ggorlen

Here are my eng values (jotted down, WIP)

fast iteration loops, merge into main and deploy often. I'm a big believer in the quantity vs quality paradox.
get strong observability right away, should be able to query database, run code in prod console, access datadog, everything hooked up to LLMs
look for the key metrics to measure work impact, figure out what's highest leverage, and establish acceptance checks for changes
understand which are the key customers and what their needs and pain points are, understand as much biz context as possible/useful--who are our competitors, what are they doing?
don't tolerate broken/flaky CI, fix/improve it right away
don't tolerate broken/flaky anything in the product or dev ex, fix/improve it right away, if possible
take action if something can be improved. now with LLMs it's generally easy to just fix stuff ~instantly instead of going through a sprint planning committe

How I Use Computers

Motivation: I often notice when standing alongside someone else that their workflow of navigating a computer differs from mine. Often, I learn new tricks from them, while other times I want to tell them about some keystroke shortcut.

Here's some tips for how I use computers to canonize the process.

General

I usually avoid desktop apps in favor of browser apps
If I have too many browser tabs open I move them to a google sheet organizer I always have running in a tab. The sheet eventually overflows and I move stuff to large markdown files. probably not the best system although it's an easy enough way to be able to find things if I need them someday, which I often do.

Predawn

last couple chords change a bit, TODO tab out the rest.

-----------------------------------------------------------
-5---------5---------5-------------------------------------
-2-4-------2-4-------2-4---------------------------------2-
---2-5-------2-5-------2-5-----------------------------2---
-----3-5-------2-5-------3-5-------------------------0-----

Novice JavaScript: the Q&A App

Context

This article discusses strategies for teaching 100% novice programmers with JavaScript--the very first program and next steps after it. Primary audience is grade and high school students, but also adults.

Problem: the quiz app

For new programmers, particularly kids, Python is the default choice for good reason: it's perfect for simple interactive apps:

Automating Life

What have we done with technology?

cars: privatized/corporatized/automated walking, biking, public transit, horses
- outcome: walkable urban centers and close-knit towns become exurbs and strip malls, ubiquitous noise, pollution and car infra, all humans required to fork over hundreds of thousands of dollars per life just to get around the way they used to be able to do for free, super long commutes, obsesity, 40k deaths per year in the US, environmental damage, lead poisoning, now cars are massively reliant on computer chips and can't be repaired easily, etc. Enabler of classical big box store style consumption.
smart phones: the control center that runs everything else and enables most other automation, replaces wallets, credit cards, traditional phones, video cameras, photo cameras, alarm clocks, music players, transit cards, watches, books, TVs, calculators, and most other single-purpose consumer goods of the 20th century
outcome: impossible to survive without technological k

Post to Slack as yourself from the CLI (macOS + Chrome)

Slack's xoxc- session tokens require an HttpOnly d cookie that JavaScript can't read. But you don't need to extract it. By running fetch() inside Chrome's Slack tab via AppleScript, the browser attaches the cookie automatically.

No bots, no webhooks, no OAuth apps. Just your existing Slack session.

Requirements

macOS (uses osascript)
Chrome with a app.slack.com tab open

Example Software Design Document

Purpose

The app will be a full-stack, multi-user wiki application, basically a simplified Wikipedia.

Technologies

Front end: React with React Router and the marked library for rendering markdown
Back end:

	# clean my ubuntu machine https://chatgpt.com/c/69aa0695-42ac-8329-8688-3684685e14fd

	# Find the largest directories in your home folder
	du -h --max-depth=1 ~ 2>/dev/null \| sort -hr

	# Find large files over 500 MB
	find ~ -type f -size +500M -exec ls -lh {} \; 2>/dev/null

	# Clear rebuildable caches (packages and browser binaries may need redownloading)
	rm -rf ~/.cache/yarn ~/.cache/puppeteer ~/.cache/pip ~/.cache/ms-playwright

	<!DOCTYPE html>
	<html lang="en">
	<head>
	<meta charset="UTF-8" />
	<title>Greg Stamp</title>
	<style>
	body {
	display: flex;
	justify-content: center;
	align-items: center;

	#!/usr/bin/env python3

	# Git stats are misleading because some people squash merge and some people
	# create merge commits, bringing their history of commits from the branch onto
	# master and making it look like they did way more work relative to squashers.
	#
	# This tries to treat merge commits as if they were squash merged so repo stats
	# are actually normalized/comparable. It also merges by identical name and email.
	#
	# I'm sure this isn't 100% accurate, particularly for LOC, but anecdotally seems