): name-based taint propagation; let-only
candidates; tail-position / app-arg / ctor-field / lam-capture
all taint.
- codegen: three sites switched (lower_ctor, lam env, closure
pair). Save/restore non_escape across lambda thunk emit.
- examples/escape_local_demo.{ailx,ail.json}: focused fixture
exercising both branches (peek and recursive count).
- e2e + escape unit tests: 133 → 141 (+8).
Stop-gate: 17a closes the queue. Memory-management discussion
(GC scope) is the next user-driven step. Per-fn arena observations
appended to JOURNAL — including the headline finding that 0/270
shipped allocations are flagged non-escaping under this rule
(real AILang threads ctors directly between fns; build-locally /
consume-locally is not idiomatic).
All existing fixture stdouts and content hashes bit-identical;
IR snapshots unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Lifts the Int-only restriction on `==`. Declared type becomes
forall a. Fn(a, a) -> Bool; codegen monomorphises and dispatches:
Int → icmp eq i64
Bool → icmp eq i1
Str → call @strcmp + icmp eq i32 0
Unit → constant true (operands still emitted for side effects)
ADT / Fn / other → CodegenError::Internal
This unblocks 16c's build_eq for non-Int lit patterns. == joins
__unreachable__ as the second polymorphic builtin (same Forall
machinery).
- check/builtins.rs: == registered as Forall(a, Fn(a, a) -> Bool).
- codegen: lower_eq dispatch table; @strcmp declared in IR header
alongside @printf/@GC_malloc/@puts.
- examples/eq_demo.{ailx,ail.json}: covers all four supported
scalars including a Str-lit-pattern match.
- IR snapshots refreshed: only +declare i32 @strcmp(ptr, ptr) in
the header; every define body bit-identical.
- e2e + check + codegen tests: 124 → 133 (+9, of which +1 is the
e2e fixture and the rest exercise the new dispatch / typecheck
surface).
Other comparison ops (<, <=, >, >=, !=) remain Int-only — out of
scope for this iter.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Eliminates the Unit-typed chain default that 16c left in place.
Adds __unreachable__ : forall a. a — codegen lowers to LLVM
unreachable. The 16a/16c chain machinery now uses it as the
deepest fall-through, so exhaustive matches with non-Unit arms
no longer need a (case _ ...) workaround.
- check/builtins.rs: register __unreachable__ as Forall(a, a).
- codegen: Term::Var "__unreachable__" emits LLVM unreachable
and sets block_terminated; If/Match/fn-body gate downstream
work on that flag.
- desugar: chain default switched from Lit{Unit} to Var.
- examples/lit_pat: categorize_first's trailing _ arm removed.
Hash changes from c4faec3abc2ed388 to 644de0c0ec15fc17.
- examples/unreachable_demo: positive fixture using safe_div
with __unreachable__ in the impossible branch.
- e2e + desugar tests: 122 → 124 (+2).
Path-a from 16b.2 planning: real primitive over a desugar-time
exhaustiveness pre-check (which would need cross-module type
registry access from desugar).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Inner LetRec capturing outer LetRec's params now lifts cleanly via
the existing post-order traversal — outer's params are KnownType
in inner's outer-scope, so 16b.2's fast-path handles them. Inner
LetRec capturing outer's NAME remains rejected (closure-of-self
in body, deferred separately).
Plus hardening: lift.rs's deferred path tracks the outer LetRec
name in a parallel `enclosing_letrec_names` set so the symmetric
inner-captures-outer-name rejection fires there too — without it
a silent lift produced a runtime arity mismatch.
- examples/nested_let_rec.{ailx,ail.json}: outer counts down via
inner that captures the outer's threshold param.
- desugar.rs: clarified panic message for outer-name capture.
- lift.rs: enclosing_letrec_names tracking.
- e2e + desugar tests: 120 → 122 (+2).
End-of-16b series: feature-complete except for closure-of-self
in body and closure-with-polymorphism, both queued separately.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Lifts the monomorphic-only restriction. Lifted top-level fn now
becomes Forall(α..., Fn(t..., T...) -> tr) when the enclosing fn
is polymorphic; capture types may mention outer type vars and
the existing mono pipeline (Iter 12b/14a) specializes them
correctly at call sites.
- desugar.rs / lift.rs: scope-building unified — Forall fn-params
are now KnownType, not LetBound. Lift wraps augmented_ty in
Forall mirroring the enclosing fn.
- Name-as-value-in-in-term in a polymorphic enclosing fn is
rejected (eta-Lam wrap from 16b.5 has no Forall). Queued
separately as closure-conversion-with-polymorphism.
- examples/poly_rec_capture.{ailx,ail.json}: apply_n_times :
Forall(a). drives at Int and Bool to exercise two mono
instantiations.
- e2e + check + desugar tests: 116 → 120 (+4).
Codegen pipeline untouched — apply_subst_to_type substitutes
through Fn.params uniformly (original params + appended captures).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Drops the direct-call-only restriction on the LetRec name when it
appears in `in_term`. No new ABI: the lift wraps the in-term in
`(let f (lam P (app f$lr_N P CAPTURES)) in_term')` so codegen's
existing Iter-8b closure-pair machinery handles it. Body-position
name-as-value still rejected (deferred — eta-of-self is harder).
- desugar.rs / lift.rs: split find_non_callee_use into body
(panic) and in_term (wrap).
- examples/local_rec_as_value.{ailx,ail.json}: no-capture path
(factorial passed to apply5 → 120).
- examples/local_rec_as_value_capture.{ailx,ail.json}: capture
path; lifted fn has captures, eta-Lam picks them up via 8b.
- e2e + desugar tests: 113 → 116 (+3).
Codegen Lam machinery handled the wrapped form on the first try.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Flips the desugar's MatchArm-capture rejection from panic to defer.
The 16b.3 lift_letrecs pass already handled match-arm bindings via
type_check_pattern_for_lift (Pattern::Var, Pattern::Ctor with sub-
patterns, ctor-field substitution against scrutinee args), so this
iter is a single-line classification change in desugar plus the
fixture and tests that exercise it.
- desugar.rs: MatchArm classification now defers (same arm as
LetBound); EnclosingLetRec panic remains for 16b.7.
- examples/local_rec_match_capture.{ailx,ail.json}: enclosing fn
pattern-matches on Pair<Int,Int>, inner LetRec captures both
match-arm bindings simultaneously. Lifts to
loop$lr_0(i: Int, threshold: Int, n: Int) -> Int.
