Logo
    Search

    675: Potluck × Bun Thoughts × Guesting on Syntax × Why Rust?

    enOctober 04, 2023

    Podcast Summary

    • Listeners share feedback on favorite episode formats and upcoming Toronto meetupPodcast solicits listener feedback, announces free meetup, and discusses new JavaScript runtime Bun and its controversy

      Syntax, a web development podcast, is seeking feedback from listeners through a survey about their favorite episode formats and other preferences. They also announced a free meetup in Toronto on October 10th at the Mascot Brewery. Wes shared an experience of encountering an issue with Get Display Media on his Mac, emphasizing the importance of having visibility into errors when pushing apps to production. The discussion then revolved around Bun, a new JavaScript runtime built on Zig, which has recently gained popularity and controversy. Bun offers support for Node modules, NPM, TypeScript, and JSX, and claims to be much faster than Node.js. The hosts expressed their support for having multiple JavaScript runtimes, emphasizing the importance of coding for JavaScript standards rather than specific runtimes.

    • New JavaScript runtime Bun aims for interoperability with CommonJS and ESMBun, a new JavaScript runtime, enables developers to use CommonJS and ESM systems together, simplifying upgrades and transitions, and offers flexibility to write and deploy code where it's most suitable.

      Bun, a new JavaScript runtime, is making strides towards interoperability with both CommonJS and ESM (ECMAScript Module) systems, allowing developers to use both side by side without issues. This could make upgrades and transitions between different JavaScript systems easier, as developers won't have to refactor all their files at once. Additionally, the ability to write JavaScript code and deploy it wherever it's most appropriate for the application is a significant advantage in the modern development landscape. However, it's important to note that Bun and Deno, another new runtime, are both for-profit companies, and their business models could potentially introduce new challenges. The speaker also mentions that they've been able to build a significant portion of their new syntax site using only fetch, a modern method for downloading data, instead of relying on Node.js-specific features. Overall, the interoperability and flexibility offered by these new runtimes and systems could lead to easier development and deployment processes. However, it's crucial to consider the potential business implications and challenges that may come with them.

    • Node.js advances and service workers for secure image requestsNode.js improves with new features, but for secure image requests in HTML content loaded from APIs, service workers provide the best solution by intercepting and modifying requests at a network level.

      Node.js is continuously improving with new features such as dotenv support, test runner enhancements, and even experimenting with a permissions model. These additions make development easier and faster for developers. However, when it comes to securing image requests for HTML content loaded directly from an API into the browser, a service worker is currently the best approach. It allows intercepting and modifying requests at a network level, adding custom headers before they reach the server. Although it requires the service worker to be installed first, it's an effective solution for this specific scenario. Other methods like Cloudflare Worker could also be considered if access to the image server is available. Overall, Node.js's progress and the use of service workers for specific security requirements demonstrate the evolving landscape of web development.

    • Protecting Images and Building APIsUse cookies for convenient image protection, service workers for offline access, or query parameters and tokens for server-side validation. Choose between REST, GraphQL, and tRPC APIs based on use case and requirements.

      When it comes to serving and protecting images on a website, there are several methods to consider. The speaker suggests using cookies as the most convenient and efficient way, as they are automatically sent with every request to the server. However, if access to the server is not available, using a service worker is the next best option. Another method mentioned is using query parameters and tokens on image URLs, which is what Amazon S3 does to protect assets. This method requires the server to check for the validity of the token before serving the image. When it comes to APIs, the speaker discusses the differences between REST, GraphQL, and tRPC. REST APIs involve hitting a URL and receiving data in response, while GraphQL involves sending a request with a specific query string in the body to get the desired data. tRPC is a library that provides a more type-safe and efficient way of building APIs compared to REST and GraphQL. The choice between these methods depends on the specific use case and requirements. For example, REST APIs are simpler and more widely used, while GraphQL offers more flexibility and the ability to retrieve multiple related pieces of data with a single request. tRPC, on the other hand, provides a more type-safe and efficient way of building APIs, making it a good choice for complex applications.

    • Comparing GraphQL, REST, and tRPC for API BuildingChoose GraphQL for type safety and control over data requests, REST for simplicity, or tRPC for ease of use and automatic type safety, depending on project needs and resources.

