19%

Section 7A: Core Built-ins

Goal: Type conversions, assertions, I/O, and core built-in functions

SPEC: spec/annex-c-built-in-functions.md PROPOSALS:

  • proposals/approved/as-conversion-proposal.md — Type conversion syntax
  • proposals/approved/developer-functions-proposal.md — Developer convenience functions
  • proposals/approved/embed-expression-proposal.md — Compile-time file embedding
  • proposals/approved/char-byte-classification-proposal.md — Char/byte classification methods
  • proposals/approved/byte-string-access-proposal.md — Byte-level string access

7A.1 Type Conversions

PROPOSAL: proposals/approved/as-conversion-proposal.md

Type conversions use as/as? syntax instead of int(), float(), etc. This removes the special-case exception for positional arguments.

Partial implementation note (verified 2026-03-29): as/as? syntax IS parsed (ExprKind::Cast), type-checked (infer_cast), evaluated (eval_can_cast), and lowered through ARC IR (lower_cast). The evaluator handles int-to-float, int-to-byte, int-to-char, char-to-int, byte-to-int, str-to-int (as?), str-to-float (as?) via hardcoded dispatch — NOT trait-based as the As/TryAs proposal envisions. BUG-7A-05: tests/spec/expressions/type_conversion.ori is ENTIRELY COMMENTED OUT — zero active tests.

  • Implement: As<T> trait — infallible conversions

    • Rust Tests: ori_types/src/check/traits/as_trait.rs — As trait tests
    • Ori Tests: tests/spec/stdlib/conversions.ori
    • LLVM Support: LLVM codegen for As trait
    • LLVM Rust Tests: ori_llvm/tests/conversion_tests.rs — As trait codegen
    • AOT Tests: No AOT coverage yet

  • Implement: TryAs<T> trait — fallible conversions returning Option<T>

    • Rust Tests: ori_types/src/check/traits/try_as_trait.rs — TryAs trait tests
    • Ori Tests: tests/spec/stdlib/conversions.ori
    • LLVM Support: LLVM codegen for TryAs trait
    • LLVM Rust Tests: ori_llvm/tests/conversion_tests.rs — TryAs trait codegen
    • AOT Tests: No AOT coverage yet

  • Implement: x as T syntax — desugars to As<T>.as(self: x)

    • Rust Tests: ori_eval/src/interpreter/as_conversion.rs — as syntax tests
    • Ori Tests: tests/spec/expressions/as_conversion.ori
    • LLVM Support: LLVM codegen for as syntax
    • LLVM Rust Tests: ori_llvm/tests/conversion_tests.rs — as syntax codegen
    • AOT Tests: No AOT coverage yet

  • Implement: x as? T syntax — desugars to TryAs<T>.try_as(self: x)

    • Rust Tests: ori_eval/src/interpreter/as_conversion.rs — as? syntax tests
    • Ori Tests: tests/spec/expressions/as_conversion.ori
    • LLVM Support: LLVM codegen for as? syntax
    • LLVM Rust Tests: ori_llvm/tests/conversion_tests.rs — as? syntax codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Standard As implementations

    • impl int: As<float> — widening (infallible)
    • impl int: As<str> — formatting (infallible)
    • impl float: As<str> — formatting (infallible)
    • impl bool: As<str> — “true”/“false” (infallible)
    • impl char: As<int> — codepoint (infallible)
    • Ori Tests: tests/spec/stdlib/as_impls.ori
    • LLVM Support: LLVM codegen for standard As implementations
    • LLVM Rust Tests: ori_llvm/tests/conversion_tests.rs — As implementations codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Standard TryAs implementations

    • impl str: TryAs<int> — parsing (fallible)
    • impl str: TryAs<float> — parsing (fallible)
    • impl int: TryAs<byte> — range check (fallible)
    • impl int: TryAs<char> — valid codepoint check (fallible)
    • Ori Tests: tests/spec/stdlib/try_as_impls.ori
    • LLVM Support: LLVM codegen for standard TryAs implementations
    • LLVM Rust Tests: ori_llvm/tests/conversion_tests.rs — TryAs implementations codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Compile-time enforcement — as only for infallible conversions

    • Rust Tests: ori_types/src/check/as_conversion.rs — enforcement tests
    • Ori Tests: tests/compile-fail/as_fallible.ori
    • LLVM Support: LLVM codegen for as conversion enforcement
    • LLVM Rust Tests: ori_llvm/tests/conversion_tests.rs — as enforcement codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Float truncation methods (not as)

