Luke Miles qpwo

Minimal Unreal New Project Setup

Skip the Launcher and create a new project directly

What is the minimal number of required files and folders needed to create an Unreal Engine Project? Surprisingly little -- one, one-line file to be specific -- and you can launch without needing to use the Launcher or have unreal prompt for a Template project.

Maybe the interesting thing here is that this is an option for some kind of automation. It's literally just create a one-line file, "open" it, and (maybe) select your chosen Unreal version.

Given the number of dev and test files I create, this is a nice direct way to get up and going.

READ pallets/flask#1938

	class Agent:
	def __init__(self, name, ports=None):
	self.name = name
	self.ports = ports if ports else []
	# Each element of ports is either None or a tuple (other_agent, other_port_index)

	def __repr__(self):
	return f"{self.name}"

	def connect(a, pa, b, pb):



	class Rezero(layers.Layer):
	def __init__(self):
	super().__init__()
	self.alpha1 = tf.Variable(0.0, trainable=True)

	def call(self, inputs, training):
	return self.alpha1*inputs
	class CustomRezero(tf.keras.layers.Layer):

	function import-ts-repl() { (
	set -e
	file=$1
	rm -f ${file}.js ${file}.js.map
	esbuild ${file} --target=node14 --format=cjs --sourcemap --outfile=${file}.js --platform=node --bundle
	node -r ${file}.js
	rm -f ${file}.js ${file}.js.map
	); }

	// January 2022, Luke Harold Miles, public domain obviously
	// Adapted from: https://github1s.com/expressjs/express/blob/master/examples/auth/index.js
	// You can run with `ts-node express-typescript-auth.ts`
	// Source for this file to get updates: https://gist.github.com/qpwo/4b2ae3aaaf222d099b34898152118c43

	import { pbkdf2Sync, randomBytes, timingSafeEqual } from 'crypto'
	import type { NextFunction, Request, Response } from 'express'
	import express from 'express'
	import type { Session } from 'express-session'
	import session from 'express-session'

	"""
	A simple predator-prey multi-agent reinforcement learning environment in
	python without dependencies, and an implementation of an example q-learning agent.
	Python 3.9
	https://gist.github.com/qpwo/a19f43368afd77288bf3b7db81fdc18b/
	February, 2021 -- Public domain dedication
	"""

	from dataclasses import dataclass, field
	import random

	// Compile with gcc -std=c99 mersenne_twister.c -o /dev/random
	// ps the -o /dev/random part is a joke, it's not actually meant to be a random device.
	// Absolutely no warranty, Kopyleft I guess who gives a shit
	#include <stdint.h>
	#include <stdio.h>
	uint32_t x[624];
	uint32_t index = 0;
	void twist() {
	for (int k = 0; k < 624; k++) {
	x[k] = x[(k+397)%624]

	import math
	import scipy.stats as st

	# assumes bivariate normal, dichotomised groups
	def dichotomy_r_to_d(r) :
	d = 2r / (math.sqrt(1 - r*2))
	return d

	# Equation 9
	# https://sci-hub.tw/10.1037/1082-989X.11.4.386

	# Experiments in capturing combined output streams from subprocesses.
	#
	# It turns out it's kind of hard to get the interleaved results of stdout and
	# stderr in Python. However, in a lot of situations where Python is calling
	# out to other processes, you probably want to swallow the child process's
	# stderr when things to right and print it when things go wrong. Since some
	# programs may be outputting results on stdout and warnings and errors on
	# stderr, it makes sense that you'd want it all, and all in the order it was
	# printed when things go wrong.
	#