Core-13 pure: add CI workflows, VM e2e tests, LLVM parity bridge (minimal); do not touch private docs

This commit is contained in:
Tomoaki
2025-09-07 07:28:53 +09:00
parent 07350c5dd9
commit d62114c705
383 changed files with 15221 additions and 382 deletions

View File

@ -2,11 +2,15 @@
This changelog tracks highlevel milestones while Core MIR and Phase 12 evolve. For detailed perfile history, see git log and docs under `docs/development/roadmap/`.
## 20250906
- Core13 flip complete: code/tests enforce Core13 minimal kernel. Normalizations (Array/Ref→BoxCall, TypeCheck/Cast/Barrier/WeakRef unification) are ON by default via env (NYASH_MIR_CORE13=1). New tests validate normalization.
- Docs synced: step50 marked done; DEV quickstart points to Core13 reference.
## 20250904
- Phase 12.7A complete: peek, continue, `?` operator, lambda, field type annotations. Language reference updated.
- Phase 12.7B (basic) complete: parserlevel desugaring for `|>`, `?.`, `??`, `+=/-=/*=/=`, `..` behind `NYASH_SYNTAX_SUGAR_LEVEL`.
- Docs: language reference and Phase 12.7 README updated to reflect basic completion; extensions tracked under gated plan.
- MIR Core migration: enforcing Core15 in code/tests during transition; Core13 target defined in docs; final flip planning in progress.
- MIR Core migration: previously enforcing Core15 during transition; superseded by 20250906 Core13 flip.
## 20250903
- Nyash ABI TypeBox integration stabilized across core boxes; differential tests added; loader defaults adjusted (builtin + plugins).
@ -14,5 +18,5 @@ This changelog tracks highlevel milestones while Core MIR and Phase 12 evolve
---
Notes
- Core15 vs Core13” migration: Implementation currently enforces 15 for stability; docs include Core13 target reference. Final flip (docs/refs/entrypoints) is tracked under `docs/development/roadmap/mir/core-13/step-50/`.
- Core13 is canonical minimal kernel. Historical Core15 notes remain under `docs/development/roadmap/` for reference.
- Phase 12.7B desugaring is gated by `NYASH_SYNTAX_SUGAR_LEVEL`; tokenizer additions are nonbreaking.