If you have an example of the comp code switching sides I would love to see it. It would make a good test sample for future testing.
I ran these on my office computer which just got updated to XX.50 yesterday and it looks like there have been some fingers in the cookie jar. Comp is even more bizarre now! Now it is switching sides even on connected arcs. This example shows how MACH inserts tiny arcs . . a bonus. Another new behavior is that there is no separation between the original and comp toolpaths, Note the tool diameter and also note the scales. First screen shot is with both axis at -1 scale. Second screen shot only one axis at -1 . . BOOM! That is opposite of what it was doing in the earlier version in the shop. The behavior in the earlier version is also different in that the comp arc started on one side and endied on the other side of one of the long arcs that make up the sides of this part. If I changed the scale from -1 to -.999, it still switched sides, but at a different location on the tool path where I could live with it, so I did that and got the parts cut. As I siad, I do not use comp much at all, but typically, I find that comp seems to work OK on simple unscaled toolpaths, especially closed paths, but if you go 'outside the box' . . . it can get ugly.
I cannot release the G-code on this part. If this occurs on releasable code, I will be happy to provide the code. I understand it is difficult to debug without a repeatable example, sorry. I just came in to add some new video links to my 'mini machining center' thread and thought I'd poke around a bit and see if I could be useful, however I am unsubsribing now. I hope my comments were helpfull, but just do not have time to participate in-depth on forums.