    • float.truncate() -> int — toward zero
    • float.round() -> int — nearest
    • float.floor() -> int — toward negative infinity
    • float.ceil() -> int — toward positive infinity
    • Ori Tests: tests/spec/stdlib/float_methods.ori
    • LLVM Support: LLVM codegen for float truncation methods
    • LLVM Rust Tests: ori_llvm/tests/conversion_tests.rs — float truncation codegen
    • AOT Tests: No AOT coverage yet

  • Remove: int(), float(), str(), byte() function syntax

    • These are replaced by as/as? syntax
    • No migration period needed if implementing fresh
    • LLVM Support: LLVM codegen removal of legacy conversion functions
    • LLVM Rust Tests: ori_llvm/tests/conversion_tests.rs — verify legacy functions removed
    • AOT Tests: No AOT coverage yet

  • BUG-7A-05: Un-comment and fix tests/spec/expressions/type_conversion.ori — restore test coverage for hardcoded cast dispatch

    • Verify all currently-working conversions (int-to-float, int-to-byte, int-to-char, char-to-int, byte-to-int, str-to-int/float via as?)
    • Fix or #skip any tests that depend on unimplemented features (str-to-bool)

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.1) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.1 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.1: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.2 Assertions

  • Implement: assert(cond:) [done] (2026-02-10) (verified 2026-03-29)

    • Ori Tests: 310+ usages across 20+ spec test files, all types exercised
    • LLVM Support: LLVM codegen for assert — works transitively (desugars to if !cond then panic(...) and panic has LLVM support)
    • LLVM Rust Tests: ori_llvm/tests/assertion_tests.rs (file does not exist)
    • AOT Tests: No AOT coverage yet

  • Implement: assert_eq(actual:, expected:) [done] (2026-02-10) (verified 2026-03-29)

    • Ori Tests: 78+ usages across int, str, bool, float, list, map, tuple, Option, Result, struct
    • BUG-7A-02: Spec says assert_eq<T: Eq + Debug> but implementation is assert_eq<T: Eq> — uses str(actual) instead of actual.debug(), poor messages for non-Printable types
    • WEAK TESTS: No negative test for non-Eq types
    • LLVM Support: LLVM codegen for assert_eq — works transitively through panic
    • LLVM Rust Tests: ori_llvm/tests/assertion_tests.rs (file does not exist)
    • AOT Tests: No AOT coverage yet

  • Implement: assert_ne(actual:, expected:) [done] (2026-02-10) (verified 2026-03-29)

    • Ori Tests: WEAK TESTS — only 3 usages, only int and str tested
    • WEAK TESTS: Missing matrix coverage for bool, float, list, map, struct, Option
    • LLVM Support: LLVM codegen for assert_ne
    • LLVM Rust Tests: ori_llvm/tests/assertion_tests.rs (file does not exist)
    • AOT Tests: No AOT coverage yet

  • Implement: assert_some(x) — spec/annex-c-built-in-functions.md § assert_some [done] (verified 2026-03-29)

    • Implementation: library/std/testing.ori — implemented in Ori stdlib
    • BUG-7A-04: Spec says assert_some<T>(option: Option<T>) -> T (returns inner value) but implementation returns void
    • NEEDS TESTS: Zero spec tests use assert_some
    • LLVM Support: LLVM codegen for assert_some

  • Implement: assert_none(x) — spec/annex-c-built-in-functions.md § assert_none [done] (verified 2026-03-29)

    • Implementation: library/std/testing.ori — implemented in Ori stdlib
    • NEEDS TESTS: Zero spec tests use assert_none
    • LLVM Support: LLVM codegen for assert_none

  • Implement: assert_ok(x) — spec/annex-c-built-in-functions.md § assert_ok [done] (verified 2026-03-29)

    • Implementation: library/std/testing.ori — implemented in Ori stdlib
    • BUG-7A-04: Spec says assert_ok<T, E>(result: Result<T, E>) -> T but implementation returns void
    • NEEDS TESTS: Zero spec tests use assert_ok
    • LLVM Support: LLVM codegen for assert_ok

  • Implement: assert_err(x) — spec/annex-c-built-in-functions.md § assert_err [done] (verified 2026-03-29)

