The stable release of Svelte 4 was released this week, according to the blog post announcing the release. This is primarily a maintenance release to prepare for Svelte 5. The latest release leverages several improvements in Node.js and browser APIs, the post says.
The post states that the new release will have smaller and faster hydration code. For example, kit.svelte.dev reduced the size of generated JS across the site by 12.7% (from 126.3 kB to 110.2 kB).
Svelte 4 reduces the size of the Svelte package by nearly 75%, down to 2.8 MB. The team noted that this reduces latency for npm installs.
“This improvement is especially noticeable for users loading the interactive learning experience on learn.svelte.dev for the first time, users of the Svelte REPL, and users with limited connectivity,” the post said. “Most of the remaining package size is eslint support, which requires distributing CJS builds. Once the eslint rewrite is complete, the Svelte package size could be reduced by another 50% or more.”
Additionally, the number of dependencies has been reduced from 61 to 16. This means not only will your downloads be faster, but your supply chain attacks will be less impacted.
The plan was to switch the underlying code from TypeScript to JavaScript (yes you read correctly, we triple checked). But that seems to have migrated to Svelte 5 since the post states: Good compiler and runtime. The post adds that Svelte 4 is primarily intended to lay the groundwork for future improvements by modernizing the tools and dropping support for some legacy technology versions such as older bundlers. I was.
Latest news from Deno
Users of Fresh, Deno’s new full-stack web framework, are clearly wondering if the Deno team is maintaining it. Fresh creator Luca Casonato announced this week that Marvin Hagemeister will succeed Casonato as full-time lead of the Fresh project. .
“…There’s an elephant in the room. Is Fresh something that the Deno team actually focuses on maintaining? When asked, we always said yes! The reality was more complicated,” Casonato wrote. “At the beginning of April, the Fresh repository had over 60 open (and unreviewed) pull requests for him. .”
Hagemeister is the maintainer of React, the builder of Preact DevTools, and a “top speeder in the JavaScript ecosystem,” which he says has helped speed up npm scripts from a 400ms overhead to 22ms. Kasonato added.
Casonato also details new features in Fresh 1.2, including:
- simplified test
- Passing Signals, Uint8Arrays, and Cyclic Data in Island Properties
- Pass JSX to islands and nest islands inside each other
- npm: limited support for specifiers
- Support for custom HEAD handlers
- Overriding HandlerContext.render status and headers
- subdirectories within the ./islands folder
- Asynchronous plugin rendering
Figma Enters Development Mode
Figma held its annual Config conference in San Francisco this week, and the focus was on bridging the gap between design and development. As part of the conference, Figma announced new products and features such as Dev Mode, a dedicated space for developers to work with designers to turn designs into code.
“Development mode allows designers and developers to work in different modes on the same design in the same file, making it easier for developers to get the information they need, eliminating the need for version updates and extra communication. will be gone,” the announcement said.
Image via Figma
Development mode includes the following features:
- Navigate files in development mode to understand designs. Developers can hover and click objects to get measurements, specs, styles, and assets.
- Generate production-ready CSS, iOS, and Android code snippets.
- Connect to your tools and codebases with plugins like Jira, GitHub, Storybook and more to automate your workflows.
- Synchronize sources of truth by linking your Figma design system to your component code and documentation.
- With Figma for VS Code built in, you can inspect Figma files, collaborate with designers, and receive Figma notifications without leaving your code editor.
- Use labels to track design status, organize development-ready design files, and compare changes.
Development mode is currently in open beta and free to try until 2023.
AWS Launches New Generative AI Innovation Center
Amazon Web Services on Thursday announced a new program to help customers build and deploy generative artificial intelligence (AI) solutions. AWS has invested him $100 million in a program called the Generative AI Innovation Center. This team includes strategists, data scientists, engineers, and solution architects who help customers interested in building AI solutions. AWS also offers free workshops, engagements, and training to help customers brainstorm ideas and explore business use cases.
“Customers work closely with generative AI experts on AWS and the AWS Partner Network to select appropriate models, define pathways to solve technical or business challenges, and develop proofs of concept. , so you can plan to launch your solution at scale,” the press release states. . “The Generative AI Innovation Center team provides guidance on best practices for responsibly applying generative AI, optimizing machine learning operations and reducing costs.”
Engagements also leverage various AWS AI tools, such as Amazon’s code prediction tool CodeWhisperer. Amazon’s high-performance infrastructure. Amazon SageMaker for model building, training, and deployment. Amazon Bedrock is a fully managed service that provides API access to underlying models from AI221 Labs, Anthropic, Stability AI, and Amazon Titan. You can also use SageMaker Jumpstart to deploy popular base models such as Cohere’s Large Language Model, Technology Innovation Institute’s Falcon 40B, and Hugging Face’s BLOOM.
Free LLM Course for Developers
Vector database company ActiveLoop, in partnership with Intel and AI learning community TowardsAI, is launching a free educational certification course for developers on large-scale language models. Deep Lake, a vector database for AI data. LangChain is a framework for developing applications using language models. This certification program is part of the Deep Lake Foundational Model certification program.
The first part of the course combines a comprehensive overview of foundational model theory with hands-on projects using LangChain and Deep Lake to give you the tools you need to apply large-scale language models across industries to AI. Provide to developers.
Louie Peters, CEO of Towards AI, said the course was designed to be hands-on, helping engineers get started with AI in their own processes, or use foundational models to build entirely new products. said that it helps to