      GraphQL, REST, and tRPC each offer unique benefits when it comes to building APIs. REST, which doesn't require a library, is the simplest option, but it doesn't provide type safety or an ecosystem for working with APIs. GraphQL, as a spec, requires libraries on both ends and offers the ability to request specific data, but it requires more setup and can be more complex. tRPC, a library for making API calls, provides automatic type safety and a seamless developer experience, but it requires more setup and has less adoption than GraphQL. Ultimately, the choice between these options depends on the specific needs and resources of your project. If you prioritize simplicity and don't need type safety, REST might be the best choice. If you need more control over the data you're requesting and value the benefits of a typed API, GraphQL could be the way to go. And if you want the ease of use and automatic type safety that tRPC provides, you'll need to invest in setting up the necessary infrastructure.

    • REST vs GraphQL APIs: Choosing the Right OneREST APIs are reliable and widely used, while GraphQL APIs unify data sources. Choose based on project needs and scale.

      REST and GraphQL APIs each have their unique strengths and applications. REST APIs are a reliable and widely used option, especially for teams dealing with multiple types of clients accessing the same server. GraphQL APIs, on the other hand, are increasingly popular in larger companies due to their ability to unify various data sources into a single endpoint. While both have their benefits, the choice between them depends on the specific needs and scale of the project. Additionally, the podcast hosts welcome guests with interesting stories or advice to share, not just product promotions.

    • Focusing on valuable guests in podcastsAuthenticity and value are key in podcast guests. Look for individuals who bring unique insights and avoid repetitive conversations with self-promoters. In documenting ecommerce startups, using TypeScript can help generate API docs, but documenting business logic and integrations remains a challenge.

      When it comes to hosting a podcast, it's important to focus on bringing on guests who have valuable insights to share, rather than just aiming for high-profile names to attract larger audiences. The podcast host expressed that they look for guests based on a "feel" and the value they can bring to the listeners. This approach helps to avoid the trap of having repetitive conversations with guests who are only on to promote themselves or their latest project. Additionally, documenting complex code, business logic, and integrations in an ecommerce startup can be a challenge when there is no initial documentation. The host suggested that using TypeScript can help generate API docs. However, they didn't specify an exact solution for documenting business logic and integrations. Overall, the conversation highlighted the importance of authenticity and value in podcast guests and the challenges of documenting intricate code in a growing startup.

    • Documentation doesn't have to be complicatedEffective documentation includes code comments, JSDoc, Storybook, simple guides, and clear configuration instructions to make development smoother

      Effective documentation is crucial for complex tech projects, but it doesn't have to be overly complicated or time-consuming. Comments in the codebase can go a long way in explaining what each part does and why it exists. For more in-depth documentation, tools like JSDoc and Storybook can be used to generate documentation and document UI components, respectively. At the most basic level, having simple guides or videos on common tasks can help save time and frustration. Additionally, keeping a sample file for configuration files and maintaining clear documentation on how to publish new versions can make the development process smoother. Overall, the key is to find the right balance between the level of documentation needed and the time and resources available.

    • Documenting processes for better understanding and efficiencyDocument your processes to improve understanding, save time, and reduce frustration. Learn the basics of new languages before diving deep, and consider exploring new programming paradigms for a fresh perspective.

      Documenting what you do is an essential practice for improving your own understanding and making future tasks easier. The speaker shared how he documents his own process for publishing a new version of his Versus code theme in Rust, which has helped him save time and reduce frustration. This concept applies to any programming language or project. Another key point from the discussion was the importance of understanding the basics of a new programming language before delving into more nuanced topics. The speakers emphasized that when learning a new language, it's essential to keep things simple and surface-level at first, rather than getting bogged down in too many details. Regarding the question of why someone should learn Rust instead of JavaScript, the speakers noted that Rust offers a different programming paradigm and level of typing that can provide a new perspective for developers. However, it's not necessary for everyone to learn Rust, especially if they are web developers. Ultimately, the decision to learn a new language depends on individual interests and goals. Overall, the discussion highlighted the importance of documentation, the value of learning the basics of new technologies, and the benefits of exploring new programming paradigms.

    • Discovering the Power of Rust for Programming and ProductivityRust offers unique advantages for lower-level programming and performance optimization. Consider learning it for building applications, exploring non-web projects, or creating low-level utilities. Consistently showing up to challenges and using tools that improve your life can lead to significant progress.

