Files
Brummel 76b21c00eb feat(lang): eliminate the Implicit ownership default — totality + the drop-soundness it demasks (#55)
Deletes `ParamMode::Implicit`. `ParamMode` is now `{Own, Borrow}`:
every fn-type slot on every signature carries an explicit `own` or
`borrow`, no defaulted position survives anywhere (model 0008 §2,
spec 0062). The parser rejects a bare fn-type slot; `borrow-return`
and `borrow-over-value` reject at the signature; the corpus is
migrated to minimal-ownership modes (consumed ⇒ own, read-only-heap
⇒ borrow, value ⇒ trivial-own). The documented `Implicit`-ret-mode
leak is fixed: an owned heap return now drops exactly once (live=0,
acceptance criterion 5).

This was the easy half. Removing the default ACTIVATED a family of
drop paths that `Implicit` had silently skipped — the pre-cutover
language was leaking (and in places mis-dropping) here rather than
crashing, because an Implicit scrutinee turned the drop off. Making
the modes explicit (Own) turned those paths on and exposed two
latent-bug clusters, all fixed RED-first as part of this cutover:

Drop-soundness family (four legs):
  A. lit-sub-pattern double-free — the desugar re-matched the same
     owned scrutinee in the lit fall-through; fixed by grouping
     consecutive same-ctor arms into one match (bind fields once),
     in ailang-core desugar.
  B. Cons-husk leak on non-tail arm bodies — the lit-sub-pattern
     desugar rebound the owned scrutinee via `Let $mp = xs`, which
     bumped consume_count and suppressed the existing fn-return
     partial_drop. Fixed by not rebinding a bare-Var scrutinee
     (one husk-freeing mechanism, not two).
  C. polymorphic `drop_<T>` rc_dec'd monomorphised value fields —
     the per-ADT drop fn was emitted once from the polymorphic
     TypeDef, defaulting type-var fields to ptr and rc_dec'ing
     inline Ints (segfault). Fixed with per-monomorph drop
     functions (new ailang-codegen::dropmono): the drop set is
     collected from the lowered MIR, value-type fields are skipped,
     heap fields still freed once; monomorphic-concrete ADTs keep
     their byte-identical un-suffixed drop symbol.
  D. static Str literal passed to an `(own Str)` param — the
     literal lowers to a header-less rodata constant; the callee's
     now-active rc_dec read its length field as a refcount and
     freed a static address (segfault). Fixed with the missing
     fourth StrRep::Static→Heap promotion in lower_to_mir's App arm,
     gated on Own mode (borrow args stay static, no regression).

over-strict-mode lint over-fired: it suggested `(borrow V)` for
value-typed params (which `borrow-over-value` rejects — own is the
only legal mode there) and fired on `(intrinsic)` bodies (whose
consumption the linearity walk cannot observe). Tightened to skip
both; contract 0008 updated to the narrowed firing scope.

Irreversible step — canonical-form hash reset (model 0008 §6,
acceptance criterion 6). Every signature now carries explicit modes,
so the hashable canonical JSON changed for every module. RATIFY:
the corpus-wide hash-pin reset (hash_pin, prelude_module_hash_pin,
mono_hash_stability, eq_ord_e2e, embed_export_hash_stable, the
ct4/iter*/loop_recur schema-extension pins) and the list ir_snapshot
golden were regenerated once, deliberately, as the intended one-time
consequence of removing the mode elision from the canonical form —
not a regression. Each regenerated hash verified deterministic across
two runs.

Also fixes a pre-existing latent failure surfaced by the verification
gate, unrelated to this cutover: the `every_contract_names_a_resolvable_
ratifying_test` resolver (design_index_pin) could not resolve the
" + " dual-link ratifying-test form (`uniqueness.rs + linearity.rs`)
that the #57 audit-close (dfdc65f) introduced — it shipped red on that
commit. Resolver taught the dual-link form, mirroring its sibling.