- e2e + check + desugar tests: 110 → 113 (+3).
First fixture lifting more than one capture; subst_call_with_extras
already handled it generically.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds path-2 from the 16b.2 planning entry. Desugar now defers
LetRecs whose captures include Term::Let-bound names; a new
ailang-check::lift_letrecs pass runs after typecheck and uses the
elaborated env to resolve capture types. ailang-check learns a real
Term::LetRec typing rule in synth and verify_tail_positions.
- desugar: Term::LetRec arm gains a defer-arm; helpers promoted to
pub for reuse by the lift pass; find_non_callee_use moved before
classification.
- ailang-check::synth/verify_tail_positions: real LetRec rules
(effect-subset, locals install, recursive name in body+in_term).
- ailang-check::lift.rs (new, 720 LOC): post-typecheck lift with
post-order traversal, env-walk for capture-type resolution,
fast-path skip when no LetRec is present.
- ail::main.rs: build path now does load → check → desugar →
lift_letrecs → codegen.
- examples/local_rec_let_capture.{ailx,ail.json}: new fixture
capturing a let-bound `threshold` in a recursive helper.
- e2e + check unit tests: 106 → 110 (+4).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Lift 16b.1's no-capture restriction. The desugar pass's `scope`
becomes a `BTreeMap<String, ScopeEntry>` carrying `KnownType` for
fn/Lam-params, sentinels for Let-/Match-bindings. LetRec captures
of `KnownType` names are lifted: the synthetic top-level fn's
signature gets the capture types appended, and every call site of
the LetRec name is rewritten via `subst_call_with_extras` to pass
the captures positionally. Captures from let-bindings, match-arm
patterns, polymorphic enclosing fns, and name-as-value uses are
rejected at desugar with panics pointing at 16b.3-16b.7.
Demo: `sum_below(n)` recurses via `(let-rec loop (params i) ...
(body ... uses n ...) (in (app loop 1)))`. Lifts to
`loop$lr_0(i: Int, n: Int) -> Int` with `(app loop X)` rewritten
to `(app loop$lr_0 X n)`. Outputs 0, 10, 45.
Tests: 103 -> 106 (+1 e2e local_rec_capture_demo, +2 desugar unit;
the 16b.1 capture-panic test was repurposed into the positive
fn-param-capture test).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Orchestrator-level planning artifact, not an iteration log.
Captures the design space for 16b.2 so the user can review
trade-offs before implementation. Three architectural paths
considered (desugar-only with restrictions, post-typecheck pass,
mini-inference inside desugar); recommends the first as the
incremental path. Lists restrictions for the safe subset
(fn/Lam-param captures only, direct-call only, monomorphic
enclosing fn, single-level LetRec) and queues the lifted
restrictions as 16b.3-16b.7. Adjacent open items (16d chain
terminator, 16e == extension) likewise scoped.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Desugar `Pattern::Lit` (top-level and nested in Ctor) to
`Term::If { cond = (== sv lit), then = body, else_ = fall_k }`.
After 16c, no Pattern::Lit survives the desugar pass; codegen
and typechecker never see one. `is_flat` reclassifies Lit as
non-flat to force the chain machinery; new `build_eq` helper
constructs the equality test per literal kind.
Tests: 99 → 103 (+1 e2e lit_pat_demo, +3 desugar unit). Demo
fixture exercises top-level lit arms (classify) and nested
lit-in-Ctor (categorize_first) with a local IntList; output
100/200/999/-1/0/7.
Known limitations documented in DESIGN.md and JOURNAL: (a)
chain machinery's Unit terminator means non-Wild-terminated
exhaustive matches with lit arms still need a trailing `_`
catch-all to type-check; (b) `==` is currently Int-only, so
Bool/Str/Unit lit patterns desugar correctly but error at
typecheck.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add `(let-rec NAME (params ...) (type ...) (body ...) (in ...))`
as a form-A surface and a `Term::LetRec` AST variant. The 16a
desugar pass lifts each LetRec whose body has no captures from
the enclosing scope to a synthetic top-level fn `<hint>$lr_N`
and substitutes the original name; typecheck and codegen never
see LetRec. Capture detection panics at desugar time, queued
for 16b.2.
Tests: 95 → 99 (+1 e2e local_rec_factorial_demo, +2 desugar
unit, +1 parse unit). The new fixture `examples/local_rec_demo`
runs `fact` at n=1, 3, 5 → prints 1, 6, 120.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds the two canonical prefix-slicing combinators to std_list.
First fns in std_list to combine `if` + Int arithmetic + recursive
ADT pattern in one body.
take : forall a. (Int, List<a>) -> List<a>
— first n elements (or all of xs if shorter than n).
drop : forall a. (Int, List<a>) -> List<a>
— list with first n elements removed (Nil if n >= length).
Both use `(if (<= n 0) ...)` as the base-case guard since
literal sub-patterns inside Ctor patterns aren't yet supported
(queued as 16c — would let take/drop collapse the if into the
match).
examples/std_list_more_demo: builds [1, 2, 3, 4, 5] once and
exercises take/drop at n=0, n=mid, n=overflow on each. Expected
output 0, 3, 5, 5, 3, 0 (one per line).
Tests: 95/95 (e2e went from 35 to 36). std_list grew from 10 to
12 combinators; total stdlib 5 modules / 29 combinators. No new
compiler bug surfaced.
Pre-existing std_list defs are byte-identical (verified via diff
on .ailx and round-trip on .ail.json), so the five downstream
importers (std_list_demo, std_list_stress, std_either_list,
std_either_list_demo, list_map_poly) need no regeneration.
Fix the codegen asymmetry that 15g surfaced. unify_for_subst had an
arg-side-only early-return for $u-prefixed synth wildcards. The
function's prev-binding recursion can swap a $u from arg into param
position when re-unifying a previously-bound type against a fresh
arg. Reduced repro: `length [Left 1, Right 10]` — `a` first binds
to `Either<Int, $u>`, then the recursive unification against
`Either<$u, Int>` lands $u in param-pos[1] and falls through to
the catch-all error.
Fix is three lines: add a symmetric $u early-return for param side.
$u is a synth-only wildcard regardless of which side it ends up on
after the prev-binding swap. Doc comment expanded to record the
origin and justification.
std_either_list_demo refactored: mkleft/mkright workaround helpers
removed; the list is now constructed inline by mixing
(term-ctor std_either.Either Left 1) and (... Right 10) directly.
