I set out to reinvent the wheel and ended up rebuilding Windows

Anyone who works with the big electrical design programs knows the feeling of knowing exactly what you want to change and having no idea where it lives, because everything is nested, submenus inside submenus, and without training you never find half of them. And when you do change something, often you cannot even tell that anything happened. I wanted that to be different: above all it had to be findable. I wanted it to look good too, right from the start, but finding things comes first and polish comes after.

Everything about the element on the right, everything about the project on the left

That turned into two decisions that still stand today. On the right sits the inspector, which means you click an element and everything belonging to that element sits in one place and can be changed right there, no digging through eight levels, and you see the change on the sheet the same moment you make it. On the left sits the project tree, the way anyone from electrical design knows it, because there I deliberately reinvented nothing: anyone who has been drawing schematics for twenty years and moves to the Mac should find their way around in the first minute, without having to relearn anything.

Along with it came everything on the left that usually disappears into menus, so terminal plans, connector plans, cable plans, the PLC module, the parts list and the article database, all on one edge and all live, and whoever does not need the tree right now just switches it away. That layout is not a matter of looks, because whether you can find something decides whether you can work at all, and that is the part I got right.

And then I drifted

The starting point was clear, I wanted a Mac user to feel at home in Sketrix immediately, so everything should be built the way Apple builds it. I moved away from that without noticing: I had been a Windows person for years, so at some point there were Windows ideas all over the program, and on top of that came my own inventions, all of which I naturally considered brilliant at the time. I never actually decided on any of it, it was habit, and in the end it was a bit of everything and properly none of it.

The second mistake was bigger: I worked on the foundation of the app and the rebuild of the interface at the same time, like two separate building sites next to each other, and plenty got built, it even went well, I just could not check it any more at some point. In the end there were three hundred test points on my desk, and I was the only one who could work through them.

What that feels like up close shows in a single day in August. My records for the fourth alone hold: add a second sidebar, tune how the sidebars behave, fine-polish, rework in three rounds, then alignment and zoom rework. And on the same day there is an entry recording that an on-screen panel and the zoom are broken in dark mode, with double frames, a frame that fails to follow when you switch, and a bar that opens and springs straight back. Exactly the kind of fault everybody spots instantly.

The thing I spent a day on is dead

On the sixteenth of August I stopped everything, and the note from back then puts it plainly: stop until the book is written. The only exception was the one rebuild that was nearly finished, I saw that one through and then it went quiet. The same day brought the decision that the sidebar dies, of all things the sidebar that had eaten a whole day twelve days earlier, gone for good with no replacement.

Instead of building on, I sat down and decided the whole interface again on paper, in eight chapters, each with an honest inventory taken from the real code and with alternatives to choose from, and it ended in fifty-five separate decisions, every one numbered and reasoned. The sprawl of drafts turned into a book.

The goal has been fixed since then, a native Apple frame with a Sketrix skin, with Xcode, Pages and Numbers as the models, and the small things too, colour pickers and popup menus, come from Apple rather than from my own hand, because what Apple has already solved I am not solving again.

And now

The interface is not yet the one I want. Settings live in a window in the middle of the app, which is not how it is done on a Mac at all, I use the menu bar far too little, and the cabinet layout turned out really well but still feels like its own world next to the rest, so all of that needs pulling together.

For a while this bothered me, because I wanted to show how good Sketrix is and thought it would have to look good first, so for a long time I showed nothing at all. Sketrix does not look bad, it could just look considerably better, and I knew it with every screenshot.

By now I see it the other way round: as long as the foundation is not solid, there is no point putting a nice pair of glasses on it. The order underneath is built and it holds, what is missing is the shell around it, and that comes next.

Martin