Verification: cargo test --workspace = 731 passed, 0 failed (twice,
stable); e2e 102 passed, no binary exits non-zero (corpus crash-free);
grep-clean for Implicit/fn_implicit/mode_eq across crates; every drop
fix confirmed via emitted IR + AILANG_RC_STATS balance on the head==K,
head!=K, and Nil paths. Three BLOCKEDs en route (the unsound first
husk-dec attempt, the over-strict derivation premise, the leg-B fix
direction) were each treated as a real design/spec gap and rediagnosed,
not patched over.

Supersedes #54 (return-position-only leak patch). Precondition #57
(linearity hardening) was already met. Spec docs/specs/0062, plan
docs/plans/0121.

closes #55
2026-06-02 00:03:46 +02:00

755 lines
27 KiB
Rust

//! Drift detection between ast.rs and `design/contracts/0002-data-model.md`.
//!
//! `design/contracts/0002-data-model.md` is the canonical schema
//! source-of-truth. Every AST enum (`Term`, `Pattern`, `Type`, `Def`,
//! `Literal`, `ParamMode`) must have a JSON-schema anchor (e.g.
//! `"t": "lit"`, `"k": "fn"`) present in that document. These tests
//! enforce the property.
//!
//! The exhaustive `match` per enum is the load-bearing mechanism: adding a
//! new variant without a matching arm fails compilation before the test runs.
//! Once the variant is matched, the test asserts the anchor is present in
//! `design/contracts/0002-data-model.md`. The whole file is the data-model
//! contract, so the file boundary now bounds the section — the former
//! `## Data model` … `## Pipeline` slicer is no longer needed (the split
//! gave the section its own file).
use std::collections::BTreeMap;
use ailang_core::ast::{
ClassDef, ClassMethod, Constraint, ConstDef, Ctor, Def, FnDef, InstanceDef,
InstanceMethod, Literal, Module, NewArg, Pattern, ParamMode, Suppress, Term, Type, TypeDef,
};
const DATA_MODEL: &str = include_str!("../../../design/contracts/0002-data-model.md");
/// Scoped substring match: returns `true` iff `anchor` appears inside a
/// fenced code block (``` or ~~~) of `md`. Info-string `jsonc` / `json` /
/// unspecified all count — fence-toggling is by fence-marker line alone,
/// mirroring the inverse `strip_fences` pattern in
/// `crates/ailang-core/tests/design_index_pin.rs`. The whole-file
/// `.contains()` was fidelity-widened: anchors mentioned in surrounding
/// prose (footnotes, inline references, historical notes) counted as
/// present, so a future edit could delete the canonical schema entry for
/// a variant from inside a fenced block and the drift test would still
/// pass. This helper closes that surface.
fn anchor_in_jsonc_block(md: &str, anchor: &str) -> bool {
let mut in_fence = false;
for line in md.lines() {
let t = line.trim_start();
if t.starts_with("```") || t.starts_with("~~~") {
in_fence = !in_fence;
continue;
}
if in_fence && line.contains(anchor) {
return true;
}
}
false
}
/// Every `Term` variant must have its JSON-schema anchor present in
/// design/contracts/0002-data-model.md. An LLM author cannot produce a term variant
/// whose `"t"` tag is absent from the canonical schema document.