    • Implementation: library/std/testing.ori — implemented in Ori stdlib
    • BUG-7A-04: Spec says assert_err<T, E>(result: Result<T, E>) -> E but implementation returns void
    • NEEDS TESTS: Zero spec tests use assert_err
    • LLVM Support: LLVM codegen for assert_err

  • MISSING ITEM: assert_panics(expr:) — implemented in library/std/testing.ori, used in 4 test files, listed in spec annex-c but not tracked in roadmap (verified 2026-03-29)

    • Ori Tests: Used in integer_safety.ori, operators_bitwise.ori, ranges.ori, range_edge_cases.ori
    • LLVM Support: LLVM codegen for assert_panics

  • MISSING ITEM: assert_panics_with(expr:, message:) — implemented in library/std/testing.ori, listed in spec annex-c but not tracked in roadmap (verified 2026-03-29)

    • Ori Tests: Needs dedicated spec tests
    • LLVM Support: LLVM codegen for assert_panics_with

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.2) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.2 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.2: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.3 I/O and Other

  • Implement: print(x) [done] (2026-02-10) (verified 2026-03-29)

    • Ori Tests: Used pervasively in test suite
    • LLVM Support: LLVM codegen for print — _ori_print in runtime, test_aot_hello_world verifies
    • LLVM Rust Tests: ori_llvm/tests/io_tests.rs (file does not exist)
    • AOT Tests: No dedicated AOT coverage yet

  • Implement: compare(a, b) [done] (2026-02-10) (verified 2026-03-29)

    • Ori Tests: 58 tests in comparable.ori for .compare(other:) method
    • BUG-7A-03: Spec says compare<T: Comparable>(left: T, right: T) -> Ordering (generic) but implementation is compare(a: int, b: int) (int-only) — cannot compare(left: "apple", right: "banana")
    • LLVM Support: LLVM codegen for compare — works transitively as Ori function
    • LLVM Rust Tests: ori_llvm/tests/comparison_tests.rs (file does not exist)
    • AOT Tests: No AOT coverage yet

  • Implement: min(a, b), max(a, b) [done] (2026-02-10) (verified 2026-03-29)

    • Ori Tests: Dedicated tests in comparable.ori with int values including equal values, negatives
    • BUG-7A-03: Spec says min<T: Comparable> / max<T: Comparable> (generic) but implementation is int-only
    • LLVM Support: LLVM codegen for min/max — works transitively as Ori functions
    • LLVM Rust Tests: ori_llvm/tests/comparison_tests.rs (file does not exist)
    • AOT Tests: No AOT coverage yet

  • Implement: panic(msg) [done] (2026-02-10) (verified 2026-03-29)

    • Ori Tests: 20+ tests in never.ori, used in #fail test attributes
    • LLVM Support: LLVM codegen for panic — _ori_panic/_ori_panic_cstr in runtime, 9 dedicated AOT tests in panic.rs
    • AOT Tests: 33 AOT panic tests pass (cargo test -p ori_llvm -- panic)
    • LLVM Rust Tests: ori_llvm/tests/panic_tests.rs (file does not exist)

  • MISSING ITEM: hash_combine(seed:, value:) — implemented in ori_eval/src/function_val.rs (function_val_hash_combine), typed in ori_types/src/infer/expr/identifiers.rs, listed in prelude, but not tracked in roadmap (verified 2026-03-29)

  • MISSING ITEM: is_cancelled() — listed in ori-syntax.md prelude as a builtin function, but not tracked in section 7A (verified 2026-03-29)

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.3) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.3 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.3: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.4 Float NaN Behavior

Decision: NaN comparisons panic (no proposal needed — behavioral decision)

Fits Ori’s “bugs should be caught” philosophy (same as integer overflow).

Status (verified 2026-03-29): NOT STARTED. Current behavior follows IEEE 754 (NaN == NaN is false, NaN != NaN is true, no panic). Tests in operators_comparison.ori assert IEEE 754 behavior. When 7A.4 is implemented, those existing tests will need to change.

