You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Library modules with frontend weight: tuples, Proxy, Partial.Unsafe, Prim #96
Part of #76 (Phase 3). A smaller slice so Phase 3 does not become one unbounded change.
Do
Provide the library modules that carry real frontend weight, in this order:
Data.Tuple (8 imports) — maps to a closed record { _1, _2 }, which is already how FE-06 represents a tuple. Mostly a declaration.
Type.Proxy (17 imports) — the type proxy the corpus uses for kind and type-level tests. Needs data Proxy (a :: Type) with no runtime payload.
Partial.Unsafe (12 imports) — unsafeCoerce/unsafePerformEffect. These must be honest about what they bypass; DEC-07 and DEC-10 constrain the representation they can reach.
Each module is declared in stdlib/lib/trusted in prefix order, compiles, and executes under Wasmtime; docs/design/backend/wasm/wasi-platform-library.md records each with its FE/BE owner.
The 2026-10-17 L2 board reports 0 missing-library blockers. Data.Tuple, Type.Proxy, and Partial.Unsafe are in the locked package, and no in-scope passing file is first-blocked on them. Corpus cases moves from 17 to 0.
This issue stays open. The previous 17 was the Type.Proxy import count from the issue body, not a fresh first-blocker recount. Data.Symbol.unsafeCoerce is a different foreign and stays with #124.
Evidence: L2 in docs/design/D-04-suite-roadmap.md (2026-10-17, 886cd82c, vendored corpus).
Recounted 2026-10-18 on master f5abccda: Data.Tuple, Type.Proxy, Partial.Unsafe, and the Prim.* library-visible half are all in the locked package (e556bf2f), and the L2 board reports 0 missing-library blockers (Corpus cases 17 → 0, first recorded 2026-10-16). Prim.Row/Prim.RowList/Prim.Symbol/Prim.Int relations all have rule entries (#120, closed); Prim.Number is intentionally not an official child module (recorded in D-04's M2). Closing on the measured gate.
Activity
Correct the first-stage blocker decomposition and close out Phase 2
Land Data.Monoid and Data.Foldable
biuld commentedon Oct 16, 2026
The 2026-10-17 L2 board reports 0 missing-library blockers.
Data.Tuple,Type.Proxy, andPartial.Unsafeare in the locked package, and no in-scopepassingfile is first-blocked on them.Corpus casesmoves from 17 to 0.This issue stays open. The previous 17 was the
Type.Proxyimport count from the issue body, not a fresh first-blocker recount.Data.Symbol.unsafeCoerceis a different foreign and stays with #124.Evidence: L2 in
docs/design/D-04-suite-roadmap.md(2026-10-17,886cd82c, vendored corpus).biuld commentedon Oct 18, 2026
Recounted 2026-10-18 on master
f5abccda:Data.Tuple,Type.Proxy,Partial.Unsafe, and thePrim.*library-visible half are all in the locked package (e556bf2f), and the L2 board reports 0 missing-library blockers (Corpus cases17 → 0, first recorded 2026-10-16).Prim.Row/Prim.RowList/Prim.Symbol/Prim.Intrelations all have rule entries (#120, closed);Prim.Numberis intentionally not an official child module (recorded in D-04's M2). Closing on the measured gate.