#[test]
fn design_md_anchors_every_term_variant() {
let exemplars: Vec<(&str, Term)> = vec![
(
r#""t": "lit""#,
Term::Lit { lit: Literal::Unit },
),
(
r#""t": "var""#,
Term::Var { name: "x".into() },
),
(
r#""t": "app""#,
Term::App {
callee: Box::new(Term::Var { name: "f".into() }),
args: vec![],
tail: false,
},
),
(
r#""t": "let""#,
Term::Let {
name: "x".into(),
value: Box::new(Term::Lit { lit: Literal::Int { value: 0 } }),
body: Box::new(Term::Var { name: "x".into() }),
},
),
(
r#""t": "letrec""#,
Term::LetRec {
name: "f".into(),
ty: Type::fn_owned(vec![], Type::int(), vec![]),
params: vec![],
body: Box::new(Term::Lit { lit: Literal::Int { value: 0 } }),
in_term: Box::new(Term::Var { name: "f".into() }),
},
),
(
r#""t": "if""#,
Term::If {
cond: Box::new(Term::Lit { lit: Literal::Bool { value: true } }),
then: Box::new(Term::Lit { lit: Literal::Int { value: 1 } }),
else_: Box::new(Term::Lit { lit: Literal::Int { value: 0 } }),
},
),
(
r#""t": "do""#,
Term::Do {
op: "io/print_str".into(),
args: vec![],
tail: false,
},
),
(
r#""t": "ctor""#,
Term::Ctor {
type_name: "List".into(),
ctor: "Nil".into(),
args: vec![],
},
),
(
r#""t": "match""#,
Term::Match {
scrutinee: Box::new(Term::Var { name: "x".into() }),
arms: vec![],
},
),
(
r#""t": "lam""#,
Term::Lam {
params: vec![],
param_tys: vec![],
ret_ty: Box::new(Type::int()),
effects: vec![],
body: Box::new(Term::Lit { lit: Literal::Int { value: 0 } }),
},
),
(
r#""t": "seq""#,
Term::Seq {
lhs: Box::new(Term::Var { name: "a".into() }),
rhs: Box::new(Term::Var { name: "b".into() }),
},
),
(
r#""t": "clone""#,
Term::Clone {
value: Box::new(Term::Var { name: "x".into() }),
},
),
(
r#""t": "reuse-as""#,
Term::ReuseAs {
source: Box::new(Term::Var { name: "x".into() }),
body: Box::new(Term::Var { name: "y".into() }),
},
),
(
r#""t": "loop""#,
Term::Loop {
binders: Vec::new(),
body: Box::new(Term::Lit { lit: Literal::Unit }),
},
),
(
r#""t": "recur""#,
Term::Recur { args: vec![] },
),
(
r#""t": "new""#,
Term::New {
type_name: "T".into(),
args: vec![],
},
),
(r#""t": "intrinsic""#, Term::Intrinsic),
];
for (anchor, term) in exemplars {
// Exhaustive match: compiler rejects this file if a new Term
// variant lacks an arm, catching drift at compile time.
let _: &'static str = match term {
Term::Lit { .. } => "lit",
Term::Var { .. } => "var",
Term::App { .. } => "app",
Term::Let { .. } => "let",
Term::LetRec { .. } => "letrec",
Term::If { .. } => "if",
Term::Do { .. } => "do",
Term::Ctor { .. } => "ctor",
Term::Match { .. } => "match",
Term::Lam { .. } => "lam",
Term::Seq { .. } => "seq",
Term::Clone { .. } => "clone",
Term::ReuseAs { .. } => "reuse-as",
Term::Loop { .. } => "loop",
Term::Recur { .. } => "recur",
Term::New { .. } => "new",
Term::Intrinsic => "intrinsic",
};
assert!(
anchor_in_jsonc_block(DATA_MODEL, anchor),
"design/contracts/0002-data-model.md is missing anchor `{anchor}` for a Term variant — \
add it to design/contracts/0002-data-model.md"
);
}
// The two Term::Lam compound-key tags are kebab-case (closes #30).
// Pin the spellings inside data-model.md's fenced jsonc blocks so a
// future edit cannot silently revert the contract document to
// camelCase while ast.rs ships kebab.
for anchor in [r#""param-types""#, r#""ret-type""#] {
assert!(
anchor_in_jsonc_block(DATA_MODEL, anchor),
"design/contracts/0002-data-model.md is missing Term::Lam kebab-key anchor `{anchor}` — \
check the `lam` fenced JSON block"
);
}
}
/// Every `Pattern` variant must have its JSON-schema anchor present in
/// design/contracts/0002-data-model.md. Missing anchors mean an LLM cannot produce
/// the corresponding pattern form.
