Garrick Aden-Buie gadenbuie

Verification-Driven Development (VDD)

Methodology: Iterative Adversarial Refinement

Overview

Verification-Driven Development (VDD) is a high-integrity software engineering framework designed to eliminate "code slop" and logic gaps through a generative adversarial loop. Unlike traditional development cycles that rely on passive code reviews, VDD utilizes a specialized multi-model orchestration where a Builder AI and an Adversarial AI are placed in a high-friction feedback loop, mediated by a human developer and a granular tracking system.

I. The VDD Toolchain

Experimental function to concatenate lots of HTML dependencies' JavaScript files into one single one.

A number of caveats apply:

Strict mode: Many JavaScript files contain "use strict"; at the top of the file. This causes slightly different behavior in the JS engine. If the concatenated JS files have different expectations regarding strict mode, there could be problems.
Hosted scripts: This script does not attempt to concatenate JS files that are hosted on CDNs. Instead, they're left in their original form. It's possible that files we do concatenate, either assume that a CDN-hosted script is loaded before it itself is, or, the opposite. combine_js has a parameter that lets you decide whether to put the combined script first or last, there's no way for us to decide automatically (assuming there's even a right answer).
ES6 modules: We leave type="module" alone, because for ES6 modules the file boundarie

	// Moved to a github repo for versioning, the file can now be loaded from a CDN
	// <script type="module" src="https://esm.sh/gh/loueed/[email protected]/comments"></script>

	class BskyComments extends HTMLElement {
	constructor() {
	super();
	this.attachShadow({ mode: "open" });
	this.visibleCount = 3;
	this.thread = null;
	this.error = null;

	# Instructions:
	# - Download LM Studio
	# - Download Phi-3 Model (within LM Studio)
	# - Load the model into LM Studio
	# - Start the Local Server (instructions here: https://lmstudio.ai/docs/local-server)

	# Load necessary libraries
	library(httr)
	library(jsonlite)
	library(tidyverse)

	# Clone llama.cpp
	git clone https://github.com/ggerganov/llama.cpp.git
	cd llama.cpp

	# Build it
	make clean
	LLAMA_METAL=1 make

	# Download model
	export MODEL=llama-2-13b-chat.ggmlv3.q4_0.bin

	// I'm tired of extensions that automatically:
	// - show welcome pages / walkthroughs
	// - show release notes
	// - send telemetry
	// - recommend things
	//
	// This disables all of that stuff.
	// If you have more config, leave a comment so I can add it!!

	{

	# You'll also need to add roxygen2 to Suggests in the DESCRIPTION file to satisfy checks
	# or just call `usethis::use_package("roxygen2", "Suggests")`

	# first we define the tag used in examples in the doc ----------------------------------------------

	#' @export
	#' @importFrom roxygen2 roxy_tag_parse
	roxy_tag_parse.roxy_tag_tip <- function(x) {
	roxygen2::tag_markdown(x)
	}

	#!/bin/bash -e

	QUERY='{
	search(type: ISSUE, query: "is:pr state:closed repo:clio/grow", first: 100) {
	nodes {
	... on PullRequest {
	title
	number
	additions
	deletions

	# script to upgrade R and re-install all (or at least most) packages

	# Pre-upgrade: save your packages as an rds file
	my_pkg <- pak::pkg_list()

	# if loaded from RStudio
	#here::here("R", "update_packages", "my_packages.rds")
	# if loaded from finder
	saveRDS(object = my_pkg, file = "my_packages.rds")

	cran_days_since_release <- function(pkg, now = Sys.time()) {
	db <- rbind(tools:::CRAN_archive_db()[[pkg]], tools:::CRAN_current_db()[pkg,])
	now - sort(db$mtime, decreasing = TRUE)
	}

	# The CRAN rule is maximum 7 existing releases within the last 180 days
	# https://github.com/wch/r-source/blob/tags/R-4-1-2/src/library/tools/R/QC.R#L7990-L8007
	cran_wait_needed <- function(pkg, max = 7L, limit = 180, now = Sys.time()) {
	since <- cran_days_since_release(pkg, now = now)
	too_new <- since[since <= limit]