Francesco Oliva olibartfast

C++ Ranges and the Evolution of Modern Algorithms

A concise reference on how C++ algorithms evolved from raw loops to composable, lazy pipelines — and why ranges represent the most significant leap in expressive power since the STL itself.

The Timeline: From Loops to Pipelines

Pre-C++11: The Dark Ages

std::span in Modern C++

A comprehensive guide to std::span introduced in C++20.

What is std::span?

std::span is a non-owning view over a contiguous sequence of objects. It provides a safe, lightweight way to reference arrays or array-like data structures without taking ownership.

Key Characteristics

🧰 Generic Useful `kubectl` Commands (with `--kubeconfig` & port-forwarding)

A compact set of everyday-use commands for inspecting deployments, pods, logs, services, events, configmaps, and port-forwarding. Replace /path/to/kubeconfig.yaml with your actual kubeconfig file.

📦 Deployments

Get basic info

Ninja Build System Tutorial for C++ Projects

What is Ninja?

Ninja is a small, fast build system designed to have its input files generated by a higher-level build system (like CMake). It's optimized for speed and is particularly useful for large C++ projects.

Basic Setup

1. Generate Ninja files with CMake:

# Create build directory

Parallel Programming Cornerstones in CUDA

This gist outlines the core parallel programming patterns (cornerstones) used in CUDA to leverage modern NVIDIA GPU architectures. These patterns are fundamental for exploiting CUDA’s Single Instruction, Multiple Thread (SIMT) execution model, memory hierarchy, and advanced thread management features for efficient parallel computation.

A key theme in modern CUDA development is the use of highly optimized libraries. For many common patterns, NVIDIA CUB provides state-of-the-art implementations. While the original standalone CUB repository is now archived, CUB is actively maintained as part of the CUDA C++ Core Libraries (CCCL). CCCL unifies CUB, Thrust, and libcu++ into a cohesive standard library included in every CUDA Toolkit. Using libraries like CUB is the recommended best practice for achieving maximum performance and reliability.

The cornerstones covered include:

[Reduction](#1-red

	#!/bin/bash

	# Script to run YOLO predictions with two models and compose results side-by-side

	# Check if video input is provided
	if [ "$#" -lt 1 ]; then
	echo "Usage: $0 <input_video>"
	echo "Example: $0 /path/to/video.mp4"
	exit 1
	fi

	#!/bin/bash

	###############################################################################
	# extract_clip.sh
	#
	# 🎬 Extract a segment of a video from a given start time and duration.
	#
	# 📥 Usage:
	# ./extract_clip.sh input.mp4 start_time duration output.mp4 [--accurate]
	#

	#!/bin/bash

	######################################################################
	# blackout_roi.sh
	#
	# 🔳 Blackout a Region of Interest (ROI) in a video for a given time window.
	#
	# 🎯 Usage:
	# ./blackout_roi.sh input.mp4 ROI t1 duration output.mp4 [x y w h]
	#

	#!/bin/bash

	# Check if at least two inputs are provided
	if [ "$#" -lt 2 ]; then
	echo "Usage: $0 <video1> <video2> [output_file]"
	exit 1
	fi

	# Assign inputs to variables
	video1="$1"

	#!/bin/bash
	# NAS Share Mounter - Mount CIFS/SMB shares with enhanced features
	# Prerequisites: sudo apt-get install cifs-utils

	set -euo pipefail # Exit on error, undefined variables, pipe failures

	# ======= CONFIGURATION =======
	SCRIPT_NAME=$(basename "$0")
	CONFIG_FILE="$HOME/.nas_mounter.conf"
	LOG_FILE="/tmp/nas_mounter_$(date +%Y%m%d).log"