Same expected output (2, 3, 2, 3); the demo doubles as the 15g-aux
regression fixture.
Tests: 94/94, unchanged. The fix expanded what compiles without
changing observable behaviour for any prior fixture.
First stdlib fn that imports three other stdlib modules (std_list,
std_either, std_pair) and returns a compound polymorphic ADT tree
(Pair<List<e>, List<a>>). Stresses monomorphisation across nested
parameterised ADTs.
What shipped:
- examples/std_either_list.{ailx,ail.json}: 3 combinators —
- lefts : forall e a. (List<Either<e, a>>) -> List<e>
- rights : forall e a. (List<Either<e, a>>) -> List<a>
- partition_eithers : forall e a. (List<Either<e, a>>)
-> Pair<List<e>, List<a>>
lefts/rights use depth-2 nested Ctor patterns (Cons (Left l) t)
— exercises 16a's desugar pass at a depth not reached by any
prior fixture.
- examples/std_either_list_demo.{ailx,ail.json}: drives all three
combinators on a five-element List<Either<Int, Int>>; expected
output one per line: 2, 3, 2, 3.
- crates/ail/tests/e2e.rs::std_either_list_demo: e2e count 34 → 35.
- docs/JOURNAL.md: Iter 15g entry.
Compiler bug surfaced (queued as 15g-aux, not fixed):
unify_for_subst in ailang-codegen/src/lib.rs accepts $u wildcards
only on the arg side. Mixing inline (Either Left n) and (Either
Right n) in a list literal lands $u on the param side via the
outer List<a>'s binding and errors. Reduced repro: `length [Left
1, Right 10]`. Demo works around with monomorphic mkleft/mkright
helpers; workaround documented inline. Fix is a one-line symmetric
extension of the early-return.
Tests: 94/94. Stdlib: 5 modules, 27 combinators.
After six feature iters since the last docs sweep, DESIGN.md had
visible drift. Patched in place rather than queueing the next
codegen-heavy iter against a stale spec.
DESIGN.md changes (+72 LOC):
- Decision 6: form (A) marked shipped (Iter 14c, sole projection
since 15e); body kept as audit trail.
- Pipeline: desugar pass added between resolve+hash and typecheck;
invariant noted that CheckedModule.symbols hashes from the
original module, not the desugared one (so ail diff/manifest
preserve on-disk identity).
- CLI: added deps, diff, workspace, builtins (shipped earlier but
never doc'd).
- "What is not (yet) supported": re-anchored from "end of Iter 13"
to "as of Iter 16a"; removed lifted gates (cross-module ADTs,
no-GC, flat-pattern-only); added tighter follow-up gates
(literal sub-patterns, local recursive let).
- "What IS supported": promoted nested Ctor patterns (16a),
cross-module ADTs (14h), form-(A) text surface (14b/14c/15e),
Boehm GC (Decision 9 / 14f) into the smoke-test list.
- Smoke tests: added std_list_demo, std_maybe_demo, std_either_demo,
std_pair_demo, nested_pat fixtures.
JOURNAL: 16a-aux entry recording the drift sites and what was
explicitly *not* changed (Goal, Decisions 1-5, 7-9, Mangling,
Convention, Data model, Verification — spot-checked, all current).
Tests: 93/93 unchanged (doc-only). Build clean.
Fourth stdlib module. Pair<a, b> is the canonical product with two
type vars and a single constructor MkPair. Five combinators: fst,
snd, swap, map_first (Pair<a, b> -> Pair<c, b>), map_second
(Pair<a, b> -> Pair<a, c>).
Smallest dogfood for the parameterised-ADT path so far: no
recursion in the data def or any combinator, single-arm matches
throughout. Demo prints 7, 9, 9, 7, 8, 18 deterministically.
No compiler bug surfaced (fourth stdlib iter in a row to land
clean). The only fixable issue was a paren-balance typo in the
demo's seq chain.
Cumulative: 4 stdlib modules, 24 combinators. Type-system surface
exercised end-to-end now spans 1/2/3-type-var data, 1/2/3-type-var
fns, recursive ADTs, cross-module imports of all of the above,
flat and nested patterns, TCO via monomorphised musttail, GC.
Tests: 93/93 (e2e 33 → 34).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Lifts the gate that rejected `(pat-ctor Cons a (pat-ctor Cons b _))`.
Approach: a pure AST → AST pass `ailang_core::desugar::desugar_module`
that flattens nested-Ctor sub-patterns into chains of single-level
matches with let-bound fresh vars and duplicate fall-through. Both
ailang-check and ailang-codegen call the pass at pipeline entry.
Hash-relevant canonical bytes are untouched: the pass runs after
load_module, in memory only. `check`'s returned CheckedModule.symbols
still carries hashes of the original defs so `ail diff` / `ail manifest`
see on-disk identities.
Lit sub-patterns inside a Ctor still rejected — that's the narrower
scope of `nested-ctor-pattern-not-allowed` post-16a; the variant doc
reflects the new meaning.
Fresh-name safety: `$` is a valid ident character in form (A), so
the Desugarer pre-walks every Term::Var / Pattern::Var name into a
BTreeSet and bumps its counter past any user-spelled `$mp_N`.
Already-flat matches go through `is_flat()` and emit identical AST
shapes; every existing fixture produces the same output as before.
New: examples/nested_pat.{ailx,ail.json} prints 30 from
first_two_sum on [10, 20, 30]. e2e test
`nested_ctor_pattern_first_two_sum` guards the path.
Tests: 92/92 (e2e 32→33, ailang-core unit 10→12).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The CLI help text claimed `render` was symmetric to `parse`, but
`render` was wired to ailang_core::pretty::module (an older
human-pretty form) while `parse` consumed form (A). Two different
"text projections" coexisted under one name.
Fix: Cmd::Render and Cmd::Describe (both branches) now call
ailang_surface::print, the actual inverse of `parse`. Round-trip
gate via `ail render | ail parse` is now an e2e test in
crates/ail/tests/e2e.rs in addition to the existing unit-level
round-trip across all 25 fixtures in ailang-surface/tests.
Cleanup: `ailang_core::pretty::module` and its three private
helpers (`def_block`, `term_block`, `term_inline`) deleted —
~260 LOC of duplicate-purpose code removed. pretty.rs goes
from 457 to 196 LOC. Surviving surface is intentionally narrow:
manifest, type_to_string, pattern_to_string — the diagnostic
stringification used in error messages, where one-line ML
notation reads better than form (A)'s nested S-expressions.
