sliptonic

There are a few factors working against the idea that a vibe coded solution will pop up and be viable. The first is access to actual hardware. A vibe-coder can build something for their own use but can't test it on equipment they don't have access to. This is non-trivial when you're talking about machines with really elaborate kinematics like rotating heads and tilting beds. That means a community has to form around the software.

The other thing is the same problem that has held back CAM development until now - domain expertise. Knowing the correct thing to build means having real-world experience as a machinist. Workholding, feeds and speeds, order of operations, tool properties, gcode quirks, postprocessing, etc, etc, etc. Modern CAM systems aren't just fancy toolpath algorithms. They manage the whole process of articulating the steps to manufacture the part. Unfortunately, the people that really understand that seldom know enough about software development to even vibe code it effectively. Maybe a team of those people could do it collectively but then you're back to building a community.

A community is what FreeCAD has now. We have about a dozen people working on CAM. That's why my post calls on the manufacturers to get involved.

A 2D solver works great for sketches and some kinds of assemblies. However, it is insufficient for many kinds of 3D assemblies. The solver Ondsel released (LGPL) which underlies the new integrated assembly workbench is a true 3D solver. https://github.com/Ondsel-Development/OndselSolver

Correct. Variable sets (varsets) is partially merged into upstream main development branch. Other work is ongoing. Varsets are much more important than just our web service though. Being able to configure and control variants in a design is underappreciated and will be essential to having complex assemblies with multiple copies of identical parts. Imagine a hinge that is reused in many places in the design. The current angle of the hinge should be controlled on a per-copy basis. When you change the angle, you don't want all copies everywhere to reflect the new state. This is a variant.

We're an open-core company building commercially around FreeCAD. We're leading the development of the new integrated assembly workbench and contributing to many parts of the application. We have our own build of FreeCAD which includes our latest contributions.

We're also building an optional web-based file sharing and collaboration layer.

It's great to see others investing time and talent into open-source CAD. At Ondsel, we believe open-source CAD is incredibly important and we're thrilled to see innovation in this area.

I guess the FreeCAD project should gear up for another wave of refugees

How long before people start to catch on?

The article barely touches on open-source but highlights the frustration that users have felt from the vendor lock-in that is typical of Autodesk.

Access to a real usability lab is one thing FOSS lacks. I've never figured out how to crack that nut. Maybe there's some way academic institutions could help fill the gap.

Part 5: https://ondsel.com/blog/default-assembly-workbench-5/

Next up is the assembly 3 workbench

Part 4: https://ondsel.com/blog/default-assembly-workbench-4

LWN article about the FreeCAD non-profit arm

Part 3: https://ondsel.com/blog/default-assembly-workbench-3

Part 2: https://ondsel.com/blog/default-assembly-workbench-1

The article links to the github issue for tracking progress. There are also several good reference links that go in-depth on the problem and algorithm. https://github.com/FreeCAD/FreeCAD/issues/8432

When your Definition of Done is a barrier to contributors (and other reasons working with first-timers is hard)

It was good to see the summary of FreeCAD mention the formation of the FreeCAD Project Association. Features, stability, and speed are absolutely critical but more and more for FOSS software to have long-term momentum, it needs a fundraising wing as well.