I share this which illustrates what, to me, is odd behavior. YMMV.
This comes about from an edge fillet that was a nightmare to get as a solid (trying combinations of radii and fillet types).
Problem FIllet.3dm.zip (1.5 MB)
I am using a display mode with tangent lines disabled to try to find edges that should be tangent. This may be problem in itself.
#1 I have the original solid
#2 I have edge filleted, rolling ball at 1/32" but expanding to 1/8" at right.
My display shows the edges at the corner (which has been the problem area) are not tangent. At that point, I probably should have just celebrated that I finally got a closed polysrf and ignored this small, apparent defect.
#3 However, to try to illustrate the problem in a small subset, I used boolean intersect to extract the corner.
#4 Then I fillet at 1/32"
Lo and behold, the corners show up as tangent when done on the extracted polysrf where they are not in the original???
#5 From #2 (filleted original), I extract and delete the two corner surfaces. Then I take #4 and extract its corner surface and graft it into the openings (with a mirror copy for the other side).
And now I appear to have a closed solid polysrf with all the edges that are supposed to be tangent looking like they are tangent.
Assuming there is no flaw in my method of diagnosing edges that are not tangent, it seems odd to me that the corners in #1 would not be tangent while those in #5 appear to be so.
I share in case this illustrates something to improve EdgeFillet.