Epic: Evaluate TS-facade migration for remaining dual-C# Node modules

[nickna]

Goal

Evaluate (do not big-bang) migrating the remaining dual-hand-written-in-C# Node modules to the proven TS-facade-over-thin-primitive pattern that already collapsed fs/os/process. The compiled IL-emitter side alone is large: http ~5,600 LOC, stream ~4,600, dns ~4,400, net ~4,300, crypto ~2,900, tls ~1,600 — each mirrored by a separate interpreter impl.

The win is real but uneven: facade migration is most valuable where a module is logic-over-simple-primitives. Socket/stream modules have large stateful primitives, so the win there is the API-glue layer, not the core. Best driven per-module, coordinated with the existing coverage epics (e.g. child_process #1012).

Sub-issues (rough priority order)

• Spike: facade-migrate one low-risk dual-C# module (e.g. zlib) as proof-of-pattern #1147 — Spike one module (e.g. zlib/dgram) to the facade pattern; measure the LOC delta.
• Assess facade-ability of the stateful socket/stream cluster (http/net/tls/stream/dns/crypto) #1148 — Assess the stateful cluster (http/net/tls/stream/dns/crypto): what's facade-able vs primitive.

---

Working agreement for every child PR

• Green on dotnet test, SharpTS.Test262, and SharpTS.TypeScriptConformance (no baseline regression).
• Interpreter and compiled output must remain behaviourally identical; add dual-mode tests where applicable.
• Standalone compiled DLLs must not gain a SharpTS.dll dependency (see CLAUDE.md "Standalone DLL Constraint").

Source: codebase-wide maintainability audit, 2026-06-29.

[nickna]

Epic status:

• Spike: facade-migrate one low-risk dual-C# module (e.g. zlib) as proof-of-pattern #1147 (spike) — DONE: zlib migrated to stdlib/node/zlib.ts over primitive:zlib in PR refactor(zlib): #1147 migrate zlib to TS facade over primitive:zlib (epic #1096 spike) #1209. Realized: −823/+535 lines (constants/codes + all 11 async callback forms + arg shuffling now one TS implementation shared by both modes). Compiled perf flat within noise, zlib.constants access 67× faster compiled, and a latent interp↔compiled async divergence fixed. Full verification green (xUnit 15,234/0, Test262/TSConf baseline-identical, standalone preserved). Writeup with friction points on Spike: facade-migrate one low-risk dual-C# module (e.g. zlib) as proof-of-pattern #1147.
• Assess facade-ability of the stateful socket/stream cluster (http/net/tls/stream/dns/crypto) #1148 (assessment) — DONE and closed: per-module facade-ability assessment posted, with recommended sequence dns (S) → net partial (M) → tls partial (M) → http partial (L); crypto opportunistic-only (~5% glue), stream infeasible (~1% glue, classes are the API), dgram skip (~3%).
• Follow-up filed: CJS require() of a stdlib facade fails when its primitive:* deps were not preloaded #1210 (pre-existing require()-of-facade primitive-preload gap, shared with fs since fs: adopt primitive:fs + TypeScript facade architecture (decision) #969).

Suggested next child if the epic proceeds: facade-migrate dns/dns/promises — one-shot request/response shape, ~50% glue including the entire duplicated promise surface (RuntimeEmitter.DnsPromises.cs deletes outright), the exact fs/fs/promises derivation pattern.