      While JavaScript is great for web development, Rust offers unique advantages for lower-level programming and performance optimization. The speaker shares his personal experience of discovering a newfound appreciation for Rust after trying it out for productivity and accomplishing tasks beyond JavaScript's capabilities. He also suggests reasons to consider learning Rust, such as building applications, exploring non-web projects, or creating low-level utilities. Additionally, the speaker shares his dedication to using home gym equipment, specifically the Tonal, and encourages listeners to stick with things that improve their lives. He emphasizes the importance of showing up consistently and seeing the big results that come from it. Regarding a technical question, a listener encountered an issue using fetch instead of Axios for making requests to a custom WooCommerce REST API. The speaker suggests that fetch may not assume JSON or even that a response is coming back, so it's essential to ensure proper JSON stringification and handling of potential redirects. The listener is advised to log the request on the API side and debug the fetch request to identify the issue.

    • Comparing Fetch and Axios API callsWhen using Fetch and Axios for API calls, be aware of discrepancies in error handling and request/response formats. Use tools like Proxy Man or Chrome DevTools to compare requests and responses for a better understanding.

      When dealing with API calls using different libraries like Fetch and Axios, it's essential to understand their differences and be aware of potential discrepancies in request and response handling. For instance, Fetch doesn't throw errors on 404 responses, while Axios does. To identify such differences, use tools like Proxy Man or Chrome DevTools to examine the requests and responses side by side. These issues can be frustrating when expecting a specific result, but with the right tools and understanding, they can be resolved. Additionally, the discussion touched upon the importance of being open-minded to various forms of media, even if one may not initially seem appealing. For example, the show "Glow Up" is a makeup competition reality show, which may not be everyone's cup of tea. However, its unique charm and outrageous elements can make for an enjoyable viewing experience. Lastly, a recommendation was made for the website or service called Send Cut Send. It's a useful tool for sending large files, especially when email size limitations are a concern. It offers a simple and efficient solution for sharing large files with others.

    • Design and fabricate custom parts using Send Cut SendIndividuals can create custom hardware parts using Send Cut Send by uploading SVG or EPS files, choosing materials like HDPE, and receiving cost-effective, shipped parts with logo options

      Send Cut Send is a website that makes it possible for individuals, including web developers, to design and have custom parts fabricated from various materials such as metal, PVC, ABS, wood, and more. The user doesn't need to know 3D modeling as they can upload an SVG or EPS file designed using tools like Figma. The company offers a range of materials, including HDPE, and can create parts from thin materials that may not cut well or overheat. The process is cost-effective, and the parts can be shipped in Canada and the US. The user can even add their logo to the parts using laser cutting. It's like a 3D printing or T-shirt printing service but for hardware. The advantage is that one doesn't need to mass-produce these parts, making it a convenient solution for creating custom parts for personal use. Send Cut Send offers materials such as cork, copper, stainless, and acrylic. The company is helpful in suggesting modifications to the design if needed. Overall, Send Cut Send makes the world of custom metal parts accessible to individuals, making it an excellent resource for creating unique parts for various projects.

    Recent Episodes from Syntax - Tasty Web Development Treats

    790: State of JS 2023 Reactions

    790: State of JS 2023 Reactions

    Scott and Wes dive into the 2023 State of JavaScript survey, breaking down the latest trends and pain points in front-end frameworks, build tools, and JavaScript runtimes. Tune in for their hot takes and insights on what’s shaping the JavaScript landscape this year!

    Show Notes

    Sick Picks

    Shameless Plugs

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott: X Instagram Tiktok LinkedIn Threads

    Randy: X Instagram YouTube Threads

    789: Do More With AI - LLMs With Big Token Counts

    789: Do More With AI - LLMs With Big Token Counts

    Join Scott and CJ as they dive into the fascinating world of AI, exploring topics from LLM token sizes and context windows to understanding input length. They discuss practical use cases and share insights on how web developers can leverage larger token counts to maximize the potential of AI and LLMs.

    Show Notes

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott: X Instagram Tiktok LinkedIn Threads

    CJ: X Instagram YouTube TwitchTV

    Randy: X Instagram YouTube Threads

    788: Supabase: Open Source Firebase for Fullstack JS Apps

    788: Supabase: Open Source Firebase for Fullstack JS Apps

    Scott and CJ chat with Paul Copplestone, CEO and co-founder of Supabase, about the journey of building an open source alternative to Firebase. Learn about the tech stack, the story behind their excellent documentation, and how Supabase balances business goals with open-source values.