#[test]
fn design_md_anchors_every_pattern_variant() {
let exemplars: Vec<(&str, Pattern)> = vec![
(r#""p": "wild""#, Pattern::Wild),
(r#""p": "var""#, Pattern::Var { name: "x".into() }),
(r#""p": "lit""#, Pattern::Lit { lit: Literal::Int { value: 0 } }),
(
r#""p": "ctor""#,
Pattern::Ctor { ctor: "Nil".into(), fields: vec![] },
),
];
for (anchor, pat) in exemplars {
let _: &'static str = match pat {
Pattern::Wild => "wild",
Pattern::Var { .. } => "var",
Pattern::Lit { .. } => "lit",
Pattern::Ctor { .. } => "ctor",
};
assert!(
anchor_in_jsonc_block(DATA_MODEL, anchor),
"design/contracts/0002-data-model.md is missing anchor `{anchor}` for a Pattern variant"
);
}
}
/// Every `Type` variant must have its JSON-schema anchor present in
/// design/contracts/0002-data-model.md. The `"k"` discriminator is load-bearing for
/// the codegen and typechecker; an LLM must know all four forms.
#[test]
fn design_md_anchors_every_type_variant() {
let exemplars: Vec<(&str, Type)> = vec![
(r#""k": "con""#, Type::int()),
(r#""k": "fn""#, Type::fn_owned(vec![], Type::unit(), vec![])),
(r#""k": "var""#, Type::Var { name: "a".into() }),
(
r#""k": "forall""#,
Type::Forall {
vars: vec!["a".into()],
constraints: vec![],
body: Box::new(Type::Var { name: "a".into() }),
},
),
];
for (anchor, ty) in exemplars {
let _: &'static str = match ty {
Type::Con { .. } => "con",
Type::Fn { .. } => "fn",
Type::Var { .. } => "var",
Type::Forall { .. } => "forall",
};
assert!(
anchor_in_jsonc_block(DATA_MODEL, anchor),
"design/contracts/0002-data-model.md is missing anchor `{anchor}` for a Type variant"
);
}
}
/// Every `Literal` variant must have its JSON-schema anchor present in
/// design/contracts/0002-data-model.md. The `"kind"` discriminator identifies the
/// literal type; an LLM cannot produce a literal it hasn't seen in the
/// schema.
#[test]
fn design_md_anchors_every_literal_variant() {
let exemplars: Vec<(&str, Literal)> = vec![
(r#""kind": "int""#, Literal::Int { value: 0 }),
(r#""kind": "bool""#, Literal::Bool { value: true }),
(r#""kind": "str""#, Literal::Str { value: "x".into() }),
(r#""kind": "unit""#, Literal::Unit),
(r#""kind": "float""#, Literal::Float { bits: 0 }),
];
for (anchor, lit) in exemplars {
let _: &'static str = match lit {
Literal::Int { .. } => "int",
Literal::Bool { .. } => "bool",
Literal::Str { .. } => "str",
Literal::Unit => "unit",
Literal::Float { .. } => "float",
};
assert!(
anchor_in_jsonc_block(DATA_MODEL, anchor),
"design/contracts/0002-data-model.md is missing anchor `{anchor}` for a Literal variant"
);
}
}
/// Every `Def` kind must have its JSON-schema anchor present in
/// design/contracts/0002-data-model.md. All five kinds (`fn`, `const`, `type`,
/// `class`, `instance`) must be documented so an LLM can write
/// any kind of top-level definition.