  • Implement: NaN comparison panics

    • NaN == NaN → PANIC
    • NaN < x → PANIC
    • NaN > x → PANIC
    • Rust Tests: ori_eval/src/interpreter/binary.rs — NaN comparison tests
    • Ori Tests: tests/spec/types/float_nan.ori
    • LLVM Support: LLVM codegen for NaN comparison panic
    • LLVM Rust Tests: ori_llvm/tests/float_tests.rs — NaN comparison panic codegen
    • AOT Tests: No AOT coverage yet

  • Implement: NaN-producing operations don’t panic (only comparisons)

    • 0.0 / 0.0 → NaN (allowed)
    • Using NaN in arithmetic → NaN (allowed)
    • Comparing NaN → PANIC
    • Ori Tests: tests/spec/types/float_nan_ops.ori
    • LLVM Support: LLVM codegen for NaN-producing operations
    • LLVM Rust Tests: ori_llvm/tests/float_tests.rs — NaN operations codegen
    • AOT Tests: No AOT coverage yet

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.4) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.4 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.4: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.5 Developer Functions

PROPOSAL: proposals/approved/developer-functions-proposal.md

todo, unreachable, and dbg for developer convenience. These provide semantic meaning (unfinished vs. impossible code) and inline debugging.

  • Implement: todo() and todo(reason:) — Mark unfinished code [done] (verified 2026-03-29)

    • Returns Never, panics with “not yet implemented at file:line”
    • Location information captured at compile time
    • Rust Tests: FunctionExpKind::Todo in ori_eval/src/interpreter/can_eval/function_exp.rs
    • Ori Tests: tests/spec/types/never.ori lines 76-89 — tests todo() and todo(reason:) as Never-producing expressions with coercion to int and str (semantic pin)
    • LLVM Support: LLVM codegen for todo
    • LLVM Rust Tests: ori_llvm/tests/developer_tests.rs — todo codegen
    • AOT Tests: No AOT coverage yet

  • Implement: unreachable() and unreachable(reason:) — Mark impossible code [done] (verified 2026-03-29)

    • Returns Never, panics with “unreachable code reached at file:line”
    • Semantically distinct from todo (impossible vs. not done)
    • Rust Tests: FunctionExpKind::Unreachable in ori_eval/src/interpreter/can_eval/function_exp.rs
    • Ori Tests: tests/spec/types/never.ori lines 92-105 — tests both forms (semantic pin via coercion tests)
    • LLVM Support: LLVM codegen for unreachable
    • LLVM Rust Tests: ori_llvm/tests/developer_tests.rs — unreachable codegen
    • AOT Tests: No AOT coverage yet

  • Implement: dbg(value:) and dbg(value:, label:) — Debug printing

    • Generic: dbg<T: Debug>(value: T) -> T
    • Writes to stderr via Print capability
    • Output format: [file:line] = value or [file:line] label = value
    • Returns value unchanged for inline use
    • Rust Tests: ori_eval/src/function_val.rs — dbg tests
    • Ori Tests: tests/spec/stdlib/developer_functions.ori
    • LLVM Support: LLVM codegen for dbg
    • LLVM Rust Tests: ori_llvm/tests/developer_tests.rs — dbg codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Compile-time location capture for all three functions

    • Location passed implicitly by compiler, not visible in user signature
    • Rust Tests: ori_types/src/infer/expr/identifiers.rs — location capture tests
    • Ori Tests: Verify location appears in panic messages/dbg output

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.5) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.5 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.5: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.6 Additional Built-in Functions

Proposal: proposals/approved/additional-builtins-proposal.md

Formalizes repeat, compile_error, PanicInfo, and clarifies ?? operator semantics.

repeat Function

  • Implement: repeat<T: Clone>(value: T) -> impl Iterator — infinite iterator of cloned values [done] (verified 2026-03-29)

    • Rust Tests: function_val_repeat in ori_eval/src/function_val.rs, type signature in ori_types/src/infer/expr/identifiers.rs
    • Ori Tests: tests/spec/traits/iterator/infinite.ori — 14 tests covering repeat with int, str, bool, take(0), map, filter, enumerate, skip, count, fold, any, all, zip
    • LLVM Support: LLVM codegen for repeat
    • LLVM Rust Tests: ori_llvm/tests/iterator_tests.rs — repeat codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Clone requirement enforcement — T must implement Clone

    • Rust Tests: ori_types/src/infer/expr/identifiers.rs — repeat type checking
    • Ori Tests: tests/compile-fail/repeat_not_clone.ori

  • Implement: Integration with Iterator trait — .take(), .collect(), etc.