    Show Notes

    • 00:00 Welcome to Syntax!
    • 00:30 Who is Paul Copplestone?
    • 01:17 Why ‘Supa’ and not ‘Super’?
    • 02:26 How did Supabase start?
    • 08:42 Simplicity in design.
    • 10:32 How do you take Supabase one step beyond the competition?
    • 12:35 How do you decide which libraries are officially supported vs community maintained?
      • 15:17 You don’t need a client library!
    • 16:48 Edge functions for server-side functionality.
    • 18:51 The genesis of pgvector.
    • 20:59 The product strategy.
    • 22:25 What’s the story behind Supabase’s awesome docs?
    • 25:26 The tech behind Supabase.
    • 35:46 How do you balance business goals with open source?
    • 42:01 What’s next for Supabase?
    • 44:15 Supabase’s GA + new features.
    • 48:24 Who runs the X account?
    • 50:39 Sick Picks + Shameless Plugs.

    Sick Picks

    Shameless Plugs

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott: X Instagram Tiktok LinkedIn Threads

    CJ: X Instagram YouTube TwitchTV

    Randy: X Instagram YouTube Threads

    787: You Should Try Vue.js

    787: You Should Try Vue.js

    Scott and CJ dive deep into the world of Vue.js, exploring what makes this frontend framework unique and why it stands out from React and Svelte. CJ gives a comprehensive tour, covering everything from getting started to advanced features like state management and Vue’s built-in styles.

    Show Notes

    Vue.js: The Documentary.

    Sick Picks

    Shameless Plugs

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott: X Instagram Tiktok LinkedIn Threads

    Randy: X Instagram YouTube Threads

    786: What Open Source license should you use?

    786: What Open Source license should you use?

    Scott and CJ dive into the world of open source, breaking down its meaning, benefits, and the various types of licenses you’ll encounter. From permissive licenses like MIT and Apache 2.0 to copy-left licenses such as GNU GPLv3, they’ll help you choose and apply the right license for your project.

    Show Notes

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott: X Instagram Tiktok LinkedIn Threads

    Randy: X Instagram YouTube Threads

    785: What’s Next for NextJS with Tim Neutkens

    785: What’s Next for NextJS with Tim Neutkens

    Scott and Wes dive into the world of Next.js with special guest Tim Neutkens from Vercel. They explore the latest updates, including the React Compiler and React Server Components, discussing their impact on developer workflows and the future of Next.js development.

    Show Notes

    • 00:00 Welcome to Syntax!
    • 00:30 What does the React Compiler do?
    • 05:04 Will React Compiler help with managing Context?
    • 06:39 What happens if you’re not using a React Compiler?
    • 09:30 Will this work on any NextJS version?
    • 12:18 What are React Server Components?
    • 16:28 Shipping all the data inside an encapsulated component.
    • 20:17 Clearing up the frustrations around retrofitting server components.
    • 23:13 Handing migration.
    • 28:30 Is this just a fetch request with props?
    • 36:41 How closely are the NextJS and React teams working?
    • 41:53 Will we ever get Async Client Components?
    • 43:52 Async Local Storage API.
    • 45:31 Turbopack.
    • 57:51 Sick Picks & Shameless Plugs.

    Sick Picks

    Shameless Plugs

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott: X Instagram Tiktok LinkedIn Threads

    Randy: X Instagram YouTube Threads

    784: Logging × Blogging × Testing × Freelancing

    784: Logging × Blogging × Testing × Freelancing

    In this Potluck episode, Scott and Wes tackle listener questions on modern blogging, website environmental impact, and using LangChain with LLMs. They also cover CSS hyphens, unit vs. integration testing, and balancing web development with new parenthood.

    Show Notes

    Sick Picks

    Shameless Plugs

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott: X Instagram Tiktok LinkedIn Threads

    Randy: X Instagram YouTube Threads

    783: How We Built a Netflix Style “Save for Offline” Feature Into Syntax

    783: How We Built a Netflix Style “Save for Offline” Feature Into Syntax

    Scott and Wes dive into the world of browser caching for audio files, exploring the File System API and the Cache API. They discuss size restrictions across different browsers, how tools like Riverside.fm leverage IndexedDB, and walk through code examples for creating, retrieving, and managing cached audio data.

