Waqas Shahid

That’s a solid approach to gradual migration. Keeping breakpoints as Sass variables while exposing colors and spacing through CSS custom properties gives you the best of both worlds, especially when runtime theming is involved. For nesting, I’d generally aim to keep it around two or three levels deep. Beyond that, selectors can become harder to follow and tightly coupled to the HTML structure. There are exceptions, of course, but if nesting keeps growing, it’s usually a good signal to simplify the selector or split the styles into smaller components. Do you find that a strict nesting limit works well across a team, or is a shared guideline with code reviews enough?

That’s a great point! I think the naming problem mostly shifts rather than disappears. Tailwind removes the need to invent names for every small styling pattern, but once you start extracting components, you’re back to deciding what deserves a name and how reusable it should be. I usually find that extracting repeated patterns like buttons and cards early helps, as long as we avoid creating abstractions for every tiny variation. The goal is to keep things consistent without making the component structure more complicated than the styling itself. How do you usually decide when a pattern deserves its own component?

Good point. hx-boost does make progressive enhancement more interesting once redirects, flash messages, and CSRF protection enter the picture. Those details need to be handled deliberately on the server side rather than assuming every response can be swapped into the page. I agree with your point about collaborative and offline-first applications, too. HTMX works best when the server remains the source of truth and most interactions don’t require complex client-side state. Thanks for adding this perspective!