Well you should have included something for guys like me: “I have no idea what Tailwind or DaisyUI bring to the table, I just use them because the team decided on it before my time”.

Daisy is fine for simple stuff but their no-js stance makes some of the more advanced ui very clunky using checkbox tricks etc

I like daisyUI because it provides a set of components which are extensible with Tailwind. I have found it much easier to extend daisyUI’s components than I ever could when using e.g. Bootstrap. Conceptually, I think it definitely goes against the grain of what Tailwind stands for, but it gives me some sane options out-of-the-box, which I can then modify as needed.

Personally, I’m in no position to start composing my own class-soup components, so I like having daisyUI as a baseline. Plus, it has good theming support.

I don’t like that the new versions are so bleeding edge in terms of browser support, but that will work itself out over time. It is nice that I no longer need to use npm to install daisyUI, which greatly simplifies the dependency stack for my back-office applications.

Like Tailwind, Daisy to me feels like a further step in the wrong direction.

But with that said I’m pretty sympathetic to the argument from the Phoenix team. This community simply does not have a large enough pool of talented designers to maintain our own components. And having things look good out of the box is needed for adoption. Given these constraints I’m having a hard time coming up with any other solution.

I think a better compromise would be to strengthen --no-tailwind into a new flag that simply elides all generated frontend code and leaves you with a bare project. No core components, no homepage, etc. Something that would require minimal effort to maintain.

I think that would do a lot to pacify those who want nothing to do with Tailwind or Daisy (a group of which I am actually a member, despite my recent comments).

Discussing Tailwind vs “Vanilla” CSS is not a binary discussion for me nor I won’t go into that here as I would need to write a book length explanation. I’ll just say both have their ideal scenarios and there isn’t a silver bullet for everything.

But regarding DaisyUI, I think it was a good choice as a base for Phoenix, it serves as a good base that is easy to extend or to rip out if not needed. I agree with @olivermt too.