    Show Notes

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott:X Instagram Tiktok LinkedIn Threads

    Randy: X Instagram YouTube Threads

    782: The Developer’s Guide To Fonts with Stephen Nixon

    782: The Developer’s Guide To Fonts with Stephen Nixon

    Scott and CJ are joined by Stephen Nixon of ArrowType to delve into the world of fonts and type for developers. They explore the intricacies of font creation, the utility of variable fonts, and offer tips for making visually appealing typography on the web.

    Show Notes

    Sick Picks

    Shameless Plugs

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott:X Instagram Tiktok LinkedIn Threads

    Randy: X Instagram YouTube Threads

    781: Potluck - The Value of TypeScript × Vue vs Svelte × Leetcode

    781: Potluck - The Value of TypeScript × Vue vs Svelte × Leetcode

    In this potluck episode of Syntax, Scott and CJ serve up a variety of community questions, from the nuances of beginner vs. advanced TypeScript to the pros and cons of SvelteKit. They also discuss falling out of love with React, shipping private packages via NPM, and the eternal struggle of always starting but never finishing projects.

    Show Notes

    Sick Picks

    Shameless Plugs

    Hit us up on Socials!

    Syntax: X Instagram Tiktok LinkedIn Threads

    Wes: X Instagram Tiktok LinkedIn Threads

    Scott:X Instagram Tiktok LinkedIn Threads

    Randy: X Instagram YouTube Threads

    Related Episodes

    Rust for JS Devs — Part 2

    Rust for JS Devs — Part 2

    In this episode of Syntax, Wes and Scott jump into part 2 of their look at Rust for JavaScript developers, including variables in Rust, type systems in Rust, signed and unsigned integers, and more.

    Show Notes

    let x = 5;  // x is immutable let mut x = 5; // x is mutable const MAX_POINTS: u32 = 100_000; // must be defined at compile time 
    • 10:42 Type System in Rust
    • 15:52 Types in Rust
    • 19:06 Why does Rust have signed and unsigned integers?
    • 23:35 Slicing strings with &str
    • 27:35 enum
    • 27:55 struct
    • 28:19 Vec
    • 29:33 HashMap and HashSet
    • 33:00 Converting Signed to Unsigned Numbers
    let unsigned_value: u8 = 200; let signed_value: i8 = unsigned_value as i8; 
    • 36:12 What’s up with &str?
    • 43:31 Rust error messages
    • 45:28 What is a Struct?
    struct User {    username: String,     email: String,     sign_in_count: u64,     active: bool, }  // You can create an instance of a struct like this: let user1 = User {     email: String::from("someone@example.com"),     username: String::from("someusername123"),     active: true,     sign_in_count: 1, };  impl User {     fn login(&mut self) {         self.login_count += 1;     } } 
    • 49:17 SIIIIICK ××× PIIIICKS ×××

    ××× SIIIIICK ××× PIIIICKS ×××

    Shameless Plugs

    Tweet us your tasty treats

    Dokumentacja w projekcie - Software Engineering at Google

    Dokumentacja w projekcie - Software Engineering at Google
    Wracamy z kolejnym odcinkiem, gdzie analizujemy temat z książki Software Engineering at Google. Tym razem bierzemy na tapet dokumentację w projekcie.

    Dyskutujemy:

    - czy rzeczywiście dokumentacja to pierwsza potrzeba w projekcie?
    - kiedy potrzebujemy dokumentacji?
    - kto powinien być odpowiedzialny za jej utrzymywanie?
    - jakie projekty szczególnie potrzebują dokumentacji?
    - jakie korzyści ma tworzenie dokumentacji?
    - czy możemy zmierzyć ROI z dokmentacji?

    Na te i wiele innych pytań staramy się odpowiedzieć w tym odcinku. Zapraszam do wysłuchania rozmowy na Spotify, Apple Podcasts, Google Podcasts i https://poprostujs.pl

    Rust for JavaScript Developers - Node vs Rust Concepts

    Rust for JavaScript Developers - Node vs Rust Concepts

    In this Hasty Treat, Scott and Wes talk about Rust, and how the concepts might translate for JavaScript developers looking to try writing in Rust.

    Show Notes

    Tweet us your tasty treats

    L'IA aura-t-elle la peau des développeurs ?

    L'IA aura-t-elle la peau des développeurs ?