Module-level doc rewritten to nail down the single-purpose
framing.
Tests: 89/89 (e2e +1, ailang-core unit -1 since the deleted
test exercised the deleted fn).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Third stdlib module. Either<e, a> is the first 2-type-var data def,
and the eliminator `either : (e -> c) -> (a -> c) -> Either<e, a> -> c`
the first fn with three type vars on top of the data — the deepest
polymorphism shipped end-to-end so far.
Five combinators: from_right, is_left, is_right, map_right, either.
Demo exercises every one and runs to deterministic output. IR shows
six distinct monomorphisations including two `from_right` variants
(Left=Int vs Left=$u depending on call site) and `either__I_I_I`.
No new compiler bugs surfaced: 14a / 14h / 15b coverage was wide
enough to handle 2-type-var data plus 3-type-var fns out of the box.
Discovered (queued as 15e): `ail render` and `ail parse` are not
symmetric. Form-(A) printer in ailang_surface::print is the actual
inverse of `parse` (round-trip test covers it across all 25 fixtures
including std_either) but is not exposed via the CLI; `render` still
calls the older ailang_core::pretty::module printer.
Tests: 89/89 (was 88, +1 e2e for std_either_demo).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Small empirical iter confirming the 14e tail-call story works
dogfood-practical. New fixture std_list_stress.ailx builds a
1000-element List<Int> via recursive build, folds it with both
fold_left (tail-marked in std_list) and fold_right (constructor-
blocked, unmarked), prints both sums (500500 each).
IR evidence at the monomorphised fold-recursion sites:
fold_left__I_I: musttail call i64 @ail_std_list_fold_left__I_I(...)
fold_right__I_I: plain call i64 @ail_std_list_fold_right__I_I(...)
The tail: true marker on std_list.fold_left's recursive call
survives monomorphisation through the __I_I specialisation.
fold_right correctly emits a plain call (recursive call is
second arg to f, not tail position).
Empirical findings at N=1000:
- fold_left runs in constant stack (musttail).
- fold_right runs in ~1000 frames. No segfault; default 8MB
Linux stack absorbs it comfortably.
- build (recursive, unmarked) likewise fits.
- End-to-end ~40ms wall (build + clang link); program <1ms.
- Boehm GC handles 2 × 1000 Cons allocations without symptom.
Tests 87 -> 88. Cumulative 30 e2e tests. cargo doc 0 warnings.
Cumulative state, post-15c: 2 stdlib modules (std_maybe,
std_list), 14 combinators, cross-module recursive ADTs working,
TCO surviving monomorphisation, GC at 1000-element scale. Four
compiler bugs surfaced + fixed in dogfood since 14a.
Natural pause point. Queue for future iters: 15d (std_either),
15e (std_pair), 16a (nested patterns), 16b (local rec let),
17a (per-fn arena, Decision 9 future-iter). None block further
stdlib work.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Second stdlib module. Tester wrote std_list.ailx (10 combinators,
164 LOC) and a consumer demo. std_list typechecked standalone;
demo did not, surfacing three compiler bugs:
1. Check-side: Iter 14h's qualify_local_types was applied to
Term::Var cross-module lookup but not to ctor-field types in
Term::Ctor synth or Pattern::Ctor resolution. First recursive
cross-module ADT (List has Cons a (List a) — recursive Con
self-ref) triggers the bug. std_maybe slipped through because
Maybe's ctors have no recursive Con field.
2. Codegen-side: same gap mirrored across 4 sites in codegen
(Term::Ctor synth, lower_ctor, lower_match) plus a tweak to
unify_for_subst (recurse on re-bind instead of strict equality
so sibling-derived List<Int> accepts nullary-ctor's List<$u>
wildcard).
3. Const codegen: emit_const rejected non-literal const bodies.
The demo's xs : List<Int> = Cons 1 (...) requires it. Fix:
per-module const table, Term::Var resolution loads literal
consts from global, inlines non-literal bodies. Bare and
qualified refs both supported.
All three fixes carry an "Iter 15b" code comment at their site.
~349/25 LOC across ailang-check, ailang-codegen, e2e.rs.
Tests 85 -> 87. New e2e std_list_demo asserts 11-line stdout:
length 5, is_empty false/true, head via from_maybe, tail length,
append length, reverse head, map double head, filter is_even
length, fold_left sum, fold_right sum. New ailang-check unit
test cross_module_recursive_adt_term_and_pat_ctor covers both
the original bug and the symmetric pat-ctor latent twin.
Hash invariance: all pre-15b fixtures + std_maybe defs
bit-identical. 14a / 14e / 14h regressions all green.
Cumulative state: 2 stdlib modules (std_maybe, std_list), 14
combinators, cross-module recursive ADT working end-to-end.
Three compiler bugs surfaced + fixed in dogfood since 14a (each
dogfood iter has surfaced ≥1).
Authoring observation: form (A) at 10 combinators is fine; main
friction is paren-counting in nested seq chains, not the form
itself. n-ary seq would help but is sugar.
Plan 15c: 1000-element list stress test for fold_left (tail-call-
marked) vs fold_right (constructor-blocked).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The 15a tester surfaced a real compiler limitation: cross-module
type and ctor references were not implemented. Iter 5b only
carried fns + consts via module_globals; types/ctors stayed
module-local with an explicit DESIGN comment. This iter completes
the cross-module mechanism using the Iter-5b convention:
qualified-only access via module.Name.
Implementation:
- ailang-check: Env.module_types populated by build_module_types.
Qualified resolution in Type::Con, Term::Ctor, with cross-module
fallback for pat-ctor (local wins, multi-import collision -> new
ambiguous-ctor diagnostic). Four new unit tests.
- ailang-codegen: workspace-level module_ctor_index replaces
per-Emitter table. lookup_ctor_by_type / lookup_ctor_in_pattern
thread qualified type names through box-tag and field-type
resolution.
- examples/std_maybe_demo.{ailx,ail.json}: type-name slots now
qualified (std_maybe.Maybe).
- New e2e test cross_module_maybe_demo asserts the demo prints
["7","99","true","true","42"].
Net diff ~550 LOC. Tests 80 -> 85. All Iter 14a regressions
(parameterised_box_round_trip, parameterised_maybe_match,
list_map_poly_inc_then_prints, polymorphic_id_at_int_and_bool)
verified green — the 14h derive_substitution change (default
unpinned forall vars to Unit for monomorphiser) sits on a
different layer than 14a's $u-wildcard fix and they coexist.
