nimony-hl — shared High-Level IR

The shared lowering for the idiomatic Nim-to-high-level backends: it turns nimony’s sem’d IR into a small, target-neutral High-Level IR (HL-IR) that the thin language emitters — nimony-ts and nimony-py — render into TypeScript and Python.

Status: early scaffold · private repo. No lowering code has landed yet. Access via Discord timbuktu_guy.

Why this repo exists

TypeScript and Python differ almost entirely in surface syntax, not in semantic model — both have real objects with tracing GC, reference semantics, native closures, try/finally, and list/dict/set. So ~85% of an idiomatic backend is the same for both: the Nim-semantic lowering. That shared part lives here, once, instead of being duplicated across the two language repos.

nimony  ──▶  <module>.s.nif   (sem'd, pre-hexer)
                  │
                  ▼
   ┌──────────────────────────────────┐
   │  nimony-hl  (this repo)           │   the shared ~85%
   │  sem'd NIF  →  High-Level IR      │
   │   • structured control flow       │
   │   • inline iterators; defer→finally│
   │   • operator calls resolved        │
   │   • Option→nullable, distinct→base │
   │   • value-copy placement from      │
   │     nimony's move/copy hooks       │
   │   • every node carries an HL type  │
   └──────────────────────────────────┘
        │                          │
        ▼                          ▼
   nimony-ts emitter          nimony-py emitter
   (HL-IR → TypeScript)       (HL-IR → Python)

HL-IR is a small target-neutral AST, each node tagged with an HL type (Int/Big/Float/Bool/Str/List/Dict/Set/Obj/Ref/Union/Func/Opt) so the emitters can make the few divergent decisions (e.g. TS number/bigint, Map keying). Full design in docs/design.md in the repo.