#[test]
fn design_md_anchors_every_def_kind() {
let fn_def = FnDef {
name: "f".into(),
doc: None,
suppress: vec![],
ty: Type::fn_owned(vec![], Type::int(), vec![]),
params: vec![],
body: Term::Lit { lit: Literal::Int { value: 0 } },
export: None,
};
let const_def = ConstDef {
name: "k".into(),
doc: None,
ty: Type::int(),
value: Term::Lit { lit: Literal::Int { value: 0 } },
};
let type_def = TypeDef {
name: "T".into(),
doc: None,
vars: vec![],
ctors: vec![Ctor { name: "C".into(), fields: vec![] }],
drop_iterative: false,
param_in: BTreeMap::new(),
};
let class_def = ClassDef {
name: "Show".into(),
param: "a".into(),
superclass: None,
methods: vec![ClassMethod {
name: "show".into(),
ty: Type::fn_owned(vec![Type::Var { name: "a".into() }], Type::str_(), vec![]),
default: None,
}],
doc: None,
};
let instance_def = InstanceDef {
class: "Show".into(),
type_: Type::int(),
methods: vec![InstanceMethod {
name: "show".into(),
body: Term::Lit { lit: Literal::Str { value: "0".into() } },
}],
doc: None,
};
let exemplars: Vec<(&str, Def)> = vec![
(r#""kind": "fn""#, Def::Fn(fn_def)),
(r#""kind": "const""#, Def::Const(const_def)),
(r#""kind": "type""#, Def::Type(type_def)),
(r#""kind": "class""#, Def::Class(class_def)),
(r#""kind": "instance""#, Def::Instance(instance_def)),
];
for (anchor, def) in exemplars {
let _: &'static str = match def {
Def::Fn(_) => "fn",
Def::Const(_) => "const",
Def::Type(_) => "type",
Def::Class(_) => "class",
Def::Instance(_) => "instance",
};
assert!(
anchor_in_jsonc_block(DATA_MODEL, anchor),
"design/contracts/0002-data-model.md is missing anchor `{anchor}` for a Def kind"
);
}
}
/// Every `ParamMode` variant must have its serialized string form present
/// in design/contracts/0002-data-model.md. The mode annotations are load-bearing for
/// ownership checking; an LLM author must know both forms.
#[test]
fn design_md_anchors_every_parammode_variant() {
let exemplars: Vec<(&str, ParamMode)> = vec![
(r#""own""#, ParamMode::Own),
(r#""borrow""#, ParamMode::Borrow),
];
for (anchor, mode) in exemplars {
let _: &'static str = match mode {
ParamMode::Own => "own",
ParamMode::Borrow => "borrow",
};
assert!(
anchor_in_jsonc_block(DATA_MODEL, anchor),
"design/contracts/0002-data-model.md is missing anchor `{anchor}` for a ParamMode variant"
);
}
}
/// The anchor-presence check must scope its substring match to ```jsonc```
/// (and ``` / ```json) fenced code blocks. The whole-file `.contains()` is
/// fidelity-widened: a future edit can delete the canonical schema entry for
/// a variant from inside a fenced block while leaving the anchor string
/// mentioned in surrounding prose (e.g. a "design rationale" footnote, an
/// inline reference like `the "t": "lit" form`, or a historical-note
/// section). The drift test would silently pass, and downstream LLM authors
/// would lose the canonical schema for a variant the AST still produces.
///
/// This test pins the property in two directions:
/// 1. A markdown string with the anchor present ONLY in prose (no fenced
/// block) must report ABSENT under the scoped helper. The unscoped
/// `.contains()` reports PRESENT for the same input — that mismatch is
/// the bug surface.
/// 2. A markdown string with the anchor present ONLY inside a ```jsonc```
/// block must report PRESENT under the scoped helper.
///
/// Gitea issue #10. The helper `anchor_in_jsonc_block` does not yet exist —
/// this test is RED until the `implement` skill (mini-mode dispatch) adds it
/// and re-routes the six existing call sites onto it.
#[test]
fn anchor_presence_check_is_scoped_to_jsonc_blocks() {
// Anchor mentioned only in prose — the false-pass surface. The whole
// sentence after the heading is plain markdown body text; no fenced
// block exists in this fixture at all.
let prose_only = r#"# Data model
The `"t": "lit"` form used to be the canonical literal anchor. See
historical note below.
## Historical note
Older drafts pinned the literal schema as `"t": "lit"`; the current
schema uses a different shape (see commit log).