Hash invariance: all five std_maybe def hashes unchanged. All
80-test-suite fixtures retain bit-identical hashes — cross-module
support is purely additive at the language level.
Process note: the std_maybe.ailx file landed in the 14g commit
via a sloppy git-add-A; should have spotted it before staging.
Not a correctness issue but a hygiene one.
Implementer flagged Unit-default monomorphisation as wasteful-
but-correct; rethink if stdlib grows toward overload-resolution-
style cases needing distinct unconstrained instantiations.
std_maybe stdlib effectively ships: module + four combinators +
e2e-tested consumer demo. Plan 15b: std_list importing std_maybe,
exercising Maybe-returning head/tail and tail-call-marked
fold_left.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reconsidered 14d's removal of Term::If. The decision was wrong.
"No redundancies" requires judgment; reducibility (if -> match)
is not redundancy in the strong sense. Term::If is a primitive
control-flow shape; bool branching is the second most common
shape after sequencing, and removing it cost 3x tokens on every
branch site (`(if c a b)` 4 tokens vs the match-on-Bool form
12 tokens).
Meta-pattern fixed: I had been treating user observations as
directives. User said "if is a subset of match"; I jumped to
remove it citing CLAUDE.md, with no independent conviction.
The leak appeared in 14f's JOURNAL prose ("three lines for what
if used to do in one"), which read as regret. Two feedback
memories saved (memory/feedback_user_suggestions_not_directives,
memory/feedback_no_nostalgia_for_removed_features) to head this
off.
Implementation: mechanical reverse-application of 14d's diff at
every site (AST, check including the 14e tail-position arm,
codegen 4 sites, surface parser/printer, pretty, CLI walker,
e2e test mutation). Removed lower_bool_match helper — it existed
only because 14d's migration shape needed codegen for non-ptr
match scrutinees; with Term::If back, match-on-Bool returns to
its pre-14d unsupported state. Three fixtures (sum, sort, max3)
restored to pre-14d shape. gc_stress (added in 14f) also
migrated back to (if ...) since it was authored under the wrong
constraint.
14e (musttail) and 14f (GC_malloc) verified intact in IR.
Hashes restored to pre-14d values:
- sum.sum: db33f57cb329935e
- sort.insert: 697fcb9f30f8633a
- max3.max: 65c45d6a45dd0a72
- max3.max3: 624b14429bf302f5
All other defs across all 18 fixtures keep their post-14f
hashes. Tests 80/80 green; cargo doc 0 warnings. LOC delta
+265/-295 net -30.
DESIGN.md Decision 7 preserved with a "Status: REVERTED" header
for audit trail. Form-(A) `if-term` production restored.
Plan: back to 15a (std_maybe stdlib module).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Decision 9 ships. Through Iter 14e every ADT box, lambda env, and
closure pair was leaked. This iter substitutes GC_malloc for malloc
in all four IR allocation sites and links -lgc. No language change,
no AST change, no schema change.
Diff: 5 files modified, ~30 LOC net.
- codegen/lib.rs: 4 substitutions @malloc -> @GC_malloc.
- ail/main.rs: .arg("-lgc") in the clang invocation.
- 5 IR snapshot files: mechanical s/@malloc/@GC_malloc/, 9
occurrences. IR is bit-identical to pre-14f modulo this
substitution — exactly Decision 9's promise.
- e2e.rs: new test gc_handles_recursive_list_construction.
- examples/gc_stress.{ailx,ail.json}: new fixture, builds a 50-
element list via recursive Cons, sums it (1275).
Hash invariance verified: every existing fixture def hash
unchanged (codegen and link line are downstream of canonical
bytes; AST didn't move).
Tests 79 -> 80, all green. Existing 79 byte-identical stdout.
gc_stress -> 1275. list_map_poly -> 2/3/4 unchanged. sort
sorted-list unchanged. cargo doc 0 warnings.
GC notes (pertinent to future work):
- GC_INIT() not needed on Arch libgc 1.5.6 (auto-init via
__attribute__((constructor))).
- No conservative-scan over-retention observed.
- -lgc alone sufficient for link (pthread/dl transitive).
Pattern-shape note from gc_stress fixture writing: the post-14d
"if-then-else" replacement is `(match (app == n 0) (case
(pat-lit true) ...) (case (pat-wild) ...))`. Three lines for
what `if` used to do in one, but uniform with the language.
Worth flagging for the stdlib brief.
Language is feature-complete enough for stdlib. The three
blockers identified at the 14b boundary (redundancy 14d, tail
calls 14e, GC 14f) are all done. Plan 15a: first stdlib module
std_list.ailx with length/append/reverse/map/filter/fold_left/
fold_right/head/tail/is_empty.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Decision 8 ships. Term::App and Term::Do gain tail: bool with
serde-default false and skip-when-false serialisation. New
typecheck pass verify_tail_positions enforces tail-position rules
(Scheme-style propagation through match arms, seq.rhs, let body,
lam body). Codegen emits musttail call for marked App calls.
Hash invariance verified: only the two migrated print_list defs
(list_map_poly.print_list, sort.print_list) changed hashes; all
other defs across all 18 fixtures kept bit-identical hashes —
confirms the skip-when-false serialisation rule works.
Tests 76 -> 79: tail_call_in_non_tail_position_is_rejected,
tail_call_in_tail_position_is_accepted, plus an IR-grep e2e test
asserting that print_list's recursive call site emits musttail
in the lowered IR. Existing 25 e2e tests unchanged in behaviour
(map -> [2,3,4], sort -> sorted list).
IR evidence at the recursive site:
%v7 = musttail call i8 @ail_list_map_poly_print_list(ptr %v6)
ret i8 %v7
Two deviations called out in the implementer report and JOURNAL:
1. tail-do uses tail call, not musttail. Cross-type return
(runtime helpers return i32, AILang Unit is i8) would have
LLVM reject musttail. Path is implemented but not exercised
by any current fixture; proper fix is runtime-helper signature
change, punted.
2. block_terminated flag in codegen so tail-call emit
(musttail call + ret) doesn't get a duplicate trailing ret
from surrounding code (match-arm phi, fn-body, lambda thunk).
Internal plumbing; required for IR well-formedness.
Form (A) productions now at ~30, exactly the constraint-1
budget. Future surface additions need to retire something or
explicit-budget-rebalance in DESIGN.md.