    • Ori Tests: tests/spec/stdlib/repeat_iterator.ori

PanicInfo Type

Proposal: proposals/approved/panic-handler-proposal.md (extends basic definition)

Spec: spec/17-errors-and-panics.md § PanicInfo Type (updated with full structure)

  • Spec: PanicInfo type definition — { message, location, stack_trace, thread_id } DONE

  • Implement: PanicInfo struct type — { message: str, location: TraceEntry, stack_trace: [TraceEntry], thread_id: Option<int> } (verified 2026-03-29: partially implemented in LLVM IR in panic_trampoline.rs, AOT tests define it locally; no interpreter/typeck/prelude registration)

    • Rust Tests: ori_types/src/check/registration/mod.rs — PanicInfo type tests
    • Ori Tests: tests/spec/types/panic_info.ori
    • LLVM Support: LLVM codegen for PanicInfo
    • LLVM Rust Tests: ori_llvm/tests/type_tests.rs — PanicInfo codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Printable impl for PanicInfo

    • Ori Tests: tests/spec/types/panic_info_printable.ori

  • Implement: Debug impl for PanicInfo

    • Ori Tests: tests/spec/types/panic_info_debug.ori

  • Add to prelude: PanicInfo available without import

    • Ori Tests: tests/spec/prelude/panic_info.ori

@panic Handler

Proposal: proposals/approved/panic-handler-proposal.md

App-wide panic handler function that executes before program termination.

  • Implement: Recognize @panic as special function (like @main) [done] (verified 2026-03-29)

    • LLVM Support: Implemented in panic_trampoline.rs and entry_point.rs
    • AOT Tests: 9 dedicated panic tests in compiler/ori_llvm/tests/aot/panic.rs (33 total AOT panic tests pass)
    • Rust Tests: ori_types/src/check/special_functions.rs — @panic recognition
    • Ori Tests: tests/spec/declarations/panic_handler.ori
    • Interpreter Support: Not implemented — GAP: LLVM-only, breaks dual-execution parity

  • Implement: Validate signature (PanicInfo) -> void [done] (verified 2026-03-29)

    • LLVM Support: Signature validation in panic_trampoline.rs
    • Rust Tests: ori_types/src/check/special_fns.rs — @panic signature validation
    • Ori Tests: tests/compile-fail/panic_handler_wrong_sig.ori

  • Implement: Error if multiple @panic definitions

    • Rust Tests: ori_types/src/check/special_functions.rs — multiple @panic error
    • Ori Tests: tests/compile-fail/multiple_panic_handlers.ori

  • Implement: Implicit stderr for print() inside @panic

    • Rust Tests: ori_eval/src/interpreter/panic_handler.rs — stderr redirection
    • Ori Tests: tests/spec/declarations/panic_print_stderr.ori
    • LLVM Support: LLVM codegen for stderr redirection in @panic
    • LLVM Rust Tests: ori_llvm/tests/panic_tests.rs — stderr redirection codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Runtime panic hook installation at program start [done] (verified 2026-03-29)

    • LLVM Support: ori_register_panic_handler in ori_rt
    • AOT Tests: Tested in compiler/ori_llvm/tests/aot/panic.rs
    • Rust Tests: ori_eval/src/interpreter/panic.rs — hook installation
    • Interpreter Support: Not implemented — GAP: LLVM-only, breaks dual-execution parity

  • Implement: Construct PanicInfo (message, location, stack_trace, thread_id) on panic

    • Rust Tests: ori_eval/src/interpreter/panic.rs — PanicInfo construction
    • Ori Tests: tests/spec/runtime/panic_info_construction.ori
    • LLVM Support: LLVM codegen for PanicInfo construction
    • LLVM Rust Tests: ori_llvm/tests/panic_tests.rs — PanicInfo construction codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Re-panic detection — immediate termination if handler panics [done] (verified 2026-03-29)

    • LLVM Support: Implemented in panic_trampoline.rs
    • AOT Tests: test_panic_handler_re_entrancy in compiler/ori_llvm/tests/aot/panic.rs
    • Rust Tests: ori_eval/src/interpreter/panic.rs — re-panic detection
    • Ori Tests: tests/spec/runtime/panic_in_handler.ori
    • Interpreter Support: Not implemented — GAP: LLVM-only, breaks dual-execution parity

  • Implement: First panic wins in concurrent context

    • Rust Tests: ori_eval/src/interpreter/panic.rs — concurrent panic handling
    • Ori Tests: tests/spec/runtime/concurrent_panic.ori
    • LLVM Support: LLVM codegen for concurrent panic handling
    • LLVM Rust Tests: ori_llvm/tests/panic_tests.rs — concurrent panic codegen
    • AOT Tests: No AOT coverage yet

  • Implement: Default handler (when no @panic defined) — print to stderr [done] (verified 2026-03-29)

