Showtime Komputeishon

About

On some local forum there was a series of discussions about pure-graphical program representations. It was going on and on. It was about how great it would be, and how text representations are bad because the (anonymous) person that started the discussions had a bad time learning any conventional text-based languages.

Somewhere in the middle, when I tried to poke around, I wrote about somewhat extreme version. Imagine multiple blocks connected by wires. There is a hidden clock that drives blocks, makes them to perform calculations, or to just propagate values through. Similar to digital logic components.

At the time the idea was just stupid. But when I read it again about a week later, it changed from just stupid to stupid and fun.

The thing

Some code, and there you have it. Jumping blocks. Why not?

Some more code, and the arithmetic addition works.

As you see from the timestamps, I wasn't very quick. It took some time. Meanwhile I was asked when a new video will come. And that's why the trailer was born.

Around that time I decided that I was going to draw a circle with that thing. A circle rasterization it not a very complicated algorithm, but it's not trivial. There should be some decision logic, which can be tricky. Especially if one needs to match the clock.

Visuals didn't change so much, but there were still quite a few additions. A graph node with point converters was introduced. Node freezing became a thing. So it became possible to freeze counters, reset their values, and then to start again, without disconnecting wires.

And finally, a circle was rasterized. The scheme was smaller than I initially expected it to be, since I was thinking about avoiding multiplications if possible. But then there is no way to make a multiplication node to not multiply numbers each cycle. That's how they are: they do their calculation each time the clock ticks. And since there must be at least some of them anyway, why not to use them more?

On the other hand, the interconnections are messy. They are even harder to track than I imagined in my head. I can't say it's impossible to figure out what the scheme does and how. But will anybody do that for real?

Conclusion

I was thinking how hard would it be to code for such programming environment, and how hard it would be to read others code. Circle rasterization was the code I hoped I still can do, although with struggle. Suprisingly, it turned out that predictions were correct. It's almost painful to watch that, if you imagine for a second that you need to figure out how that works.

Next time somebody tells me that graphical coding is easier, I'll have something to show them.

And by the way, here is the final source code: https://github.com/i-rinat/jokes/tree/master/showtime-komputeishon.