    Bonjour à tous et bienvenue sur le ZD Tech, le podcast quotidien de la rédaction de ZDNet. Je m'appelle Pierre Benhamou et aujourd'hui, je vais vous parler d'un outil de Google qui pourrait bien remplacer un jour les développeurs et faire entrer la programmation dans une nouvelle ère.

    Nombreux sont aujourd'hui les étudiants à s'orienter vers des filières d'informatique, alléchés à la fois par un domaine passionnant, mais aussi rémunérateur.


    Les entreprises s'en frottent les mains, alors que les métiers du développement informatique font actuellement l'objet d'une demande croissante bien supérieure à l'offre disponible sur le marché du travail.


    Confrontés à une pénurie chronique de talents, les géants de la tech n'ont pour leur part pas attendu pour s'attaquer au problème en mettant sur pieds des modèles d'intelligence artificielle qui pourraient bien un jour assurer à elles seules les tâches qui incombent aujourd'hui aux développeurs.

    Des chercheurs de DeepMind, l'unité de Google spécialisée dans l'IA, viennent ainsi de jeter une nouvelle pierre dans le jardin de ces derniers en affirmant qu'Alphacode, leur système de programmation assisté par l'IA, s'est amélioré tant et si bien qu'il est désormais capable d'obtenir un score médian aux différents concours de programmation organisés pour les développeurs en herbe.


    C'est bien simple, le système de DeepMind s'est ainsi classé parmi les 54,3 % de participants les mieux placés dans 10 concours comptant plus de 5 000 participants organisés récemment.


    Dans ces concours, chaque participant doit analyser la description d'une énigme et développer rapidement un programme destiné à la résoudre. Confronté à ses concurrents humains, Alphacode a dû créer un algorithme puis le mettre en œuvre de manière efficace en optant pour des langages divers et variés afin de surmonter les difficultés. Rien de plus facile pour Alphacode, dont les connaissances s'étalent sur pas moins de 715 Go de code provenant de fichiers issus de dépôts GitHub écrits en langage C++, JavaScript, Python, PHP, Ruby, Rust, Scala, et j'en passe. 


    Rien de plus facile pour Alphacode, dont les connaissances s'étalent sur pas moins de 715 Go de code provenant de fichiers issus de dépôts GitHub écrits en langage C++, JavaScript, Python, PHP, Ruby, Rust, Scala et j'en passe.

    Se pose alors la question de l'avenir des développeurs, dont l'activité pourrait un jour être automatisée par Alphacode et d'autres systèmes d'IA.


    Ces derniers auront-ils la peau des développeurs ? Et bien, rassurez-vous, ce n'est pas encore prêt d'arriver, car la nouvelle percée de DeepMind n'est pas sans défauts. Le modèle même d'Alphacode, qui se base sur des programmes existants pour s'entraîner, fait que ce dernier n'est pas sans générer du code comprenant des faiblesses exploitables. On s'en doute, en se basant sur des programmes écrits par des humains pour s'améliorer, Alphacode va logiquement reproduire les biais et les failles de ses professeurs.


    Reste également un problème pour l'heure insoluble à résoudre pour faire d'Alphacode et consorts des concurrents de taille aux développeurs humains. Il s'agit bien évidemment de l'impact environnemental de tels systèmes, dont l'entraînement a nécessité, dans le cas du système de DeepMind, des « centaines de jours pétaFLOPS » dans les centres de données de Google. De quoi rassurer pour le moment les développeurs en herbe et leurs aînés ! 

    Quel langage de programmation pour quel usage ?

    Quel langage de programmation pour quel usage ?

    Dans cette 7ème édition de nos Twitter Spaces, on va parler langages de programmation !

    Les langages de programmation sont nombreux, et ne sont pas tous adaptés à toutes les applications. Au cours de ce space nous répondrons à plusieurs questions : saviez-vous que chez Les-Tilleuls.coop nous faisons majoritairement du Javascript? Python est-il vraiment plus performant que PHP ? Pourra-t-on un jour réconciler les amateurs de Go et ceux de Rust ? Pourquoi choisir une techno plutôt qu'une autre ? Et la sobriété numérique dans tout ça ?



    Ils ont pris la parole :


    Cet épisode vous a plu ?

    N’hésitez pas à nous le dire sur Twitter @coopTilleuls!


    Musique :  Bonfire - An Jone