    • LLVM Support: Implemented — test_panic_default_nonzero_exit in compiler/ori_llvm/tests/aot/panic.rs
    • AOT Tests: Tested in compiler/ori_llvm/tests/aot/panic.rs
    • Rust Tests: ori_eval/src/interpreter/panic.rs — default handler
    • Ori Tests: tests/spec/runtime/default_panic_handler.ori
    • Interpreter Support: Not implemented — GAP: LLVM-only, breaks dual-execution parity

  • Implement: Exit with non-zero code after handler returns [done] (verified 2026-03-29)

    • LLVM Support: Implemented and tested
    • AOT Tests: test_panic_default_nonzero_exit in compiler/ori_llvm/tests/aot/panic.rs
    • Rust Tests: ori_eval/src/interpreter/panic.rs — exit code
    • Interpreter Support: Not implemented — GAP: LLVM-only, breaks dual-execution parity

compile_error Function

  • MISSING ITEM: compile_error(msg:) — listed in spec annex-c and ori-syntax.md prelude but not previously tracked in roadmap (verified 2026-03-29)

    • Implement: Compile-time error emission — stops compilation with user message
    • Rust Tests: ori_types/src/infer/expr/identifiers.rs — compile_error type checking
    • Ori Tests: tests/compile-fail/compile_error.ori
    • LLVM Support: No codegen needed (compile-time only)

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.6) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.6 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.6: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.7 Resource Management

Proposal: proposals/approved/drop-trait-proposal.md

Adds drop_early function for explicit early resource release.

drop_early Function

  • Implement: drop_early<T>(value: T) -> void — Force drop before scope exit

    • Rust Tests: ori_eval/src/function_val.rs — drop_early tests
    • Ori Tests: tests/spec/stdlib/drop_early.ori
    • LLVM Support: LLVM codegen for drop_early
    • LLVM Rust Tests: ori_llvm/tests/drop_tests.rs — drop_early codegen
    • AOT Tests: No AOT coverage yet

  • Implement: drop_early works for any type (not restricted to T: Drop)

    • Types with Drop: drop method called, then memory reclaimed
    • Types without Drop: memory reclaimed immediately
    • Ori Tests: tests/spec/stdlib/drop_early_any_type.ori

  • Add to prelude: drop_early available without import

    • Ori Tests: tests/spec/prelude/drop_early.ori

  • Update Spec: spec/annex-c-built-in-functions.md — add drop_early documentation

    • Signature: drop_early<T>(value: T) -> void
    • Semantics: Takes ownership, value is dropped immediately
    • Use case: Release resources before scope exit

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.7) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.7 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.7: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.8 Compile-Time File Embedding

PROPOSAL: proposals/approved/embed-expression-proposal.md

embed and has_embed built-in expressions for compile-time file embedding. Type-driven: str (UTF-8 validated) or [byte] (raw binary) based on expected type. Paths are const-evaluable expressions, relative to source file, restricted to project root.

  • Implement: embed(path) — Compile-time file embedding

    • Context-sensitive keyword, parsed as EmbedExpr node
    • Type-driven: str → UTF-8 read + validation, [byte] → raw bytes
    • Path must be const-evaluable str (supports interpolation, const functions)
    • Path resolution relative to source file, no absolute paths, no project escape
    • Rust Tests: ori_types/src/infer/expr/embed.rs — type inference for embed
    • Ori Tests: tests/spec/embed/embed_str.ori, tests/spec/embed/embed_bytes.ori
    • LLVM Support: Emit embedded data in .rodata section
    • LLVM Rust Tests: ori_llvm/tests/embed_tests.rs — embed codegen
    • AOT Tests: No AOT coverage yet

  • Implement: has_embed(path) — Compile-time file existence check

    • Returns compile-time bool
    • Same path resolution rules as embed
    • Rust Tests: ori_types/src/infer/expr/embed.rs — has_embed type checking
    • Ori Tests: tests/spec/embed/has_embed.ori
    • LLVM Support: LLVM codegen for has_embed
    • LLVM Rust Tests: ori_llvm/tests/embed_tests.rs — has_embed codegen
    • AOT Tests: No AOT coverage yet

  • Implement: File size limit enforcement (10 MB default)

    • #embed_limit(size:) attribute for per-expression override
    • ori.toml [embed] max_file_size for project-wide override
    • Ori Tests: tests/compile-fail/embed_size_limit.ori

