The Forth Deck mini: a portable Forth computer with a discrete CPU

Enthusiasm for the Forth Deck mini & discrete CPU

• Many are delighted by a hand‑built Forth machine with a discrete, microcoded CPU.
• The minimal ALU (single‑bit NOR, multi‑cycle 8‑bit ops) and tiny chip count are praised as elegant engineering.
• People see it as a fun, educational platform for playing with algorithms, games, and small tools directly on hardware.

“AlphaSmart for programmers” / Model 100 successors

• Several want a modern, distraction‑free portable like an AlphaSmart or TRS‑80 Model 100:
  • Real keyboard, low‑power reflective display, long battery life, no internet.
• The Model 100 is repeatedly cited as near‑ideal: great keyboard, ROM tools, AA batteries; main complaints are tiny 8‑line display and clunky data transfer.
• Some propose a $200 modern unit; BOM constraints (especially display cost) are discussed.

Ultra‑low‑power “zorzpad” and longevity goals

• A related project aims at a solar‑powered, batteryless notebook (zorzpad) with:
  • Sub‑milliwatt consumption, 53‑year design life, no charging ports, no commodity key switches.
• Emphasis is on survivability without industrial supply chains: no replace‑often batteries, minimal moving parts, heavy use of redundancy and potting.

Displays: e‑ink vs memory‑in‑pixel LCD

• E‑ink is seen as attractive but probably too power‑hungry and slow for continuous editing on microwatt budgets.
• Sharp memory‑in‑pixel monochrome LCDs are championed:
  • True bistability, ~100 µW level, fast refresh, good sunlight readability.
• Proposals for hybrid designs (small “live” LCD line + large e‑ink page) are compared; memory LCD is argued to be superior overall except for grayscale and size.

Batteries, energy storage, and reliability

• Strong criticism of batteries for ultra‑long‑life devices:
  • Limited shelf life (~10 years), frequent failures (swelling, leaks, thermal issues), and dependence on future markets.
• Others suggest using standard, easily replaced cells (AA/AAA/coin) to gain practicality.
• Alternatives discussed:
  • Supercaps with careful derating and encapsulation.
  • Mechanical or pneumatic charging (pull‑string, bike pump, compressed air) as external, replaceable modules.

Programming models: Forth, UXN/Varvara, CHIP‑8, Lisps

• Forth is valued for interactivity and bootstrapping, but many find it hard to read at scale.
• UXN/Varvara gets praise as a frugal, portable VM with a growing, self‑hosting ecosystem, but:
  • Critics argue it’s CPU‑inefficient (interpretive overhead ~20×), limited to 64K RAM, and unfriendly for complex, data‑heavy tasks (e.g., Norvig‑style spell corrector).
  • Varvara’s 2‑bit planar graphics don’t match 1‑bit memory LCDs well; full GUI at 60 Hz is seen as incompatible with ultra‑low‑power goals.
• CHIP‑8/Octo are noted as fun but IO‑poor and community‑fragmented.
• There is an extended side discussion on Lisp vs Python:
  • Trade‑offs between flexibility, readability, dynamic typing, and collaboration; Lisp seen as powerful but socially and technically challenging for large shared codebases.

Other hardware & prior art

• References to:
  • AlphaSmart, TRS‑80 Model 100, Psion 5, Oric‑1, Jupiter Ace, Open Firmware machines, OLPC XO, GEOS on C64.
  • Modern ESP32 devices: M5Stack Cardputer, Devterm, meshtastic boards; some already have Forth ports.
  • Commercial “distraction‑free writers” (e.g., Freewrite) viewed as too expensive or ergonomically off.

Design trade‑offs & UX

• Debates over GUI vs pure text:
  • Argument that with modern low‑power MCUs and reflective LCDs, efficient GUIs are feasible even under 1 mW; historical “text is cheaper” assumptions may no longer hold.
• Text editors:
  • Benchmarks show wide power gaps: ultra‑minimal editors can be ~1000× more instruction‑efficient than feature‑rich IDEs.
  • Discussion on designing new editors specifically for tiny, slow, 1‑bit screens, including block editors, single‑symbol views, and strategies to minimize screen updates.