G4Exception in zero angle sbend

[William Shields]

The line:

BM10: sbend, angle=0.0, l=0.2;

causes the following exception:

-------- EEEE ------- G4Exception-START -------- EEEE ------- *** G4Exception : GeomMgt0002 issued by : G4SmartVoxelHeader::BuildNodes() PANIC! - Overlapping daughter with mother volume. Daughter physical volume BM10_1_of_1_pv is entirely outside mother logical volume World_lv !! *** Fatal Exception *** core dump *** -------- EEEE -------- G4Exception-END --------- EEEE -------

[Laurie Nevay]

Yes, the sbend and rbend geometry aren't particularly tolerant of 0 angle bends just now as different geometry is actually required - the same Geant4 primitive solids cannot be used in the case of zero angle.

This is unusual but a not impossible scenario. I will look into using different solids or making it tolerant of zero angle.

Just to check, there are hkick and vkick objects that are dipole correctors that do not bend the reference frame and the geometry will work with zero angle. Is the object genuinely meant to be an sbend?

[Laurie Nevay]

Fixed with commit #29df7fa. More testing used, better length calculation in case of 0 angle and no overlaps now.

In future, this could be achieved by creating a generic new BDSAcceleratorComponent that's like a collimator / shielding block but with flexible beam pipe type and a circular outside. This may be useful for general shielding. I guess that wouldn't produce the right shape for a poled magnet, but perhaps still of use and can be switched upon in BDSComponentFactory -> easier code-wise.