Ok, some good news - I have been able to tune up some things to make this work for the next beta. I think this tune-up will help fix several boolean problems.

Here is a view from the side:




The distance shown in red there is pretty small, about 0.02 units. There was a problem where intersections that were close to other edges by about this distance would confuse the booleans and make it think that those areas should be merged together into one edge.

This was because of too loose of a tolerance value for the edges involved - once I added a mechanism to tune up the tolerance values on edges it allows the boolean to function properly.

This was kind of a general problem for any boolean that had an intersection that came to within maybe 0.04 units or so of another edge (without actually touching the edge).

I have to test this for a while to make certain that it doesn't mess up anything else, but I think it should be a good general improvement overall. There are still other things I'm working on for the next beta, so I don't think the next beta will be out until next week sometime.

I've attached the resulting model here. This was very helpful to get a problem example that had simple shapes in it, thanks!

- Michael

> Ok,thanks, here's another one, the same situation

Yup, looks like the same or related issue - I undid the boolean by separating out the surface and untrimming them (selecting edges and hit delete), then re-booleaned it with the new tune-up in place, and it seems to fix this one as well (model attached).

This version had several little curly-cue loops around the edges - I have seen these in a couple of models but I didn't know what was causing them, it looks like they are another side effect of too loose edge tolerances.

It seems like this may be a pretty significant fix for the next beta.

- Michael

Very glad to help! So, do I get a free license ??? :))

Geez, I figured that fixing bugs that you report would make it more likely that you'd want to buy it, not less likely ! :) :) :)

- Michael