If a face's edge lies on linear curve and (?) pcurve scaling produces wrong model.
- pload MODELING
- pcylinder p 10 50
- explode p f
- tscale p_1 0 0 0 0.001 #p_1 is a cylindrical face
- tscale p_2 0 0 0 0.001 #p_2 is top planar face
#p_1 is wrong, p_2 is fine
- cylinder c 50
- mkface f c 0 5 0 10
- tscale f 0 0 0 0.001
# f is wrong
My assumption is that tscaling sets wrong SameRange/SameParameter flags - it keeps them true while ranges have changed and no longer match each other.
Can't check if it's a regression vs 6.3.0 or is an old issue.
Hope this helps.