687: Hosting NextJS × Spicy Sidedishes × Modern Forms × Abandoning TypeScript

Episode Summary

Executive Summary: This Syntax potluck covers practical web-dev tradeoffs: Next.js hosting and vendor lock-in, browser and CSS evolution, database-as-a-service choices, CMS-driven forms in modern stacks, TypeScript usage changes in Svelte/Turbo, React Native vs Electron/Tauri, error-throwing conventions, progressive enhancement with newer CSS features, and foundational runtime concepts like processes and daemons. The hosts emphasize picking tools by needs, not hype, and note how platform choices shape developer experience and lock-in.

Main Topics: Next.js hosting and vendor lock-in: Discussion of whether Next.js ties teams to Vercel, how Netlify/Cloudflare/SST/OpenNext attempt to support the framework, and how runtime differences affect portability. Browser politics, CSS progress, and Firefox/Safari comparisons: A debate about whether Firefox is the 'new IE,' the reality of browser feature lag, and praise for Safari's recent leadership in shipping advanced CSS/JS features. Database-as-a-service categories: Comparison of hosted database products like Supabase, PlanetScale, Neon, and related platforms, with emphasis on distinguishing pure databases from full-stack data platforms. Forms in modern JS apps: Pain points around recreating WordPress-style editable forms in headless CMS/React stacks, with frustration toward rigid SaaS form builders and lack of a good bridge between CMS and custom UI. TypeScript, Svelte, Turbo, and migration concerns: Clarification that Svelte/Turbo changes are about authoring and build steps, not abandoning types; also why many companies hesitate to migrate large React apps to Next.js. React Native vs Electron/Tauri: Explanation that React Native targets true native mobile UI, while Electron/Tauri wrap web apps in desktop webviews, making them different tools for different platforms. JavaScript/runtime fundamentals and workflow conventions: Definitions of process, worker, job/task, and daemon, plus a brief answer on 'throw new Error' as mostly a convention rather than a functional difference.

Key Arguments: Next.js can be hosted elsewhere, but Vercel offers the smoothest access to framework-specific features like middleware and serverless/edge deployment. Platforms like SST and OpenNext exist largely to reduce lock-in by recreating Vercel-like ergonomics on AWS/serverless infrastructure. Browser feature gaps are no longer about one browser being universally worse; different vendors lead in different areas, especially Firefox and Safari for advanced standards. The modern web stack still lacks a strong solution for editor-managed, CMS-linked, custom-branded forms; in many cases developers still end up hard-coding forms. Supabase-like platforms and PlanetScale/Neon serve different needs: some are full application platforms, others are closer to standard databases with serverless-friendly scaling. Svelte's move away from TypeScript authoring is about reducing complexity and build overhead, not rejecting types in shipped code. React Native is fundamentally different from Electron/Tauri because it targets native UI layers rather than browser webviews. Many newer browser APIs can be used progressively: if unsupported, the app can often simply degrade gracefully without breaking core functionality. 'throw new Error' is largely a convention for explicitness; the functional outcome is the same as throwing an existing error object. Workers, jobs, tasks, processes, and daemons describe different layers of computation and background execution, useful for understanding web app infrastructure.

Data Points: Podcast episode reference: 551 - Referenced as a prior episode about serverless databases Syntax team summit location: Toronto - Hosts discussed meeting listeners and holding a team/content summit Sentry acquisition timing: end of March - Wes noted Syntax joined Sentry about six months earlier, around end of March Listener survey frequency: monthly - They said Potluck questions are done once a month PowerBlock increment example: 10 pounds - Scott described his adjustable dumbbells as changing in 10-pound increments Additional PowerBlock handle increment: 2.5 pounds - Scott noted extra small plates in the handle allow finer adjustments PowerBlock age estimate: 20-25 years - Scott said his set was found at a yard sale and was likely this old Form submission cap example: 100 submissions - They mentioned a Typeform pricing limit that required upgrading after 100 submits Browsers/platforms mentioned: 4 browsers - Discussion of modern browser interoperability and standards compliance Episode number mentioned for CSS survey: 3 - Wes said the new fuzzy finder helped locate episode 3

Pivotal Quotes: "I think that is by design. It's probably the reason why they've gone so hard into Next.js and have invested so much money on, like, kind of, I don't want to say taking over the React platform, but really becoming the React platform." — Scott: On Vercel/Next.js ecosystem lock-in and strategic platform control "There is no service, as far as I know, that bridges that gap." — Scott: On the lack of a good modern service for editable CMS-driven forms in custom JS apps "The progressive enhancement for that is to just not have an animation." — Wes: On using the View Transitions API with graceful fallback

Implications: Teams should choose stacks based on deployment needs, workflow fit, and lock-in tolerance. The episode suggests the web is improving fast, but developers still need pragmatic fallbacks, especially for forms, hosting, and cross-browser support.