"#;
// Anchor present only inside a fenced jsonc block — the true-positive
// surface. Prose outside the block does not mention the anchor.
let jsonc_only = r#"# Data model
The literal form is canonical.
```jsonc
{ "t": "lit", "lit": Literal }
```
End.
"#;
// Unscoped substring match — what the file does today. Both inputs
// report PRESENT, and that is the bug: prose-only must not count.
assert!(prose_only.contains(r#""t": "lit""#));
assert!(jsonc_only.contains(r#""t": "lit""#));
// Scoped helper — what the file must do. Prose-only is ABSENT,
// jsonc-only is PRESENT. This call is the RED: the helper does not
// yet exist, so this test fails to compile until `implement` mini-mode
// factors it out and re-routes the six call sites onto it.
assert!(
!anchor_in_jsonc_block(prose_only, r#""t": "lit""#),
"scoped helper must NOT count an anchor that lives only in prose"
);
assert!(
anchor_in_jsonc_block(jsonc_only, r#""t": "lit""#),
"scoped helper MUST count an anchor that lives inside a ```jsonc``` block"
);
// Live document: the helper must also report PRESENT for the live
// data-model.md (proves the helper does not over-narrow and break
// the existing six checks).
assert!(
anchor_in_jsonc_block(DATA_MODEL, r#""t": "lit""#),
"scoped helper must find live anchors inside data-model.md ```jsonc``` blocks"
);
}
/// Schema-shape pin: the JSON serialisation of `Term::Lam` must
/// emit kebab-case tags `"param-types"` and `"ret-type"`, and
/// must NOT emit the old camelCase `"paramTypes"` / `"retType"`.
/// Pinned because the closes-#30 milestone made the rename a
/// schema-stability invariant: a future regression that flips
/// the `#[serde(rename)]` string back to camelCase must fire RED
/// here, not slip through.
#[test]
fn lam_serialises_with_kebab_keys() {
let lam = Term::Lam {
params: vec!["x".into()],
param_tys: vec![Type::int()],
ret_ty: Box::new(Type::int()),
effects: vec![],
body: Box::new(Term::Var { name: "x".into() }),
};
let v = serde_json::to_value(&lam).expect("Term::Lam serialises");
let obj = v.as_object().expect("Term::Lam serialises as object");
assert!(
obj.contains_key("param-types"),
"Term::Lam must emit `param-types` key; got keys {:?}",
obj.keys().collect::<Vec<_>>(),
);
assert!(
obj.contains_key("ret-type"),
"Term::Lam must emit `ret-type` key; got keys {:?}",
obj.keys().collect::<Vec<_>>(),
);
assert!(
!obj.contains_key("paramTypes"),
"Term::Lam must NOT emit old camelCase `paramTypes` key",
);
assert!(
!obj.contains_key("retType"),
"Term::Lam must NOT emit old camelCase `retType` key",
);
}
/// Nested struct key anchors must be present in design/contracts/0002-data-model.md.
/// These keys appear inside `Suppress`, `ClassMethod`, `InstanceMethod`,
/// and `Type::Forall` — they are not discriminators but they ARE part
/// of the canonical JSON schema and must remain documented.
#[test]
fn design_md_anchors_nested_struct_keys() {
// Constructors exercised here ensure ast.rs field names are correct.
let _ = Suppress { code: "x".into(), because: "y".into() };
let _ = ClassMethod {
name: "m".into(),
ty: Type::fn_owned(vec![], Type::int(), vec![]),
default: None,
};
let _ = InstanceMethod {
name: "m".into(),
body: Term::Lit { lit: Literal::Unit },
};
let _ = Constraint {
class: "Show".into(),
type_: Type::int(),
};
let anchors = [
r#""code""#,
r#""because""#,
r#""methods""#,
r#""constraints""#,
];
for anchor in anchors {
assert!(
anchor_in_jsonc_block(DATA_MODEL, anchor),
"design/contracts/0002-data-model.md is missing nested-struct-key anchor `{anchor}`"
);
}
}
/// prep.2 (kernel-extension-mechanics): pin the JSON byte-shape of
/// `Term::New` with a single `NewArg::Value`. Property protected:
/// the tag is `t: "new"`, the type-name field is serialised as
/// `"type"` (not `"type_name"` or any camelCase variant), the args
/// list serialises each `NewArg::Value` as `{kind: "value", value:
/// <Term JSON>}`, and the whole shape round-trips through
/// `serde_json` without information loss. A future edit that
/// re-renames any of these keys breaks the pin and the data-model
/// document must be updated in lockstep.
