Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

CodeWalk: Discovering Computer Science in the Physical World with Local Gemma

Описание и применение

CodeWalk позволяет разработчикам наблюдать реальные проявления фундаментальных концепций компьютерных наук, таких как FIFO‑очереди, и автоматически сопоставлять их с теоретическими моделями, повышая понимание и уменьшая время, проведённое за экраном.

Задача

Software developers spend thousands of hours in front of screens, missing real‑world analogues of core computer science concepts.

Результат

A 10‑minute field test at a coffee shop demonstrated that CodeWalk successfully identified a FIFO queue, mapped customers to buffer elements, the cashier to a single consumer thread, and a complex order to head‑of‑line blocking, achieving a 95% confidence strong match.

Рабочий процесс

  • Choose a concept and time
  • Receive a physical mission from a local Gemma model
  • Lock phone and step outside
  • Observe the target phenomenon for the set time
  • Return and record observations
  • Gemma analyzes the notes, maps real‑world actors to CS concepts, and suggests the next walk

Доказательства

CodeWalk is an open-source, local-first companion application that challenges developers to close their laptops, put their smartphones away, and step into the physical world to observe foundational computer science concepts in action

Источник

The app instructed me: "Walk to a nearby coffee shop counter or bus stop. Stand at a respectful distance, put your phone away, and observe how people arrive and leave for 10 minutes. Notice: Where do new people join? What happens when an order takes extra time?"

Источник

Customers joined strictly at the rear (enqueueing at the tail). The cashier processed one transaction at a time from the front (dequeueing from the head). When one customer placed an elaborate order with five custom drinks, the line behind them swelled from 2 to 9 people—a vivid physical manifestation of Head-of-Line (HoL) Blocking and single-threaded backpressure!

Источник

Verdict: STRONG_MATCH (95% Confidence)

Источник

Real-World Mapping: Customers in line → Buffer / Queue elements; Cashier register → Single Consumer thread; Complex 5-drink order → High-latency task causing Head-of-Line blocking

Источник

  • Дата записи: 2026-10-11T19:41:18.817Z