New Feature: Bridging Curve Plus

S1mon · January 23

I don't have an icon or an explainer page yet, but I wanted people to bang on this and tell me if there are any obvious issues. So things might change, but I hope this is a reasonable approach.

For a long time I've wanted to have a way to optionally make both sides of a Bridging curve work the same. I actually did 98% of this just by telling ChatGPT what I wanted.

https://cad.onshape.com/documents/0ae92c6d28bf061856d24872/w/2814604ba93ce6fdb7b1eb7b/e/436d3a65e3074595b77ece52

Screen Recording 2026-01-23 at 11.34.33 AM.gif

I fed ChatGPT the existing bridgingCurve.fs code from the standard library, asked it to summarize the functions and generate an inverted call tree to help me understand the code. I've reviewed the bridgingCurve.fs code a bunch of times in the past, and have some understanding of bits and pieces of it, but this was interesting and useful to me. Not sure if it was necessary to create this plus version.

I then asked it to do this:

I'd like to create a "Bridging curve plus" function which would add an "equal sides" toggle which would appear under the "edit control points" if it's on to make the two side controls equal. I'd like to import as much of the existing code as possible and if there's a good way to do this, make my version only override the things that absolutely need to be different so I can leverage any important updates from Onshape.

At first the result had a bug so I told it:

That almost works perfectly. One bug I found is if I type values into the "end magnitude" "end curvature offset" or "end flow offset", they do not do anything or get copied to the other side. The manipulators on both sides work as requested. Just tell me the changes you would make to fix this.

Then there was this bug:

Almost. So now if I type in values for side 2, it updates both sides, but not if I type into side 1. Typing values into either side should work.

With a couple of small edits and a little prompting, I got a working feature.

So what's wrong with this approach? What else would you like to see?

One surprise is that I assumed that it would be better to have the "equal sides" force both sides to have the same tangency option (i.e. G1G1, G2G2, or G3G3, but not G1G3). This doesn't do that, but the result seems still useful.