lab/sx
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs(reify): Phase 4 self-reference = explicit reserve()/complete()

Browse Source 
User-directed API decision: replace the reify_rec((self)=>...) closure
with an explicit reserve() -> Type handle + complete(handle, info) pair.
reserve() returns a forward nominal Type usable freely in any later
TypeInfo (*List, []List, and across types for mutual recursion the
one-self closure couldn't express); reify(info) stays as the one-shot
sugar. Maps onto existing reserve->complete machinery. Captured in
PLAN-REIFY Phase 4 + Contract 5 + CHECKPOINT-REIFY.
This commit is contained in:
agra
2026-06-16 18:38:49 +03:00
parent 353109206b
commit ae5de1e687
2 changed files with 34 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
current/CHECKPOINT-REIFY.md
View File

@@ -61,12 +61,16 @@ only at the `E :: reify(...)` const-decl site (`decl.zig`) and reads a LITERAL
(and likely generalize the literal-AST reader, or evaluate via the interpreter, for
non-literal `TypeInfo`).
OPEN DESIGN Q (raised by user, = Phase 4): self-referencing enum via `*Self`/`[]Self`
payloads. Current `reifyType` resolves each payload eagerly through `buildEnumInfo`,
so a `*Self` payload has no type to point at yet — Phase 4 needs the
reserve-placeholder→complete path (`nominal.zig` reserve + `types.zig:442`): intern
the enum's nominal slot FIRST, expose it as `Self`, then resolve payloads. By-value
recursion stays rejected (infinite size).
SELF-REFERENCE = Phase 4, API DECIDED (user-directed): explicit **`reserve()` →
`complete()`** (NOT a `reify_rec((self)=>…)` closure). `reserve()` returns a forward
nominal `Type` handle (named from the `::` LHS) usable freely in any later `TypeInfo`
— `*List`, `[]List`, and across types for MUTUAL recursion the one-`self` closure
couldn't express; `complete(handle, info)` fills the body. `reify(info)` stays as the
one-shot sugar. Maps onto existing machinery: reserve = empty nominal slot
(`reserveShadowEnumSlot`-style), complete = `buildEnumInfo` + `updatePreservingKey`,
struct-stub→tagged_union re-key already handled (`internNamedTypeDecl`'s
`adoptsForwardStructStub`). Loud invariants: never-completed reserve = hard error;
by-value self-inclusion rejected (infinite size). Full write-up in PLAN-REIFY Phase 4.
## Known issues
None yet.