Hacker News Reader

Comment by jillesvangurp | Hacker News Reader

Basically this is what Figma could be if you could add behavior to the UI. Somehow the web has been running backwards for years now. We lost flash in favor of HTML 5, which sort of fizzled out in terms of creative people actually doing a lot with it. We used to have simple application builders in IDEs like Visual Basic, Delphi, Borland JBuilder. Even Eclipse had a simple swing UI builder thing for a while. For the web we had dream weaver, frontpage, etc.

I'm not saying those tools were perfect; because they weren't. But we never really got good substitutes for them. We now have people "designing" stuff in Figma and then handing over for implementation to some developer team that essentially then recreates the whole thing pretty much from scratch. There's something deeply stupid/wasteful about needing to do that.

View on HN

> We now have people "designing" stuff in Figma and then handing over for implementation to some developer team that essentially then recreates the whole thing pretty much from scratch. There's something deeply stupid/wasteful about needing to do that.

I'm not sure how wasteful it is really. There is some overlap, but not loads in my experience. Most of the time spent on frontend is the stuff behind the components: behaviour; device interaction; security; API calls.

I've had a weirdly large fraction of my career spent on making UI rather than business logic.

That said, the business logic in apps is, half the time, a needlessly complex shim on what is ultimately "here's a REST API you can do in one or two lines via `Codable` etc. except we're anticipating needing to support several API providers… oh no the money ran out".

> I've had a weirdly large fraction of my career spent on making UI rather than business logic.

Same here. None of my employers could go more than 2 years or so, before the head of UX would declare "Our app is no longer 'fresh' enough, and our competitors are adding 'pop' and 'zing' to their apps! We must stop what we are doing and totally redesign our app, create custom controls, and flex our brand onto the user!" Too much of my career has been turning a designer's Photoshop graphic pixel-for-pixel into an application, without regard to the OS's standard controls or conventions.

Yup, which is where I suspect most of the "visual UI builders are not the right tools to express dynamic constraints" objections come from: people are over-constraining the problem space with irrelevant nonsense. Half of the problems stem from abandoning OS standard controls, many of the rest are managing complexity of a large and complex UI codebase that didn't need to exist in the first place.

It's not like most webapp UIs are particularly well-aligned or consistent anyway, nor they are easy to visually overhaul on marketing teams' whim - so, what's the point? Why use tools that make important and impactful things between difficult and impossible?

(A possibly legitimate argument against RAD tools is that they encourage mixing business logic with UI code (static and dynamic). I used to be receptive to that, but I long come into conclusion that "separation of content and presentation" is bullshit both conceptually and in practice. And guess what, with React & all, we're all back to mixing UI and business logic - or adding extra layers of complexity to pretend we're not.)

Not mixing business logic kind of stops being possible when the UI needs to follow your business logic. We used to do much less complex, less specialized apps. And on the other hand - building an universal business rule engine just because you needed a wizard with three business process steps seems excessive.

There is quite some overlap, but the CSS generated by Figma is not really useful and of a grade I’d want in my application. On the other hand, there is a gradient where building in all the behaviour you could want in a designer application, makes the application so complex that it become unusable for designers.

> I'm not saying those tools were perfect

Bite your tongue. Flash was perfect.

I don't think visual UI builders were ever the way forward. These work well for static UIs, but once you try to express dynamic constraints, they fall apart.

Figma shows this: I find it useful to absolutely position elements (and maybe work with symbols) for a quick prototype, but once I bring in autolayout, I end up dealing with exactly the same issues as with CSS-flexbox and grids - except in a vastly inferior system, where I don't have version control, refactoring tools and chatGPT. Also, subjectively, I find it much easier to work directly with XML-structure, than to double-click (or cmd-click) through 15 groups in order to select a specific component, and then modify its properties in a textual form.

I'm guessing that's why Apple and Android deprecated their interface builders in favor of DSLs.

Reply on news.ycombinator.com