In web design, everything easy is hard again

>simply npm your webpack via grunt with vue babel or bower to react asdfjkl;lkdhgxdlciuhw

>[...]

>I had to install a package manager to install a package manager.

And that's coming from a designer, not a programmer.

A few years ago I had to take over a project that required installing something like 6 different package managers, 2 build systems, and a dozen different languages (counting things like jade and coffeescript), including dependencies that somehow required conflicting versions of the same libraries! (It took building pieces of it on 2 separate VMs and then merging them to get it to work.) All to build what turned out to be a simple little web site that in the end was basically 12 static pages and a simple form. That doesn't make things way easier.

When I started programming, the key important things were to focus on getting the data model accurate, getting the business logic correct, validating input properly, and producing the desired output. Now that is maybe 5-10% of our workload, the other 90+% is dealing with dependencies, build tools, frameworks, 57 layers of generic abstraction, and all the fragile interconnections and leaky abstractions and compatibility issues between all of those things needed primarily just to simply display a page of content or handle a CRUD form.

Sure, they're beautiful and interesting from an intellectual standpoint, and when they're working properly, they do make things that were once quite difficult a lot easier. But they also add a lot of complexity, require a lot of maintenance, and make the simple things much more complicated. Often, it's as if we're flying a squadron of helicopters down to the corner store to get milk (because isn't that cool and fun?!), and by the time the operation's completed and we've returned home and got all the helicopters repaired, we realize that we forgot the milk, or it's already spoiled, or we ended up with a potato instead.

I do love playing with the new toys. React is really cool, a lot of the new stuff is interesting. But all those layers, all that indirection, all that distances us from the domain data and business logic can make it a lot harder to solve real problems and produce real value. Sometimes I just want to build something and not have to deal with all that mess.

Actually that’s pretty much what I would expect if I tried to build interactive applications on top of a declarative document format whose standardization was co-opted by the largest advertisement company, while working with novices straight out of last week’s coding bootcamp eagerly reinventing the wheel out of their lack of experience and mentoring.

You're trying to conflate document design with UI design. By the writing of the author, he is clearly doing just fancy documents. He didn't even touch the problematic issues you normally face in web development.

You can do pretty neat things with trivial html/css these days, where you needed quite a bit of JS just 5 years ago.

Platform GUI toolkits are designed for controlling an interactive program from the start, a very different usage scenario. They were never designed to be documents to be read from top to bottom. Traditionally, GUI toolkit almost never included fancy text layout boxes just for text display. To give you an example, a simple text label to be used next to a button almost universally could be of a single font with the same style throughout.

I wouldn't even consider this a limitation, because on toolkits with advanced theming (GTK/QT), this allowed for a very consistent layout across all programs, which is clearly a much better user experience for the user as opposed to styled individual applications. Somehow, this crucial bit of /actual/ usability /design/ was lost.

The irony? Current GUI platform toolkits are actually copying the document model of web pages instead, so that your layout skill set can be more easily transferred across systems. So you can finally style your button's label to your hearth's content. XUL is probably the first example of this.

As if "layout" was the major issue in desktop user interfaces (HINT: it's not).

Now we have GUI toolkits which are basically web views that maybe call some native code. In my opinion: a huge failure in engineering and user experience.

I think it has a great deal to do with abandoning something simple not because it's not good enough to do the job, but because of fear of being left behind, fear of becoming irrelevant in a world where "new and improved" is introduced at an ever growing frequency.

Sure, improvement is a good thing, but the question I think we sometimes forget to ask ourselves is when does the cost involved in certain technological improvements (more learning curves, higher process complexity, energy required to stay on top of the rapidly changing technologies and tools etc.) start to become higher than the benefits produced by the so called improvement.

I've been in this field long enough to see how many times hype beats common sense, unnecessary complexity beats simplicity, and emotion beats logic (heck, I'd bet it's a common theme outside the technology sector as well) but I'm optimistic - the bottom line is that things are improving and even though things sometimes seem circular, it's more of an upward spiral.

Website, ha! There's a webpage pretending to be a programming IDE; even though it needs GBs of memory just for its overhead, it's hugely popular. shrug (Ditto anything else electron-ish)