  • Implement: File dependency tracking in Salsa

    • Hash-based invalidation (content hash, not mtime)
    • Embedded file changes trigger recompilation
    • has_embed file existence changes trigger recompilation
    • Rust Tests: oric/src/query/embed.rs — dependency tracking

  • Implement: embed/has_embed error diagnostics — add error codes to ori_diagnostic for file-not-found, invalid path, bad UTF-8

    • File not found (with “did you mean?” suggestions)
    • Absolute path error
    • Path escapes project root error
    • Invalid UTF-8 error (when str expected)
    • Cannot infer embed type error
    • File exceeds size limit error
    • Ori Tests: tests/compile-fail/embed_errors.ori

  • Implement: Binary deduplication — multiple references to same file share one copy

    • LLVM Rust Tests: ori_llvm/tests/embed_tests.rs — deduplication
    • AOT Tests: No AOT coverage yet

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.8) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.8 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.8: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.9 Char and Byte Classification Methods

Proposal: proposals/approved/char-byte-classification-proposal.md

Standard classification methods on char and byte types for character category testing. char methods are Unicode-aware; byte methods are ASCII-only with short aliases.

Char Unicode Methods

  • Implement: char.is_alphabetic() (as is_alpha) — Unicode L* categories [done] (verified 2026-03-29)
    • Registry: compiler/ori_registry/src/defs/char.rs
    • Evaluator: compiler/ori_eval/src/methods/variants.rs
    • NEEDS TESTS: Zero spec tests — needs tests/spec/types/char_classification.ori
    • Note: Registered as is_alpha, not is_alphabetic per spec — naming gap
  • Implement: char.is_digit() — Unicode Nd category [done] (verified 2026-03-29) — NEEDS TESTS: zero spec tests
  • Implement: char.is_alphanumeric() — L* or Nd
  • Implement: char.is_whitespace() — Zs + control whitespace [done] (verified 2026-03-29) — NEEDS TESTS: zero spec tests
  • Implement: char.is_uppercase() — Lu [done] (verified 2026-03-29) — NEEDS TESTS: zero spec tests
  • Implement: char.is_lowercase() — Ll [done] (verified 2026-03-29) — NEEDS TESTS: zero spec tests
  • Implement: char.is_ascii() — U+0000..U+007F [done] (verified 2026-03-29) — NEEDS TESTS: zero spec tests
  • Implement: char.is_control() — Cc
  • Implement: Unicode lookup tables from UCD — compressed tables for L*, Nd, Zs, Lu, Ll categories

Char ASCII Methods

  • Implement: char.is_ascii_alphabetic() — a-z, A-Z
    • Rust Tests: ori_eval/src/methods/char/tests.rs — ASCII classification
    • Ori Tests: tests/spec/types/char_ascii_classification.ori
  • Implement: char.is_ascii_digit() — 0-9
  • Implement: char.is_ascii_alphanumeric() — a-z, A-Z, 0-9
  • Implement: char.is_ascii_whitespace() — space, tab, newline, CR, VT, FF
  • Implement: char.is_ascii_uppercase() — A-Z
  • Implement: char.is_ascii_lowercase() — a-z
  • Implement: char.is_ascii_hex_digit() — 0-9, a-f, A-F
  • Implement: char.is_ascii_punctuation() — ASCII punctuation ranges
  • Implement: char.is_ascii_control() — 0x00..0x1F, 0x7F

Byte Methods (Full + Short Aliases)

  • Implement: byte.is_ascii(), byte.is_ascii_alpha(), byte.is_ascii_digit(), byte.is_ascii_whitespace() [done] (verified 2026-03-29)
    • Registry: compiler/ori_registry/src/defs/byte.rs — 4 methods registered
    • Evaluator: compiler/ori_eval/src/methods/variants.rs — 4 methods implemented
    • NEEDS TESTS: Zero spec tests for all 4 methods — needs tests/spec/types/byte_classification.ori
  • Implement: Remaining byte methods + 7 short aliases — 6+ full methods not yet implemented
  • LLVM Support: LLVM codegen for byte/char classification methods
    • LLVM Rust Tests: ori_llvm/tests/aot/primitives.rs — byte/char classification codegen
    • AOT Tests: No AOT coverage yet

Conversion Methods

  • Implement: char.to_ascii_uppercase(), char.to_ascii_lowercase() — returns self if not ASCII letter

