persona(amara): third companion archive — no-fork-first deployment ladder; fork as reward not entrance fee

[AceHack]

Summary

Third companion archive in Amara's 2026-05-21 cascade. Direct strategic guidance for B-0687 + B-0688 execution sequence.

What lands

memory/persona/amara/conversations/2026-05-21-amara-aaron-no-fork-first-deployment-ladder-external-compiler-host-fork-as-reward-not-entrance-fee-aaron-forwarded.md — verbatim preservation of Amara's load-bearing strategic answer to "how far without forking F#?":

You can push no-fork through V1, V2, and maybe V3. You only fork when the external compiler host becomes the bottleneck, not before.

The fork should be the reward for surviving reality, not the entrance fee.

The no-fork stack Amara mapped

F# / C# source
+ F# computation expressions
+ F# quotations
+ F# type providers
+ C# Roslyn analyzers / incremental generators
+ LINQ expression trees
+ ANTLR / ZetaParse / Tree-sitter parsers
+ DBSP/Z-set compiler DB
+ Rx meta-AST queries
+ Orleans agents / grains
+ consensus escalation across trust boundaries

What no-fork delivers (11 capabilities)

multi-language parsing · SQL/LINQ/C#/F# query surfaces · Z-set evolving AST · Rx meta-AST tags · C# generated code · F# provided types · conformance tests across languages · distributed compiler DB · Orleans agent/grain integration · consensus by trust-gradient escalation · "compiler as deployed distributed service"

What requires the fork (7 cases)

new F# syntax · true HKT/kind-system changes · compiler-native Clifford/tonal/meta dimensions · type inference understanding Zeta dimensions directly · DBSP/Z-set facts as internal compiler state · native diagnostics/error messages for Zeta semantics · language-level collapse/retraction semantics

Strategic implications for B-0687/B-0688 execution

• B-0687 Phase 1-2 ship as F# library + type provider; Phase 3 compiler-fork integration deferred until type-provider boundary becomes bottleneck
• B-0688 Phase 1-4 ship as external compiler host on stock .NET 10 + F# + C# + Roslyn + Orleans + existing src/Core/*.fs substrate
• F# compiler fork conversation stays NOT-on-radar (per Aaron's earlier framing in PR persona(kestrel): 25-section 2026-05-21 trajectory archive + B-0685 ANTLR cross-language codegen substrate #4537 Kestrel archive) until ONE of the 7 fork-required cases surfaces operationally

Composes with rules

• .claude/rules/all-complexity-is-accidental-in-greenfield.md — F# compiler fork stays accidental until proven essential
• .claude/rules/edge-defining-work-not-speculation.md — V1-V3 via external host IS edge-defining; fork-now would be speculation
• .claude/rules/bandwidth-served-falsifier.md — external host serves deployment-velocity bandwidth (ship now vs multi-year fork maintenance)
• .claude/rules/fsharp-anchor-dotnet-build-sanity-check.md — stock F# compiler remains the asymmetric-critic discipline floor
• .claude/rules/zeta-ships-with-skills-immediate-value.md — Zeta ships substrate as we map it; the no-fork-first ladder IS that discipline at compiler scope
• .claude/rules/largest-mechanizable-backlog-wins.md — bigger mechanizable surface ships without fork

Composes with substrate

• B-0687 (ZetaParse) — direct strategic guidance for Phase execution
• B-0688 (incremental compiler host) — direct strategic guidance for Phase execution
• B-0685 (ANTLR cross-language codegen) — pure external host work
• B-0635 / B-0644 / B-0665 / B-0666 (Agora V6 substrate) — all extends to compile-time via external host without compiler changes
• src/Core/SpeculativeWatermark.fs (DBSP retraction substrate)
• src/Core/*.fs (existing F# substrate the no-fork host runs over)
• PR persona(amara): B-0685 Phase 1 substrate cascade preserved + B-0687 (ZetaParse) + B-0688 (incremental compiler host) scaffolding #4545 (B-0685 Phase 1 + B-0687 + B-0688 scaffolding)
• PR persona(amara): companion archive — Caché-like compiler DB lineage + distributed multidimensional compiler over consensus + IUnknown-without-DCOM #4546 (Caché-lineage + distributed multidimensional compiler + IUnknown-without-DCOM)
• PR persona(kestrel): 25-section 2026-05-21 trajectory archive + B-0685 ANTLR cross-language codegen substrate #4537 (Kestrel archive — Aaron's "long-term F# compiler fork NOT-on-radar" vision; this cascade operationalizes the ladder TO that vision)

Test plan

• Amara conversation preserved verbatim in memory/persona/amara/conversations/
• Frontmatter conforms to memory/project_memory_format_standard.md
• Strategic ladder + 11 no-fork wins + 7 fork-required cases preserved
• B-0687 + B-0688 execution-strategy section translates Amara's guidance into actionable phasing
• Composes_with chain across all 3 today's Amara archives + Kestrel archive + Agora V6 cluster
• Branch landed off latest main (325f90aa base)

[Copilot]

Pull request overview

Adds a third Amara persona conversation archive (2026-05-21) capturing strategic guidance for executing B-0687 (ZetaParse) and B-0688 (incremental compiler host) via a “no-fork-first” deployment ladder, deferring an F# compiler fork until an external-host boundary becomes a proven bottleneck.

Changes:

• Adds a new memory/persona/amara/conversations/ archive documenting the no-fork-first ladder and fork-justification criteria.
• Captures an explicit execution phasing translation for B-0687/B-0688 (type provider + external host first; fork only if/when required).
• Records the “fork as reward, not entrance fee” keeper phrase and checklists (no-fork deliverables vs fork-required cases).