Skip to content

Pumice desktop app integration

Status: open/draft

Implement D-0016-pumice-app-integration: bring the standalone pumice Tauri v2 + Nuxt 4 desktop application into this monorepo as apps/pumice, proving D-0012-monorepo-tooling’s polyglot workspace against a real, load-bearing application rather than a thin wrapper. Land it in two phases — co-tenant first (shared tooling, no code coupling), then converge (shared crates, adapter seam) — so each step is independently reviewable.

  • apps/pumice exists and builds under moon (moon run pumice:build), with a thin moon.yml shelling to pumice’s existing commands under the system toolchain.
  • Pumice’s frontend runs on Bun; the three pnpm lockfiles are retired and no stray pnpm-lock.yaml remains (per D-0012-monorepo-tooling).
  • Pumice’s Rust crates compile under moon’s rust toolchain as an isolated cargo unit; the 1.93.0-vs-stable divergence is resolved.
  • The ontogen codegen pipeline is a moon task with correct inputs/outputs, and the codegen-drift gate runs in CI.
  • Pumice’s checks (frontend, Rust, codegen-drift) run under moon ci on a pumice-affected path that installs Tauri’s Linux system dependencies.
  • Pumice’s template-sync and helper scripts work from the apps/pumice location (or the sync is retired deliberately).
  • Phase 2: at least one genuinely shared Rust crate is promoted to packages/rust/, and the convergence question with T-CKC1 is resolved.

Member tasks, sequenced by wave (dependency order). Each links its full implementation-ready spec.

Wave 4 — convergence + extraction (Phase 2)

Section titled “Wave 4 — convergence + extraction (Phase 2)”

The shared-codegen convergence question (ontogen vs the zod dashboard-client pipeline) is framed in D-0017-ontogen-shared-codegen-convergence; its outcome lands in T-Z5MK once a concrete shared-model trigger appears.

  • Phase 1 (Waves 1–3) leaves pumice a self-contained co-tenant: it shares the moon graph, CI, justfile, and lefthook, but no application code.
  • Phase 2 (Wave 4) is a strong candidate for its own follow-up milestone — it is gated on the co-tenant landing being stable and on a concrete shared-code need, mirroring how T-CKC1 is framed as a later, separately-scheduled capstone.

← Back to Milestones