The real problem is that BREP CAD kernels are hard. A few of proprietary kernels dominate the scene: Parasolid powers NX, SolidWorks, Fusion, and Onshape, while ACIS (owned by Dassault) is used by Inventor and BricsCAD. Catia uses Dassault's own CGM kernel. The open-source world relies mostly on OpenCASCADE, which is unfortunately a lot less capable than any of these.

Fillets and chamfers are a good example. They seem simple but are geometrically non-trivial, and OCC will fail on cases that Parasolid handles without complaint. You can push either kernel to its limits if you try hard enough, but OCC hits that ceiling much sooner. So any CAD tool built on top of it inherits that ceiling too.

> the UIs demand continuous, thoughtful development

The current AutoCAD GUI is essentially unchanged from the 80s, so this shouldn't really be much of an issue. They added a ribbon probably 15 years ago at this point, but I can't think of any other major/recent changes. (But maybe there are some changes that I'm not familiar with, since I don't use AutoCAD very often and only started using it relatively recently)

> I feel CAD is one area where open source does not shine.

Many such cases, not only in CAD area. Good non-dev FOSS software is exception, not a rule, and these exceptions pretty often have some corporate backing and are not purely community-driven. And even for dev tools there are proprietary offerings that are light-years ahead of anything FOSS, though people here are never going to admit it, as TTY clone running vi clone is supposedly all you need.

I don't state this with satisfaction, quite the opposite, but it is long since I became disillusioned.

fusion 360 is exactly "free" for that reason, to make people reason like you do. They do not want a blender moment.

The reason being that Open Source is a bunch of people who approach EVERYTHING as a programming problem, and they are chronically allergic to graphics, graphical UIs, and any kind of sense of what user interactions are a good experience.

They don't start with "how do users want this to operate?" They start with a weekend of coding, applying their preconceived notions, a library of fancy algorithms that are not directly motivated by an actual feature, and they go from there. This does not lead to a good product, as in something that could earn you money on an open market. It only prevails, in spite of nobody wanting to pay for it, because they give it away for free, and they sink their own "disposable time" (and maybe even income) into the project.

I'm use Vectorworks professionally, but that's because it's become the de-facto program for drawing in the entertainment / production space, and uses Parasolid. I'd love to have any, ANY alternative to VW and Nemetschek's awful pricing / subscription scheme, but I don't see FreeCAD being a competitor in this space for many, many years. Maybe for hobbyists, but I agree that the problem space is oversized for a community project like FreeCAD.

Another advantage of OpenSCAD (if you can call it that) is that LLMs seem to be able to work with it pretty well. A few days ago I asked chatgpt to make me a box for storing batteries, and it came out perfect on first try without any modification. It also made an okay-ish looking 3D pelican after some back-and-forth.

I never understood this UI problem.

Because... You can copy the UI of the leader and problem solved.

There you have GIMP with an absolute nightmare UI to use, but people keep saying, just get used to it. On the other hand, a single developer, in javascript, made a copy of photoshop, and most people I know prefer to use that over GIMP...

Just copy the UI that works, if you can't research your own UI.

I don’t know, I tried FreeCAD a few months ago and it was buggy as hell. I did some really basic extrusions and distance constraints. But ended up with non-perpendicular entities despite not constructing it like this.