float64

Decision log

Lightweight ADR-style format: each decision has a status (decided / open), context and consequence. Open decisions are this project’s real backlog.

DEC-001 — Project name: float64 ✅ decided

Floating-point home, double precision. Runner-up jangada-de-pedro stays reserved as the future blog brand (domain jangadadepedro.pt to be registered). drop-anchor and waterfall honourably discarded.

DEC-002 — Repo description ✅ decided

Double-precision floating home. Plans, research and telemetry for migrating one engineer (and one dog) from solid ground to the Tagus.

GitHub topics: houseboat, liveaboard, tejo, lisboa, boatops. Website field reserved for jangadadepedro.pt. The original PT tagline (“Casa de vírgula flutuante, precisão dupla…”) stays reserved for the blog — the pun only works in Portuguese.

DEC-003 — Methodology: measure before choosing ✅ decided

Requirements derived from data, not brochures. Before any platform decision: 3–4 weeks of telemetry on current life (ESP32/M5Stack already in the drawer — humidity and temperature per room, smart plugs on key loads), 8–10 interviews with actual liveaboards, and an activity→requirement matrix with MoSCoW priorities (Phase 1.3). The matrix constrains the boat model, not the other way around. The January POC gets acceptance criteria defined before stepping aboard; a NO-GO in February is also a success — it cost €400 to find out instead of €140k.

DEC-004 — Home berth: Oeiras vs Parque das Nações ⚠️ OPEN

Context: the crossing scenario test (research/oeiras-estuary-crossing.md) revealed a requirements conflict: Apartboat XL + Oeiras berth + weekend crossings don’t coexist.

Option A — Oeiras is a Must: the model stops being a houseboat and becomes something CE-C that can handle the estuary mouth — trawler, motorsailer, or a liveable power cat à la Overblue. You trade square metres for seaworthiness.

Option B — Anchoring getaways in home mode are a Must: keep the houseboat, base at Parque das Nações (Mar da Palha: sheltered, no bar; Seixal and Alcochete 1h–1h30 away). Oeiras becomes 20 minutes by train.

Status: undecided. Note: this may not need deciding yet — Phase 1 telemetry and the interviews should reveal which of the two uses is actually a Must and which is a Nice-to-have.

DEC-005 — Repo language: English ✅ decided

Content switched from PT-PT to English on 2026-07-28 (folders renamed: planos→plans, investigacao→research, telemetria→telemetry). Portuguese terms kept where they’re the proper noun (nortada, Mar da Palha, carta de Marinheiro).

DEC-006 — Site architecture: plain Jekyll on GitHub Pages ✅ decided

The repo doubles as the project site, built by GitHub Pages’ built-in Jekyll — no Actions, no npm, no build step. Custom layouts + CSS; blog entries in _posts/, boat archive as the _boats collection, scenario suite as the _scenarios collection (front matter: id, status pass/fail/open, dimension). Adding a markdown file is publishing. Constraint accepted: only GitHub-whitelisted Jekyll plugins.

DEC-007 — Knowledge before contact ✅ decided

The marina email (SPK-1) is deferred until Phase 0 builds real knowledge: a boating course/licence path chosen and started, a couple of nights actually slept on a rented houseboat, and the boat archive populated with real pricing, delivery conditions and used-market data. Rationale: walk into the marina conversation as an informed counterpart, not a dreamer — better questions, better credibility, better negotiating position. Trade-off accepted: the one project-killing answer (liveaboard policy, SCN-012) arrives later; mitigation is to keep Phase 0 short and time-boxed.

Immediate backlog (from roadmap v0.2)

TODO: export the “Roadmap casa flutuante” document (artifact from the claude.ai conversation) into plans/roadmap-v0.2.md — it’s the source of these phases and the full backlog. Translate to English on the way in.