[Gmsh] Problem: extrusion with rotation

[Robert Funnell]

On Sun, 28 Dec 2008, Christophe Geuzaine wrote:

> Robert Funnell wrote:
>> I am trying to create a geometry corresponding to a cylinder which has a 
>> bend. As shown in the Gmsh code appended below, I am using an extrusion 
>> with translation, a 2nd extrusion with translation + rotation, and a 3rd 
>> extrusion with just translation.
>> 
>> The 2nd extrusion begins from the end of the 1st extrusion, as intended. 
>> However, the final circular surface of the 2nd extrusion isn't located at 
>> the end of the extrusion, but rather is offset in the y and z directions, 
>> so the 3rd extrusion doesn't start in the right place. (The effect can be 
>> seen in the screenshot at 
>> http://audilab.bmed.mcgill.ca/~funnell/gmsh_bad_extrude.png .)
>> 
>> If I do either translation or rotation alone in the 2nd extrusion, I get 
>> the results that I expect.
>> 
>> Could someone please explain what I'm doing wrong here?
>> 
>
> Your rotation looks weird. Here's an example:
> ...
> extr2[] = Extrude { {1,0,0}, // direction of rotation axis
>                    {0,2,20}, // a point on the rotation axis
>                    -Pi/2 } { // the rotation angle
>   Surface{extr1[0]};
> };
> ...

Ah, how embarrassing. I was somehow thinking that I needed to combine 
an extrusion with the rotation in order to get the desired radius of 
curvature. And if one combines an extrusion with a rotation in a 
nonsensical way then one gets a nonsensical result :-) Thank you.

- Robert