GC notes from implementer survey land in JOURNAL:
- Allocations cluster in lower_ctor; every term-ctor does
malloc(8+8n).
- Tail recursion does not reduce alloc pressure, only stack.
For map-style ctor-blocked recursions, allocation IS the
bottleneck.
- Per-fn arena is sound only when fn return type contains no
boxed ADT. Most current fixtures violate this.
Plan 14f: Boehm conservative GC (GC_malloc, -lgc) as a first
cut. Single-iter integration, no AST/schema change. Stress
test: build a 100k Cons list, observe RSS doesn't blow up.
After 14f the language is feature-complete enough for stdlib
work (15a).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Term::If was semantically a subset of Term::Match on Bool. Per
CLAUDE.md the language must contain no redundancies; two AST nodes
for the same operation produces an authoring decision with no
semantic content and a duplicate codegen path. Removed.
Migration shape (applied to sum, sort, max3 fixtures):
(if c a b) -> (match c (case (lit-bool true) a) (case _ b))
No schema version bump (per user direction): no third-party consumes
ailang/v0, so version ceremony is pure overhead. Edited AST and
fixtures in place; pinned hashes in hash.rs updated.
Implementer deviation, called out and justified: a tightly-scoped
lower_bool_match helper (~95 LOC) was needed in codegen because
the existing match path rejects i1 scrutinees and Pattern::Lit.
Helper accepts only the canonical two-arm migration shape, errors
on anything else, emits the same br/phi IR Term::If used to. No
generalisation of the ADT-match codegen.
Diff: 13 files, +286/-221 (net +65 LOC). AST got smaller
(one variant gone), form-(A) got smaller (one production gone),
typecheck got smaller (one branch gone). Codegen got slightly
larger by the bool-match helper.
Hash deltas: sum.sum, sort.insert, max3.max, max3.max3 changed.
All other defs (e.g. sum.main, sort.IntList, sort.sort,
sort.print_list, max3.main) kept bit-identical hashes — confirms
canonical-JSON byte format intact.
Verification: 76/76 tests green; sum->55, max3->17, sort->[1,1,2,
3,3,4,5,5,5,6,9] (identical to pre-migration). cargo doc 0 warnings.
Tail-call survey by implementer (informs 14e): print_list
recursions are already in tail position (rhs of seq inside match
arm); map/sort/insert recursions are NOT (constructor-blocked
inside Cons applications). 14e annotation will benefit terminal
recursions; ctor-blocked ones need accumulator-form rewrites in
source, not a compiler-side transform.
Decision 7 added to DESIGN.md. JOURNAL entry has the language-
completion sequence (14d done, 14e tail-calls, 14f GC, 15a stdlib).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Strictly additive new crate per Decision 6 architectural pin.
JSON-AST stays the source of truth; ailang-surface is one
producer/consumer of ailang_core::ast::Module values. ailang-check
and ailang-codegen unchanged.
Crate contents:
- src/lex.rs (~264 LOC): 3-rule lexical core. Whitespace and parens
delimit tokens; semicolon to EOL is comment; first-character
classifier (digit -> int, " -> string, else -> ident).
- src/parse.rs (~1041 LOC): hand-written recursive descent. One Rust
fn per EBNF production. No parser-combinator dep.
- src/print.rs (~371 LOC): deterministic pretty-printer. Round-trip
contract with parse() is the surface's correctness gate.
- tests/round_trip.rs (~128 LOC): integration test runs every
examples/*.ail.json fixture through print -> parse -> canonical
JSON, asserts canonical-byte equality with the original.
Two AST-driven form widenings beyond the 14b sketch (both folded
into DESIGN.md Decision 6):
- lam-term carries (typed name type) params, ret type, and
optional effects (Term::Lam has parallel param_tys/ret_ty/
effects fields).
- import-clause admits (import name (as alias)?) (Import.alias
is Option<String>).
Production count ~28, under 30-rule budget. Constraint 1
(formalisable for foreign LLM) intact.
Verification:
- cargo build --workspace green.
- cargo test --workspace: 76 tests green (was 64; +9 surface unit,
+2 round-trip integration). All 17 fixtures round-trip
byte-identical at canonical level. 3 hand-written .ailx
exhibits parse to canonical JSON identical to their .ail.json
siblings.
- cargo doc --no-deps zero warnings (DESIGN.md item 6 invariant).
Manual smoke test (ail parse → ail run): hello, box,
list_map_poly all produce expected output through the form-(A)
authoring lane end-to-end.
CLI: ail parse <file.ailx> [-o <file.ail.json>]. .ail.json
remains a first-class input to every existing subcommand.
Plan 14d: stdlib (std_list.ailx with length/filter/fold/concat/
reverse/head/tail), authored in form (A) from day one.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
User clarified the architectural intent: the data structure is
the source of truth. The textual surface is one projection
among potentially many — visual / graphical front-ends are an
explicitly anticipated future lane.
The previous draft of Decision 6 said "JSON-AST demoted to
storage and exchange only", which was the wrong framing. The
JSON-AST stays canonical; form (A) is the AI authoring
projection but does not displace the AST.
Changes:
- New "Architectural pin: data structure is the source of truth"
subsection.
- Implementation outline reworded: parser is "strictly additive",
no schema changes, .ail.json remains a first-class input to
every existing subcommand.
- New "What this Decision deliberately does not do" subsection
to lock in the architectural commitments.
No code change. Pure documentation correction to the 14b
design pass before 14c starts.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
User redirected at iter boundary: writing a stdlib in JSON was
the wrong move. The language is supposed to be the one I program
*best* in, and JSON-AST authoring is rationalisation, not
strength.
DESIGN.md Decision 6 captures the constraints that fall out of
the "formalisable for a foreign LLM" hard requirement (no
precedence, no semantic indentation, ASCII only, every AST node
a uniquely-tagged form), sketches three candidate notations with
the same `map` encoded in each, and picks form (A) — fully-tagged
S-expressions — as the first attempt with explicit rollback path
to form (C) if (A) hurts authoring.
Form (A) shape:
- 3-rule lexical core: sexpr / atom / token-classified-by-
first-character.
- Every AST node has a unique head keyword. No case-rule (no
"capitalised head means ctor"); ctors are explicit via
`(term-ctor TypeName CtorName args)` and `(pat-ctor CtorName
fields)`.
