Summary of ReactiveConf 2017 - Tom Dale: Secrets of the Glimmer VM

00:00:00 - 00:30:00

Tom Dale, creator of the Glimmer VM, discusses how the VM helps to improve web performance by breaking up work into small chunks and allowing for incremental rendering. He also explains how Glimmer helps developers reduce the amount of wasted time and effort spent optimizing updates and optimizing the performance of their applications.

• 00:00:00 Tom Dale discusses web performance in relation to lightweight and efficient apps. He argues that the web cannot compete with native apps in terms of complexity and experience, but that typescript gives glimmer a head start in this area.
• 00:05:00 Tom Dale discusses the benefits of using a functional programming language to build and update the DOM, and how the glimmer compiler helps to reduce the amount of code that needs to be sent to the browser and the amount of work that it has to do once it gets there.
• 00:10:00 Tom Dale discusses the different ways in which templates can be compiled into bytecode, and how this affects the download size of an application. He also explains glimmer, a small JavaScript runtime that efficiently encodes templates into an executive or program of Dom and component related operations.
• 00:15:00 Tom Dale explains how the JavaScript parser takes a hundred and fifteen milliseconds by comparison, while the JSON parser takes only 7.6 milliseconds, but because JSON is so much simpler of a language, parsing it is much much faster. Dale goes on to discuss sending templates, and how binary data is simply put there is no parse, so here's how you initialize the glimmer application: you grab this binary data, GPX files, and then you get the raw array buffer back from fetch. He then passes that into glimmer and so here's deep in the glimmer source code you can see they take that array buffer as literally the array buffer directly and they start running the VM against that. They create a UN sixteen array, so there's no intermediate data structure, and just see how all this goes together. In a little demo, he has a script that pulled 50 random Wikipedia articles and then converted them either into a handlebars template or into JSX. He then renders all 50 of them simultaneously in one go, not at all realistic, but the intent of this is to stress test the initial render performance of pre-act and glimmer. He then moves on to the next topic, 60 frames per second incremental rendering. Unfortunately, 16 milliseconds isn't a
• 00:20:00 Tom Dale, creator of the Glimmer VM, discusses how the VM helps to break up work into small, 16 millisecond chunks that allow the browser to remain responsive even when there is a lot of work to be done. The VM also allows for incremental rendering, which is dramatically different from the synchronous implementation of render. Finally, Dale discusses optimized updates, which use a technique called partial evaluation.
• 00:25:00 Tom Dale discusses the differences between Glimmer and React, demonstrating how Glimmer's faster initial render allows for less work overall when updating a component's state.
• 00:30:00 Tom Dale explains how glimmer, a template-based VM for running web applications, offers significant performance advantages over traditional web applications. He also discusses how glimmer helps developers reduce the amount of wasted time and effort spent optimizing updates and optimizing the performance of their applications.