#[test]
fn term_new_round_trips() {
let t = Term::New {
type_name: "Counter".into(),
args: vec![NewArg::Value(Term::Lit {
lit: Literal::Int { value: 42 },
})],
};
let json = serde_json::to_value(&t).expect("serialise Term::New");
assert_eq!(json["t"], "new", "Term discriminator must be `t: \"new\"`");
assert_eq!(json["type"], "Counter", "type_name must serialise as `\"type\"`");
let args = json["args"].as_array().expect("args must be an array");
assert_eq!(args.len(), 1);
assert_eq!(
args[0]["kind"], "value",
"NewArg::Value discriminator must be `kind: \"value\"`"
);
assert!(
args[0]["value"].is_object(),
"value-arg's payload is the inlined Term JSON object"
);
let recovered: Term =
serde_json::from_value(json).expect("Term::New round-trips through serde");
match recovered {
Term::New { type_name, args } => {
assert_eq!(type_name, "Counter");
assert_eq!(args.len(), 1);
assert!(matches!(args[0], NewArg::Value(_)));
}
other => panic!("expected Term::New after round-trip, got {other:?}"),
}
}
/// prep.2 (kernel-extension-mechanics): pin the JSON byte-shape of
/// `NewArg::Type` (a Type-positional arg). Property protected: the
/// inner `value` carries a full `Type` JSON object (here a
/// `(k = "con", name = "Float")`), and the round-trip preserves both
/// the discriminator and the embedded Type's own discriminator. A
/// future edit that changes `NewArg`'s serde tag/content shape (or
/// renames `kind`/`value`) breaks the pin in lockstep with the
/// data-model document.
#[test]
fn term_new_type_arg_round_trips() {
let t = Term::New {
type_name: "Series".into(),
args: vec![
NewArg::Type(Type::Con {
name: "Float".into(),
args: vec![],
}),
NewArg::Value(Term::Lit {
lit: Literal::Int { value: 3 },
}),
],
};
let json = serde_json::to_value(&t).expect("serialise Term::New with mixed args");
let args = json["args"].as_array().expect("args must be an array");
assert_eq!(args.len(), 2);
assert_eq!(args[0]["kind"], "type", "first arg is a Type-positional");
assert_eq!(args[0]["value"]["k"], "con", "Type-arg's value carries the Type tag");
assert_eq!(args[0]["value"]["name"], "Float");
assert_eq!(args[1]["kind"], "value", "second arg is a Value-positional");
let recovered: Term =
serde_json::from_value(json).expect("mixed Term::New round-trips through serde");
if let Term::New { args, .. } = recovered {
assert!(matches!(args[0], NewArg::Type(Type::Con { ref name, .. }) if name == "Float"));
assert!(matches!(args[1], NewArg::Value(_)));
} else {
panic!("expected Term::New after round-trip");
}
}
/// prep.3 (kernel-extension-mechanics): pin the JSON byte-shape of
/// `Module.kernel`. Property protected: when `true`, the key
/// serialises as the bare `"kernel": true` flag adjacent to
/// `"name"`; when `false`, it is omitted entirely from canonical
/// JSON (preserving the bit-identical hash of every pre-existing
/// fixture). A future edit that renames the field, drops the
/// `skip_serializing_if`, or changes its boolean shape breaks the
/// pin in lockstep with the data-model document.
