881: What Happened to Remix. Worth Using Axios? Client Only Next.js?

Episode Summary

Executive Summary: A potluck episode covering practical frontend/backend decisions and developer tooling: Axios vs fetch, where HTMX fits and where it breaks down, database hosting for local-first apps, choosing modern React frameworks when keeping backend separate, ad platforms for dev podcasts, career advice for a pilot transitioning into software, password storage basics, extracting non-exported package internals, update-tracking workflows, responsive testing, and designer feedback dynamics. The hosts emphasize choosing tools based on coupling, flexibility, and real-world maintainability rather than hype.

Main Topics: Axios vs Fetch Tradeoffs: Listeners challenged the hosts’ skepticism of Axios. The discussion highlighted Axios strengths like interceptors, error handling, and custom instances, while arguing fetch better aligns with web standards and cross-platform primitives. HTMX Limits and Client-Side Needs: HTMX works best for tightly coupled server-rendered HTML apps, but falls short for apps with local-first data, multiple frontends, or substantial client-side state management. Alpine, Inertia, and React/Svelte/Vue-style approaches were suggested alternatives. Local-First Database Hosting: The hosts compared managed Postgres/MySQL providers, platform-style services, and self-hosting. For zero/local-first, Postgres hosting like Neon was favored, but backups, pooling quirks, and proximity to app servers were emphasized. Choosing a Frontend Framework with a Separate Backend: For teams wanting a standalone API and separate frontends, Next.js is not always the best fit. TanStack Start, React Router 7, Astro, Vite + React SSR, and Waku were discussed as lighter or more flexible options. Podcast Monetization and Ad Networks: The hosts explained that developer podcasts usually monetize best via direct sponsor relationships, though Buy Sell Ads was praised as a reputable intermediary in the dev ecosystem. Career Transition into Software Development: A pilot planning a transition into tech was advised not to expect a senior role immediately. The hosts recommended building substantial real projects, freelancing, or co-founding something to gain real-world experience. Developer Workflow and Maintenance Practices: Questions on password storage, hidden package internals, update tracking, mobile emulation, and design feedback were answered with pragmatic advice: hash passwords, patch or copy missing internals responsibly, use changelogs and TypeScript, don’t rely solely on device presets, and give constructive design feedback early.

Key Arguments: Axios is still valuable for interceptors, normalized error handling, and base URLs, but fetch is preferable when you want web-standard Request/Response primitives and better interoperability. HTMX is strongest when one server owns one HTML-based UI; it becomes awkward when the app needs local-first state, multiple UIs, or native/mobile clients. Managed database platforms save ops burden, but backups are non-negotiable; hosted services should still provide a clear recovery story. If backend and frontend are intentionally separate, a framework that leads with routing and SSR/static rendering is more appropriate than a full-stack opinionated framework. Direct sponsor relationships are usually more effective than ad networks for developer podcasts because value depends on familiarity, trust, and product relevance rather than raw impressions. A newcomer from another profession should focus on shipping real software and gaining operational experience rather than expecting a senior title immediately. Hashing must be one-way; encoding like Base64 is reversible and therefore unsuitable for password storage. If a dependency hides useful functionality, patching or vendoring the code is acceptable, but upstreaming a pull request is the cleaner long-term path. Staying current requires a mix of changelogs, release notes, TypeScript compiler feedback, and selective reading of project blogs rather than one universal update system. Responsive design should be fluid and tested broadly; device presets are mostly useful for viewport sizing, not for simulating real devices accurately. Design disagreements are best handled through respectful collaboration, not silent overrides or resignation; otherwise products drift toward lower quality.

Data Points: Nothing Ear A price: $79 sale / typically $89 - Scott’s replacement earbuds after losing AirPods Pro Nothing Ear battery life: ~3x longer than AirPod Pro - Scott reported all-day battery life compared with his prior earbuds Small Rig hardware standard: Quarter-inch screw - Wes described camera/phone rig accessories built around a common mounting standard Sentry alerting: Slow query alert identified an unindexed query - Used as an example of observability replacing some need for platform-provided slow-query tooling Xior bundle size: 3 KB gzipped - Mentioned as a tiny Axios-like alternative TanStack Start composition estimate: ~87% TanStack Router + TanStack Query - Scott described the framework’s architecture informally Developer ad-block usage: ~60% - Scott cited a rough estimate for ad blockers among developers Minimum responsive width discussed: 320 px - The hosts debated whether 320px is still a relevant breakpoint floor Device examples mentioned: iPhone SE 375 px; iPhone XR 414 px - Used to argue that modern breakpoints vary widely

Pivotal Quotes: "HTMX is good if you have a server and a web-based UI that are tightly coupled in a way that the server is going to be returning HTML." — Scott: Explaining the best-fit scenario for HTMX "Hashing is one way. It is irreversible. You cannot unhash something." — Scott: Answering the password storage question "I probably would say reach for Tanstack start or reach for React Router 7." — Scott: Recommending alternatives to Next.js when backend stays separate

Implications: Listeners should choose tools based on architecture, not trends: use web standards when possible, keep server/client boundaries intentional, prefer real-world experience over résumé shortcuts, and invest in observability, backups, and collaboration to reduce long-term complexity.