BSpline Bug

If a rational periodic (U or V) bspline surface (i.e all weights = 1) is converted with SetVNotPeriodic or SetUNotPeriodic, it generates the wrong number of Poles/Knots
and a wrong Knot distribution.

I found (after much trying ) a workaround in the sources marked as


but seems like the bug is still there...
The workaround simply puts a weight (1,1) to 0.3 , does the conversion and then restores the weight to 1.0. It doest works well :)

At least put this in the documentation!!

Fabian Hachenberg's picture

What's the status of this bug? Is it a bug or not?

QbProg's picture

I did not test this in 6.5

Forum supervisor's picture

The specified reference (//szv#10:PRO19566:05Oct99) is very old.
At the moment we don't have information that this bug still exists.
If you can provide any test case to reproduce it we will recheck it
and will inform you about results.

QbProg's picture

The comment was old, but the issue permained.
I will try to do and post a test, but simply create a periodic bspline surface (periodic in U or V) , with all the weights to 1, and try to convert it to a non-periodic one.
it will likely have an uncorrect number of knots/poles in the periodic direction

I will let you know if this persisted in 6.5