Section 10: Control Flow

Goal: Complete control flow constructs

SPEC: spec/14-expressions.md, spec/11-blocks-and-scope.md, spec/16-control-flow.md, spec/17-errors-and-panics.md PROPOSALS:

proposals/approved/if-expression-proposal.md — Conditional expression semantics

proposals/approved/error-return-traces-proposal.md — Automatic error trace collection

proposals/approved/loop-expression-proposal.md — Loop expression semantics

proposals/approved/while-loop-proposal.md — While loop expression (sugar over loop)

proposals/approved/labeled-block-early-exit-proposal.md — Labeled block early exit (break:label from blocks)

10.1 if Expression

Proposal: proposals/approved/if-expression-proposal.md

10.2 for Expressions

NOTE: This is the for x in items do/yield expr expression syntax for iteration. The for(over:, match:, default:) pattern with named arguments is a separate construct in Section 8.

Imperative form (do):

Collection building (yield):

With guards:

For-yield comprehensions (proposals/approved/for-yield-comprehensions-proposal.md):

10.3 loop Expression

Proposal: proposals/approved/loop-expression-proposal.md

Labeled loops:

Labeled loop semantics (proposals/approved/labeled-loops-proposal.md):

10.3B Labeled Block Early Exit

Proposal: proposals/approved/labeled-block-early-exit-proposal.md

block:name { ... break:name value ... } — early exit from named blocks without return. Extends the label system to plain blocks.

10.3A while Expression

Proposal: proposals/approved/while-loop-proposal.md

Syntactic sugar: while condition do body desugars to loop { if !condition then break; body }. Always void type.

Sync Points: `ExprKind::While` (multi-crate sync required)

If while is represented as its own ExprKind::While variant (rather than desugared in the parser):

ori_ir — Add ExprKind::While { condition, body, label } variant
ori_types — Type checking: condition must be bool, result type is void, handle E0860/E0861
ori_eval — Evaluation: desugar to loop+break or direct while execution
ori_llvm — Codegen: br instruction with condition check, loop body block, exit block

If desugared in the parser to loop { if !condition then break; body }, only steps 1-2 (parser + IR) are needed.

Implementation

10.4 Error Propagation (?)

Error Return Traces (proposals/approved/error-return-traces-proposal.md):

10.5 Let Bindings

10.6 Scoping

10.7 Panics

10.8 Index Expressions — [partial] Evaluator Complete

Implement: # length symbol in index brackets (list[# - 1]) — spec/14-expressions.md § Index Access [done] (2026-02-10) (verified 2026-03-29)
- Parser: Parse # as ExprKind::HashLength inside [...] — ori_parse/src/grammar/expr/postfix.rs
- Type Checker: Resolve HashLength to receiver’s length type (int) — ori_types/src/infer/
- Evaluator: Evaluate HashLength as len(receiver) in index context — ori_eval/src/interpreter/mod.rs
- ARC Lowering: CanExpr::HashLength resolves to stored hash_length variable via len() call — properly flows through to LLVM
- Ori Tests: tests/spec/expressions/index_access.ori — hash_last, hash_second_last, hash_first, hash_middle, hash_arithmetic (35 total tests)
- LLVM Support: LLVM codegen for hash length in index — ARC lowering resolves HashLength to len() variable which flows through correctly
- LLVM Rust Tests: ori_llvm/tests/collection_tests.rs — hash length codegen [aspirational — file does not exist]
- AOT Tests: No AOT coverage yet

Implementation Notes:

Added IN_INDEX context flag to ParseContext
Parser recognizes # (TokenKind::Hash) as ExprKind::HashLength only inside index brackets
Type checker and evaluator already had full support for HashLength
ARC lowering properly handles HashLength by resolving to a len() call variable
/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 (10.8) — 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-10.8 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 10.8: 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.

10.9 Section Completion Checklist

All items above have all three checkboxes marked [ ]
Spec updated: spec/14-expressions.md, spec/16-control-flow.md reflect implementation
CLAUDE.md updated if syntax/behavior changed
80+% test coverage
Run full test suite: ./test-all.sh
/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.
NEGATIVE TESTS GAP: Add #compile_fail negative tests for control flow features (zero exist as of 2026-03-29) — non-bool condition, branch type mismatch, non-void if-without-else, break type mismatch in loop, continue-with-value in loop, ? in non-Result/Option function

Exit Criteria: All control flow constructs work including labeled loops, scoping, and panic handling

/tpr-review passed — independent review found no critical or major issues (or all findings triaged)
Subsection close-out (10.9) — 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-10.9 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 10.9: 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.

Verification Notes (2026-03-29)

BUG: if true then 42 compiles silently (returns void) instead of producing a type error for non-void/non-Never then-branch without else. infer_if() returns Idx::UNIT regardless of then-branch type when else is absent. Per spec (14-expressions.md Conditional), non-void/non-Never then-branches without else should produce a type error.

Stale data corrected: loops.ori 29->35 tests, block_scope.ori 3->23 tests, lambdas.ori 29->30 tests. Three [ignored] markers on scoping AOT tests removed (all pass). HashLength LLVM placeholder note removed (ARC lowering properly handles it). Wrong test reference on panic(message) corrected (was coalesce.ori which tests ??, not panic).

Status changes: 4 items changed from [ ] to [partial] — Never coercion in branches (10.1), struct literal restriction (10.1), implicit panics (10.7), all have partial implementation but missing tests.

Critical gap: ZERO #compile_fail negative tests exist for any control flow feature. 5 LLVM unit test files referenced in roadmap are aspirational (do not exist) — covered by AOT integration tests instead. 23 referenced test files total do not exist (annotated as aspirational).

Verification Stats

total_items: 82
verified: 40
correct_unchecked: 35
should_be_partial: 4
stale_data: 6
wrong_test_ref: 1
bugs_found: 1
negative_test_gap: ZERO #compile_fail negative tests for any control flow feature