I saw this issue in inkscape 1.2.1 for linux. Instead of explaining I'll upload a video, but basically doing boolean operations shifts the position of things unexpectedly. It can be easily reproduced. Pay attention to the guideline

I had a slightly newer version on windows and the issue was not there, so I thought maybe it was fixed in a later version, so I waited.

Yesterday I downloaded inkscape 1.2.2 for linux - that's a newer version than I have on windows - but the bug is still there, so perhaps it's a linux only bug?

Specific linux version: Inkscape 1.2.2 (1:1.2.2+202212051550+b0a8486541)

Windows version: Inkscape 1.2.1 (9c6d41e410, 2022-07-14)

Is this the right place to post this? thanks

Do you have line thickness for this circle?

I don´t see any possibilities in Inkscape to avoid this by using simple 4-node-circle objects. Other SVG editors show the same when not worse results. Increasing the "Numeric Precision" in Preferences or rotating the circle a tiny bit won´t help either.

But what kind of works is to use a polygon instead of the circle object with an odd amount of corners - say 17 for instance - convert with Path->Object to Path - then select every node and set to "Symmetric node" on both objects. The result contains a deviation smaller than one thousandth of a px.

So I wouldn´t call it a bug per se - more an insufficient algorithm. (VectorStyler is so far the only sufficient editor in this regard in my arsenal)

I just realized it also happens on windows. What a pity

@monsterfox no line thickness

Will have to try what Polygon says. I do tile work and ocasionally geometric work for laser cutting, things have to be perfectly lined up

The video doesn't show for some reason. Here's a gif. Hopefully it displays

It's a deficiency in cubic Beziers (or such). Polygon has the best workaround, using a multifaceted polygon converted to smooth path.

You that video I saw, two round trimmed will appear after the trimming position of the two ends of the high and low, but if you use Taylor's method, after construction, although the position will also still deviate some, but the two ends are the same. Taylor that solution is the best. These details can be ignored in my work, because his values are too small. I think your tile cutting this small problem should also be irrelevant. There will be some burrs after the comparison cut.