Result: done
Cause: 7189b7fc (S1) sysroot_package.nika: sequences::the_corpus_has_no_more_unanswered_method_calls_than_it_had, 16 against 14 - Item::Test { body, .. } | Item::Bench { body, .. } => body left body untyped (block.stmts.is_empty() and .len() unanswered); found by bisecting the file.
Fixed: pattern_parts (check_patterns.nika) types an or-pattern’s name as the type every alternative gives it, left untyped where they differ (ADR-291 D22); lower-std rerun.
Tests: corpus-count test ok (14, no ceiling change); tests/compiler 549 ok incl. new an_or_pattern_binds_the_type_its_alternatives_agree_on; it sysroot:: 15 ok; cargo fmt; the ceiling test is the guard (the new program passes before the fix too).
Found: none
Filed: #1150 (debug cargo build -p nikaia hits the recursion limit in winnow-grammar rt.rs (release/fast build fine, CI is release); not touched, probe: cargo build -p nikaia.
Files: crates/nikaia-std/src/tools/check_patterns.nika crates/nikaia-std/src/tools/lowered/check_patterns.rs tests/compiler/tests/an_or_pattern_binds_the_type_its_alternatives_agree_on.nika tests/compiler/tests/an_or_pattern_binds_the_type_its_alternatives_agree_on.stdout
Commit: check: an or-pattern’s names are typed where its alternatives agree (fix forward: corpus unanswered calls)
Body: S1’s sysroot_package.nika matched Item::Test { body, .. } | Item::Bench { body, .. }; an or-pattern’s bindings were left untyped, so two method calls on body went unanswered and the corpus count rose to 16. The names are now typed where every alternative gives the same type; where they differ nothing changes. Count back at 14.
Fixes: none
Changelog: check: a name an or-pattern binds is typed where every alternative gives it the same type
Attribution:
Co-Authored-By: Claude Sonnet 5.5 noreply@anthropic.com
Claude-Session: https://claude.ai/code/session_01GFeMqxRbtUMiGVyDMxsvEq
Pushed: local 0647154f74622e093bdf08c01f5462572b2f3212