- Bare atoms get their sort from the parent slot (type-var
inside `(con NAME args)`, term-var inside `(app HEAD args)`,
pat-var inside `(pat-ctor CTOR fields)`, integer literal in
term position, etc.).
Empirical exhibits, hand-encoded:
- examples/hello.ailx 5 LOC (JSON was 36 pretty / 21 canonical)
- examples/box.ailx 25 LOC (JSON was 160 / 88)
- examples/list_map_poly.ailx 50 LOC (JSON was 394 / 230)
4-8x line reduction, ~4x character reduction. Bigger gains on
bigger programs since overhead is proportional to AST depth.
None are parseable yet — header comments say "Iter 14b design
exhibit, parser lands in 14c".
Two small spec issues caught while writing the exhibits and
folded back into DESIGN.md before committing:
- Operator idents (`+`, `==`) need the token-by-first-char
classification rule, not a word-shaped regex.
- Bool literals (`true`/`false`) reserved in term context;
unit is explicit `(lit-unit)`.
Tests unchanged (this iter is paper). 25/25 e2e green.
cargo doc --no-deps zero warnings.
Plan 14c: new crate `ailang-surface` with PEG parser, round-trip
hash-equivalence gate against every existing `examples/*.ail.json`,
CLI subcommand `ail parse`. If round-trip holds, stdlib starts
in `.ailx` form (Iter 14d).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Dogfood payoff for parameterised ADTs (13a/b/c). Adds the first
program to nest a nullary ctor of a parameterised ADT inside a
parent ctor (Cons(Int, Nil)) and to call a polymorphic recursive
higher-order fn over a recursive parameterised ADT.
Fixture (examples/list_map_poly.ail.json):
- data List a = Nil | Cons a (List a)
- inc : (Int) -> Int = \x. x + 1
- map : forall a b. ((a) -> b, List<a>) -> List<b>
recursive, instantiated at (Int, Int)
- print_list : (List<Int>) -> Unit !{IO}, recursive
- main builds [1,2,3], maps inc, prints each: "2", "3", "4"
E2E test list_map_poly_inc_then_prints in crates/ail/tests/e2e.rs.
Bug fixed (crates/ailang-codegen/src/lib.rs, +30/-9):
synth_arg_type used Type::unit() as placeholder for ADT type
vars that the ctor's args couldn't pin (Nil for List<a>).
Inside Cons(Int, Nil), unify_for_subst then bound a=Int from
the head and collided with a=Unit from the tail. Replaced the
placeholder with a synth-only wildcard Type::Var{name:"$u"}
mirroring the checker's $m metavar convention; unify_for_subst
short-circuits on $u-prefixed arg-side vars (accept without
binding, let a sibling pin the var).
No schema or API change. No new variant. Tester's recursion
hypothesis was refuted by debugger via a non-recursive
Cons(7, Nil) repro before the fix landed.
Tests: 25/25 e2e (was 24). All Iter 12/13 regressions green.
cargo doc --no-deps zero warnings (workspace invariant from
13d/e/f preserved).
Three-agent flow (tester -> debugger, no implementer needed
since the fix was inside debugger's <50 LOC scope). Process
note in JOURNAL.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Third docwriter mission (combined). Pure rustdoc additions, no
API or behaviour change.
ailang-codegen (113 LOC of doc):
- Crate root: intra-doc links (emit_ir, lower_workspace,
CodegenError::MissingEntryMain) + precondition note (both
entry points assume type-checked input).
- /// on CodegenError + every variant, naming AST trigger.
- /// on emit_ir and lower_workspace (single vs multi-module
split, cross-linked).
ail CLI (49 LOC of doc):
- Module-level //! expanded from 5 lines to full subcommand
list with one-liners (11 subcommands verified against Cmd
enum), clang-on-PATH note, design-intent paragraph.
Verification: cargo doc --no-deps zero warnings (also under
RUSTDOCFLAGS='-D rustdoc::broken_intra_doc_links'); build green;
tests 64/64 + 3 ignored doctests green. Diff is 100% doc lines
(verified by filtering).
Findings (not fixed; orchestrator-deferred):
- CodegenError::Internal is one catch-all variant for ~30
invariant-violation sites; splitting would help test
ergonomics but is out of doc scope.
- emit_ir synthesises a Workspace with root_dir="."; harmless
today, surfaces if codegen ever reads root_dir.
Process note: my brief said the CLI had 9 subcommands; agent
found and documented 11. Useful counter-pressure on orchestrator
sloppiness — recorded in JOURNAL.
Workspace-wide rustdoc invariant (DESIGN.md item 6) now
load-bearing across all four crates. Docwriter shifts from
sweep to maintenance mode going forward.
Next: 14a (polymorphic list_map using Iter-13a parameterised
ADTs) is unblocked.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Second docwriter mission. Pure rustdoc additions (no API or
behaviour change) across the typechecker crate:
- builtins.rs: module root expanded; EffectOpSig (struct + 3
fields) and install() got /// strings.
- diagnostic.rs: Severity (+ both variants), Diagnostic (+
severity/code/message fields) and the error/with_def/with_ctx
helpers got /// strings; super:: link rewritten to crate::.
- lib.rs: CheckError + every variant, to_diagnostic, CheckedModule
(+ symbols), check, Env (+ globals/effect_ops/types/
module_globals/current_module), CtorRef (+ type_name) got ///
strings; crate-root prose upgraded with intra-doc link to
check_module.
Verification: cargo doc --no-deps zero warnings; cargo build
--workspace green; cargo test --workspace 64/64 + 3 ignored
doctests green. Diff is 188 LOC, all in /// or //! lines (verified
by filtering).
Findings (not fixed; orchestrator-deferred):
- Env is pub but only privately constructable.
- CheckError::CtorArity and ::ArityMismatch share the public
diagnostic code "arity-mismatch" by design.
- Diagnostic / Severity reachable via two paths because the
diagnostic module is pub.
JOURNAL.md updated with the Iter 13e entry. 13f (ailang-codegen
+ ail) and 14a (List a rewrite) remain queued.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds ailang-docwriter to /agents/ — a recurring role for keeping
crate-, module-, and pub-item-level rustdoc accurate. First mission:
ailang-core. Crate root, every module root, every pub item documented;
intra-doc links throughout; Iter-13a additions (TypeDef.vars, Type::Con.args)
get an explicit backwards-compat note. Two stale broken-link warnings in
ailang-check fixed in passing. cargo doc --no-deps now warning-free across
the workspace; promoted to verification invariant 6 in DESIGN.md.
