Ever since software engineering has become a matter of mostly prompting an llm I have had almost no inspiration to continue working on my side project codenamed “vantage”. I am not sure if I am going to continue working on it. When I started with the project, which was about three years ago, every feature used to be a challenge.

I still remember vividly a lot of late night programing sessions. These include the conceptualization of the main event loop, the integration of the iced runtime, the issues related to the display of modals and the resulting accomplishment when the recursive context menu was displayed correctly on top of all other ui elements. The last features I was working on were a gizmos to shift and rotate renderables and Lua-scripting to auto generate them. gizmos.rs is 100% llm generated and mostly works. However, “mostly” is certainly not good enough. I am also sure that I have no spirit left to prompt claude all day and explain the shortcomings of the produced outputs.

Continuation

The natural question is how to proceed. I already started greatly reducing the amount of my free time spent on software engineering. I do not intend to stop learning about the intricacies of software altogether. Only the amount of time I spend writing actual code (or generating it). To compensate for the missing learning opportunities I started reading Bjarne Stroustrup’s book on C++ and studying MIT opencourseware lecture “6.867-fall-2006” (machine lerning), as well as finally getting my homebuilt vfd to work.

The only remaining task to be able to try my vfd is to solder all components onto my custom pcb. I designed the pcb in kicad and built a case for it using onshape. While I did that I felt as if software tools with specialized agents are going to be able to do this for me in the (near) future. This was somewhat confirmed by recent news. After designing the case for my vfd I downloaded zoo design studio to try cad agents. I explained the cad agent what i wanted, while already having drawn the model in onshape, and waited for it to finish. Overall, it resulted in more or less the same thing in much less time.

Now why would anyone switch from pursuing software engineering projects in their free time to electrical or mechanical engineering when the current outlook is the same? I just thought that despite being able to design a cad model in minutes there is still a joy in expermination, iteration, assembly and testing. I.e. the remaining tasks left for me – the human – are a lot more interesting than what you do when creating software with software.¹

Up Next

I hope this point of view is not too negative. I am actually quite excited by the agents and capabilities coming to conventional (e)cad applications, especially for hobby projects. I am also strictly not giving up on building software in my free time. Just relocating how I spend my time. If you are interested in the source code of vantage just contact me. It could be helpful to someone using iced and building a graphics heavy application.

I expect to write a blog post about my vfd in the near future. I still need to solder roughly half of the components onto the pcb. I really did my homework for this attempt. I followed the datasheets and ratings to the T and integrated proper high power connectors into the case so the bench setup is going to be clean. I am looking forward to trying it out.²

1. For me this is mostly reviewing code. ↩︎
2. I wrote this post 100% by myself. No llm got to touch any word at all 🙂 ↩︎