Toolpath generation issue

Hey Joel,

First thing I need is what release you were running when this happened — you didn't mention it, and it changes everything below.

Also, if you can still reproduce it, save it as a JC3 file and attach it here — not just the DXF. The JC3 carries all your parameters and any overrides, and if some odd setting on your end is triggering this, that's the only way I'll see it. The raw DXF won't capture it.

I tried to reproduce this on v2.3.8 and it generates perfectly for every mild steel cut-chart selection from 26ga through 7/8 plate. Back in v2.3.6 I fixed a batch of sneaky arc-fitting bugs, and one of them is exactly how this would manifest: the fitter's deviation check was point-based, so an arc candidate sitting across a straight edge with no sample points in the middle only got checked at the two endpoints — it could pass validation while ballooning out into a big arc across the gap. That's a circle/curve cut where a flat edge belongs, just like your screenshot. So:

• If you're on anything older than v2.3.6, update to the latest (v2.3.8) — there's a real chance this is already fixed, and if you can't reproduce it after updating, that's what happened.
• If you're already on v2.3.8 and still seeing it, then your panel is hitting a case my fix didn't cover, and I'll need that JC3 (with all your parameters/overrides) to chase it down.

Sorry you wasted some 10ga over it this morning.

Sorry, I failed to mention -- running v 2.3.8. I tend to stay with latest, updating pretty regularly here. Shop laptop runs Windows 11.

JC3 file attached. If I import that DXF to the job and generate toolpaths, the same arc is produced. So I can reproduce whatever happened... It's quite possible that it's a setting on my side. You have my machine config there from a former issue, I think? I can re-send if you need it again.

Thanks,
Joel

Okay, perfect. I was just able to reproduce it but ONLY when I have your machine config loaded and selected in the "MACHINE" selection so this must be something very sneaky. I'll let you know as soon as I figure out whats causing this to happen.

Perfect, thanks!

I don't think I've taken time yet to yank out my duplicate hole cutting config you pointed out earlier. No idea whether or not that's related, just something I thought of.

You're welcome!

No this isn't rules related. This is a real bug with arc fitting which I've been gradually making better and better from a performance standpoint. It used to be MUCH slower but it did far more thorough checks although this specific case is guarded heavily but this specific geometry breaks the check but only when the kerf is set to 0.060" as you have it. At 0.056" which is the Hypertherm kerf for 10ga mild steel it doesn't do it. I'll have more info shortly. I think I have it sorted on my end and I'll follow up if it passes all the tests (over 1k geometry tests). All these tests get ran constantly per build as I add new features to catch regressions but this specific geometry at this kerf slipped through anyways