    • Ori Tests: tests/spec/types/char_conversion.ori

  • Implement: byte.to_ascii_uppercase(), byte.to_ascii_lowercase() — returns self if not ASCII letter

    • Ori Tests: tests/spec/types/byte_conversion.ori

  • Implement: char.to_digit(radix:), byte.to_digit(radix:) — returns Option<int>; radix 2..=36, panic on invalid

    • Ori Tests: tests/spec/types/char_to_digit.ori

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.9) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.9 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.9: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.10 Byte-Level String Access

Proposal: proposals/approved/byte-string-access-proposal.md

Methods for accessing raw UTF-8 bytes of a str value, enabling O(1) byte-level indexing for lexers and parsers.

str Methods

  • Implement: str.as_bytes() — zero-copy [byte] view via seamless slicing [done] (verified 2026-03-29)
    • Registry: compiler/ori_registry/src/defs/str.rs
    • Evaluator: compiler/ori_eval/src/methods/collections.rs — converts str to [byte]
    • NEEDS TESTS: Zero dedicated spec tests — needs tests/spec/types/str_as_bytes.ori
    • LLVM Support: Seamless slicing behavior not implemented in LLVM
  • Implement: str.to_bytes() — owned [byte] copy [done] (verified 2026-03-29)
    • Evaluator: compiler/ori_eval/src/methods/collections.rs
    • NEEDS TESTS: Zero dedicated spec tests — needs tests/spec/types/str_to_bytes.ori
  • Implement: str.byte_len() — O(1) UTF-8 byte count [done] (verified 2026-03-29)
    • Evaluator: compiler/ori_eval/src/methods/collections.rs
    • NEEDS TESTS: Only 1 indirect test in primitives.ori — needs dedicated tests/spec/types/str_byte_len.ori
  • Implement: Flatten behavior — as_bytes() on substring seamless slice produces single-level [byte] view

str Associated Functions

  • Implement: str.from_utf8(bytes:) — validate UTF-8, return Result<str, Error> [done] (verified 2026-03-29)
    • Registry: compiler/ori_registry/src/defs/str.rs
    • Evaluator: compiler/ori_eval/src/methods/collections.rs
    • NEEDS TESTS: Zero dedicated spec tests — needs tests/spec/types/str_from_utf8.ori
  • Implement: str.from_utf8_unchecked(bytes:) — skip validation, requires unsafe [done] (verified 2026-03-29)
    • Registry: compiler/ori_registry/src/defs/str.rs
    • Evaluator: compiler/ori_eval/src/methods/collections.rs
    • NEEDS TESTS: Zero dedicated spec tests — needs tests/spec/types/str_from_utf8_unchecked.ori

LLVM Support

  • LLVM Support: Codegen for as_bytes, to_bytes, byte_len, from_utf8, from_utf8_unchecked

    • LLVM Rust Tests: ori_llvm/tests/aot/strings.rs — byte access codegen
    • AOT Tests: No AOT coverage yet

  • /tpr-review passed — independent review found no critical or major issues (or all findings triaged)

  • /impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.

  • Subsection close-out (7A.10) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.10 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.10: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.

  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.

  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

7A.11 Section Completion Checklist

  • All items above have all checkboxes marked [ ]
  • Re-evaluate against docs/compiler-design/v2/02-design-principles.md
  • 80+% test coverage, tests against spec/design
  • Run full test suite: ./test-all.sh
  • LLVM Support: All LLVM codegen tests pass
  • /tpr-review passed — independent Codex review found no critical or major issues (or all findings triaged)
  • /impl-hygiene-review passed — implementation hygiene review clean (phase boundaries, SSOT, algorithmic DRY, naming). MUST run AFTER /tpr-review is clean.
  • /improve-tooling retrospective completed — MANDATORY at section close, after both reviews are clean. Reflect on the section’s debugging journey (which diagnostics/ scripts you ran, which command sequences you repeated, where you added ad-hoc dbg!/tracing calls, where output was hard to interpret) and identify any tool/log/diagnostic improvement that would have made this section materially easier OR that would help the next section touching this area. Implement every accepted improvement NOW (zero deferral) and commit each via SEPARATE /commit-push. The retrospective is mandatory even when nothing felt painful — that is exactly when blind spots accumulate. See .claude/skills/improve-tooling/SKILL.md “Retrospective Mode” for the full protocol.

Exit Criteria: Core built-in functions working correctly

  • Subsection close-out (7A.11) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-7A.11 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 7A.11: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.
  • /sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.
  • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.