Under OS X (10.6.3) I often experience crashes when I try to change properties of a large selection of objects or groups. I have, for example, scenes with 150 or 200 objects and try to change their colour. I select them, open Node Params and adjust a colour - after a short moment, RF crashes. This also happens when the objects are grouped and I want change the colour of the group to affect all included nodes.
This error can be reproduced.
RF crashes with group actions
RF crashes with group actions
Thomas Schlick | Next Limit Technologies
Re: RF crashes with group actions
Hi Thomas,
Are you creating these objects with the "Create Array" tool?
Are you creating these objects with the "Create Array" tool?
Re: RF crashes with group actions
Hi Alex,
No, these objects have just been imported from C4D, for example pre-fractured walls or larger numbers of other objects, e.g. spheres. Changing properties at once works for approx. 50-75 objects, higher numbers will crash RF. It makes no difference whether these objects are grouped or part of a multi-selection. Changing RBD masses, colours or other parameters will cause these crashes.
The objects are correctly triangulated and don't have degenerated faces. Also simulations work fine and the problems are exclusively limited to parameter changes of large node selections.
Best,
Thomas
No, these objects have just been imported from C4D, for example pre-fractured walls or larger numbers of other objects, e.g. spheres. Changing properties at once works for approx. 50-75 objects, higher numbers will crash RF. It makes no difference whether these objects are grouped or part of a multi-selection. Changing RBD masses, colours or other parameters will cause these crashes.
The objects are correctly triangulated and don't have degenerated faces. Also simulations work fine and the problems are exclusively limited to parameter changes of large node selections.
Best,
Thomas
Re: RF crashes with group actions
Hi again Thomas,
thank you very much for the feedback. We have been able to reproduce the issue. It happens when selection is bigger than 100 and some object has been removed previously. We'll try to figure out why that magic number.
Thanks again
thank you very much for the feedback. We have been able to reproduce the issue. It happens when selection is bigger than 100 and some object has been removed previously. We'll try to figure out why that magic number.
Thanks again
Re: RF crashes with group actions
Thanks for the confirmation, Alex.
Maybe it has something to do with RF's undo cache, because with scenes containing a larger amount of objects I can sometimes observe the following: when I remove the original objects (e.g. a pre-fractured wall), add some new nodes and then close RF, I can see how the old objects are removed from the viewport during shutdown. They're not physically part of the scene, but it seems as if they're visible when the undo stack is emptied.
Best,
Thomas
Maybe it has something to do with RF's undo cache, because with scenes containing a larger amount of objects I can sometimes observe the following: when I remove the original objects (e.g. a pre-fractured wall), add some new nodes and then close RF, I can see how the old objects are removed from the viewport during shutdown. They're not physically part of the scene, but it seems as if they're visible when the undo stack is emptied.
Best,
Thomas
Re: RF crashes with group actions
Hi again Thomas,
yes, it is definitively related to the undo stack being modified during the execution of the parameter change.
Current safest workaround would be disabling the undo/redo stack. A bigger size of this stack doesn't guarantee you will never fill it and wrap around eventually.
Thank you very much for finding it. It is apparently fixed for the patch.
yes, it is definitively related to the undo stack being modified during the execution of the parameter change.
Current safest workaround would be disabling the undo/redo stack. A bigger size of this stack doesn't guarantee you will never fill it and wrap around eventually.
Thank you very much for finding it. It is apparently fixed for the patch.