DESIGN.md: removes "No parameterised ADTs" from the gap list,
adds an entry under "what is supported" that names the per-use-
site substitution scheme and the `llvm_type(Type::Var)` hard-error
defence. Smoke-test list extended with `box.ail.json` and
`maybe_int.ail.json`. Boundary snapshot moved from "end of Iter 12"
to "end of Iter 13".
JOURNAL.md: single Iter 13 entry covering 13a/b/c. Records the
hash-invariant regression test, the architect-flagged debt I
deliberately did not touch (poly-fn-as-value asymmetry, builtins
triple-source, `synth_arg_type` shortcuts on If/Match), the
KISS observation that 13b's "no mono-queue for types" was the
right call, and the process note that this was the first iter
worked strictly through `/agents/` after the role pin in 3df943d.
Plan iteration 14 queued: list_map-as-`List a` rewrite, GC/arena,
poly-fn-as-value.
Closes the gap between Iter 13a (parameterised ADTs in the
checker) and end-to-end execution. ADT ctor code stays inlined at
every use site — no mono-queue for types, no new symbols — but
LLVM field types are now derived per use site via substitution.
`CtorRef` gains `type_vars: Vec<String>` so use sites can identify
which fields reference rigid vars. `cref.fields` (precomputed
LLVM strings) is documented as monomorphic-only and read past for
parameterised ADTs.
`lower_ctor`: derives a `BTreeMap<String, Type>` substitution from
`synth_arg_type` of each arg via `unify_for_subst`, then maps every
`ail_field` through `apply_subst_to_type` + `llvm_type` for the
per-store types. Monomorphic ADTs hit the original fast path.
`lower_match`: builds `arm_subst` from `s_ail.args` ↔ `cref.type_vars`,
substitutes through each `cref.ail_fields[idx]`, and uses the
substituted type both for the LLVM `load` AND as the AILang slot
of the local — so a downstream `unbox(b)` sees `b: Int`, not
`b: a`.
`synth_arg_type` for `Term::Ctor`: returns concrete type-args
derived from the ctor's term-args. Vars left unpinned (e.g. `None`
for `Maybe a`) fall back to `Type::unit()`.
`llvm_type(Type::Var)` now hard-errors. Earlier this silently
fell through to `ptr` (the ADT-via-ptr fallback), producing
garbage IR. Failing loudly here surfaces missed substitutions in
the test suite.
Two e2e tests (`box.ail.json`, `maybe_int.ail.json`) cover ctor
lower with substituted field types and match-arm field
substitution. 64 tests green; clippy clean (two pre-existing
warnings untouched). Hash invariant holds.
Out of scope per agent assignment: poly-fn-as-value, higher-rank
polymorphism, DESIGN/JOURNAL updates (deferred to 13c).
Adds an explicit "My role: orchestrator" section so the discipline
doesn't drift across sessions. Three concrete contracts:
- What I do myself (planning, design, JOURNAL/DESIGN) vs. what gets
delegated (implementation, testing, debugging, architectural
drift review). Trivial mechanical edits stay inline.
- Authority over /agents/: I may add, edit, retire, or replace agent
definitions when orchestration needs change, with the change going
through git like any other code.
- When NOT to delegate: direct user questions, single judgement
calls, and cases where context is already loaded.
Triggered by a session where I quietly took over implementer work
on Iter 13a instead of routing it through ailang-implementer.
Adds type parameters to ADTs. `TypeDef` gets a `vars: Vec<String>` and
`Type::Con` gets an `args: Vec<Type>`, both `skip_serializing_if =
"Vec::is_empty"` so canonical-JSON hashes of every pre-13a definition
stay bit-identical (regression test in `hash::tests`).
Checker:
- `check_type_def` installs `td.vars` as rigid vars while validating
ctor field types — `Cons(a, List a)` now resolves both occurrences
of `a` and the recursive `List a` use.
- `check_type_well_formed` accepts `Type::Con { name, args }` only
when the type is in scope and `args.len()` matches its declared
arity; primitives stay zero-arg.
- `Term::Ctor` synth instantiates `td.vars` with fresh metavars, so
`MkBox(42)` synthesises to `Box<$m0>` and unifies field types
through the surrounding context.
- `type_check_pattern` substitutes the scrutinee's concrete type-args
through ctor field types, so a `MkBox(x)` arm against a `Box<Int>`
scrutinee binds `x : Int`.
- `check_fn` validates declared param/return types via
`check_type_well_formed` so arity mismatches on parameterised
ADTs surface before the body is checked.
`unify`, `occurs`, `Subst::apply`, `substitute_rigids`, and codegen's
`unify_for_subst` / `apply_subst_to_type` recurse into `args`.
Pretty-print:
- `type_to_string` renders `Box<Int>` for parameterised cons.
- `def_block`/`manifest` carry the `[a b ...]` vars list.
Three new check unit tests cover ctor instantiation at a concrete
arg, the polymorphic-`unbox` round trip at two distinct
instantiations, and arity mismatch on `Box<Int, Bool>`. All 62 tests
green; clippy clean (two pre-existing warnings untouched). Hashes
db33f57cb329935e (sum) and b082192bd0c99202 (IntList) verified
unchanged.
Codegen still synthesises `Type::Con` with `args: vec![]` from
`Term::Ctor` — full ADT monomorphisation lands in 13b.
DESIGN.md:
- "What is not (yet) supported" updated for end-of-Iter-12 boundary:
parameterised ADTs replace HM-inside-bodies as the headline gap;
polymorphism limitations (direct-call-only, no higher-rank) are
spelled out so future me doesn't trip over them.
- "What is supported" gains the polymorphism + monomorphisation
bullet, with the descriptor scheme written out.
- Smoke-test list extended with poly_id and poly_apply.
JOURNAL.md: Iter 12a/b retrospective. Notes the metavar-encoding
choice (Type::Var{name:"$m<n>"} vs new variant) and why the
codegen-side type tracker was preferable to a typechecker
sidetable for now. Architecture self-check confirms the language
is now usable for poly-flavoured programs over primitives. Plan 13:
parameterised ADTs as the natural next step (without them, generic
map remains hand-monomorphised).
Iter 12c was originally to include a polymorphic map rewrite —
dropped because parameterised ADTs are the prerequisite. The two
new examples (poly_id, poly_apply, both already in 12b) cover the
real e2e proof.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Brummel