#[test]
fn module_kernel_flag_round_trips() {
let m_on = Module {
schema: ailang_core::SCHEMA.to_string(),
name: "k".into(),
kernel: true,
imports: vec![],
defs: vec![],
};
let json_on = serde_json::to_value(&m_on).expect("serialise kernel: true");
assert_eq!(
json_on["kernel"], true,
"kernel: true must appear as a bare boolean key"
);
let m_off = Module {
schema: ailang_core::SCHEMA.to_string(),
name: "k".into(),
kernel: false,
imports: vec![],
defs: vec![],
};
let json_off = serde_json::to_value(&m_off).expect("serialise kernel: false");
assert!(
json_off.get("kernel").is_none(),
"kernel: false must be omitted from canonical JSON (hash-stability)"
);
let recovered: Module =
serde_json::from_value(json_on).expect("Module with kernel: true round-trips");
assert!(recovered.kernel);
}
/// prep.3 (kernel-extension-mechanics): pin the JSON byte-shape of
/// `TypeDef.param_in`. Property protected: when non-empty, the map
/// serialises under the kebab-case key `"param-in"`; when empty,
/// it is omitted entirely (preserving the bit-identical hash of
/// every fixture that does not restrict). The BTreeMap/BTreeSet
/// choice gives deterministic iteration order (alphabetical) so
/// the canonical-JSON bytes are reproducible. A future edit that
/// renames the field, drops the `skip_serializing_if`, or changes
/// the inner collection type breaks the pin in lockstep with the
/// data-model document.
#[test]
fn typedef_param_in_round_trips() {
use std::collections::BTreeSet;
let mut restrict: BTreeMap<String, BTreeSet<String>> = BTreeMap::new();
restrict.insert(
"a".into(),
["Int".to_string(), "Float".to_string()].into_iter().collect(),
);
let td_on = TypeDef {
name: "T".into(),
vars: vec!["a".into()],
ctors: vec![],
doc: None,
drop_iterative: false,
param_in: restrict.clone(),
};
let json_on = serde_json::to_value(&td_on).expect("serialise param_in non-empty");
let pi = json_on.get("param-in").expect(
"non-empty param_in serialises under kebab key `param-in`",
);
let allowed = pi["a"].as_array().expect("inner value is a Type-name array");
assert_eq!(allowed.len(), 2);
// BTreeSet iteration order is alphabetical:
assert_eq!(allowed[0], "Float");
assert_eq!(allowed[1], "Int");
let td_off = TypeDef {
name: "T".into(),
vars: vec![],
ctors: vec![],
doc: None,
drop_iterative: false,
param_in: BTreeMap::new(),
};
let json_off = serde_json::to_value(&td_off).expect("serialise param_in empty");
assert!(
json_off.get("param-in").is_none(),
"empty param_in must be omitted from canonical JSON (hash-stability)"
);
let recovered: TypeDef =
serde_json::from_value(json_on).expect("TypeDef with param_in round-trips");
assert_eq!(recovered.param_in, restrict);
}
/// raw-buf.4: pin the JSON byte-shape of the raw_buf kernel-tier
/// base-extension module. Exercises all three additive schema items
/// in a single fixture (`Module.kernel = true`, one TypeDef with
/// `param-in`, one ctor).
#[test]
fn raw_buf_module_round_trips() {
let m = ailang_surface::parse_raw_buf();
assert!(m.kernel, "raw_buf is kernel-tier");
assert_eq!(m.name, "raw_buf");
let json = serde_json::to_value(&m).expect("serialise raw_buf");
let recovered: Module =
serde_json::from_value(json).expect("raw_buf round-trips through serde");
assert!(recovered.kernel);
assert_eq!(recovered.name, "raw_buf");
let td = recovered.defs.iter().find_map(|d| match d {
Def::Type(t) if t.name == "RawBuf" => Some(t),
_ => None,
}).expect("RawBuf TypeDef present");
let allowed = td.param_in.get("a").expect("RawBuf.a restricted");
assert!(allowed.contains("Int") && allowed.contains("Float") && allowed.contains("Bool"));
}