Hi Michael,
This one almost had me baffled, but I managed to get it to reproduce.
The file attached is at step 4. before the fillet, when you fillet this edge the cylinder vanishes.
The procedure was:
1. Offset surface 3mm


2. Extrude cylinder from circle.


3. boolean diff surface from cylinder


4.Fillet this edge 1mm


5. Hey presto! vanishing cylinder


---------

It "seemed" to be the seams not lining up between the curve surface and the cylinder. I rotated your surface to match the edges and then it worked fine.

Thanks for looking at that Grendel, but when you rotate the surface to match the seams the orientation of the convex shape is out by 90°, but when you mentioned that I rotated the cylinder 90° and it works also, thanks for pointing that out, but I think it should work either way.........Michael?

Cheers

I remember seeing that someone else had a similar problem due to seem alignment but I can't remember which thread it was.

Hi Grendel,

I got the same problem before then Danny come with a solution. The funny thing is now Danny on the same problem ha ha ha....Michael, its possible in the feature to refine "seam edge problem" ?

Hi Anis,

Yeah that's right! I remember now, but your problem was the opposite, you had to rotate the seams 90° away from each other for the fillet to work, funny that.
Found it ! http://moi3d.com/forum/index.php?webtag=MOI&msg=2264.14

---------

Hi Anis,

> Michael, its possible in the feature to refine "seam edge problem" ?

Yeah, it will probably continue to improve in the future as the geometry library continues to be updated and have problems fixed in it.

- Michael

Hi Danny, thanks for cooking it down to a reproducable example!

I can reproduce it with the Nov-21 v2 beta, but it seems to work properly in version 1.0, do you see the same thing over there?

So it seems to be a regression in the geometry library, but I have applied a new geometry library update for this next upcoming v2 beta and it looks like that has fixed this problem.

- Michael

Hi Michael,

> I can reproduce it with the Nov-21 v2 beta, but it seems to work properly in version 1.0, do you see the same thing over there?

Yes, it works fine in V1 for me also.

> but I have applied a new geometry library update for this next upcoming v2 beta and it looks like that has fixed this problem.

That's what you said a few betas ago, I thought an updated geometry library was a good thing ?
So when a new update arrives for the geometry library should we worry or be happy or be happily worried or worry happily or...... :)

Thanks for the update on the situation, Michael.

Cheers

Hi Danny,

> That's what you said a few betas ago,

Yeah the big update was in the Aug 1 beta. This particular case that you have here has probably been busted since then.


> I thought an updated geometry library was a good thing ?

Well, it can be kind of a mixed bag unfortunately... Whenever there is a change made to a complex set of code (and filleting is one of the most complex areas), to fix one particular case, it is pretty easy for it to have side effects that then cause a new problem in some other case. That is what is called a "regression".

In general code that processes geometric shapes can tend to suffer from this quite a bit, partly because there is such an infinite variety of shapes that can be used it isn't very easy to do something like an automated test that catches every situation that the code may be used in.


> So when a new update arrives for the geometry library should
> we worry or be happy or be happily worried or worry happily or...... :)

Probably a good dose of each is warranted! :)

But certainly the general idea is that more things should be better than be broken. If it is not, then I can always roll back to an old version.

Maybe if this new one bakes in pretty well I may not apply any more updates of this until after v2 is complete.

Actually as far as these things go, it has been pretty good so far regarding regressions.

- Michael