January 2022 Summaries

There are three essential resources that every computer needs: processor (CPU), memory (RAM), and storage. Replit users' computers have access to a virtual CPU, RAM allocation, and a virtualized filesystem. The new resource utilization component helps users understand why their repl is running slowly or misbehaving by providing information on CPU and RAM usage. Modern computers run multiple programs at once, with the operating system allocating CPU time to user applications. CPU utilization measures the difference in time spent executing instructions over a given interval divided by wall time, taking into account available CPU cores. Replit users have access to 0.5 vCPUs for free plans, 2 vCPUs for Hacker repls, and 4 vCPUs for Boosted repls. RAM is used to store program values at runtime, with more programs requiring more RAM. Users have limited control over memory usage, but can test it by running a program that allocates large amounts of memory. The new resource monitoring component shows how much storage space is being used in the repl's filesystem, which is currently limited to 1 GB. Understanding these resources and their utilization can help users optimize their Replit experience.

We have rearchitected our mobile IDE for web, focusing on a simpler & faster render tree and abstractions called "services" to drive the state of the repl UI. We moved the source of truth to the backend, using services to reduce memory footprint and make synchronizing clients across devices easier. The new version aims to improve core mobile ergonomics and run repls faster globally. We're excited to bring you the next version soon and are currently hiring for impactful projects.

Rust is a highly favored programming language, noted for being both beloved by developers and among the highest paid, as revealed by a Stack Overflow survey. For those interested in learning Rust, the Replit platform provides an accessible and collaborative in-browser IDE that supports over fifty programming languages, including Rust, making it an ideal tool for both individual and group learning. This tutorial guides users through setting up a Rust development environment on Replit and building a simple application that includes writing and running Rust code, implementing a binary search algorithm, adding unit tests, incorporating user input, and managing dependencies with Cargo, Rust's package manager. Additionally, the tutorial demonstrates how to enhance the application with third-party packages for improved console output and utilize Replit’s built-in key-value database to store and retrieve data. By following these steps, learners can establish a foundational understanding of Rust and gain practical experience using it in a modern development environment.

For the past 6 months, we have been testing Teams Pro while in private beta, releasing it today to the world. ` Teams Pro is Replit for work, offering a collaborative repository for sharing and editing code among team members. Users are utilizing Teams Pro for various purposes such as hosting slack bots, conducting remote interviews, rapidly prototyping, pairing programming, accessing Linux packages and programming languages, sharing private repositories, and asynchronous conversations with their team. The goal of Teams Pro is to redefine collaborative programming at work and invite users to join the journey.

After many years of running on Heroku, Replit has fully migrated to Google Cloud Platform with the goal of achieving greater flexibility and control over its infrastructure to support its mission of bringing a billion software creators online. The migration involved four stages: demoing a prototype, migrating data from Postgres to self-hosted databases in Google Cloud, moving to Google Cloud SQL for Postgres administration offloading, and finally, moving the front-end to utilize the Cloudflare Load Balancer's controls to redirect requests based on percentage. The migration was a significant project requiring extensive planning and practice to ensure a seamless experience for users, with the team rehearsing failure modes multiple times and using tools like Redis Labs' Google Cloud marketplace product to validate their strategy.