nerd-toolkit / nerd

http://nerddoc.x-bot.org
Other
8 stars 0 forks source link

Simulator/Physics: Creation of CollisionObjects #23

Open josefbehr opened 11 years ago

josefbehr commented 11 years ago

Is there a specific reason that CollisionObjects are mostly created in the constructor of a simulation object, but never explicitly de-registered or destroyed? Would it be reasonable to include the creation and destruction of CollisionObjects into the setup() and clear() processes?

cybott commented 11 years ago

Die you check whether the collision objects are automatically destroyed in the simbody base Class? Currently I cannot check this myself. Setup and cleanup are called during each reset, so collision objects should not be created and destroyed there to avoid the unnecessary overhead.

Josef Behr notifications@github.com schrieb:

Is there a specific reason that CollisionObjects are mostly created in the constructor of a simulation object, but never explicitly de-registered or destroyed? Would it be reasonable to include the creation and destruction of CollisionObjects into the setup() and clear() processes?

— Reply to this email directly or view it on GitHub.

josefbehr commented 11 years ago

Apparently, there once was code in SimBody.cpp (https://github.com/nerd-toolkit/nerd/blob/master/nerd/simulator/simulator/Physics/SimBody.cpp#L156) that removed the CollisionObjects, but it has been commented out and stated the owners should take care of that. Either that or the code is de-commented, I guess.

cybott commented 11 years ago

Then the best measure for now is to keep it unchanged and to do a check for memory leaks later. I try to find out why I commented out the delete statements in the first place.

Josef Behr notifications@github.com schrieb:

Apparently, there once was code in SimBody.cpp (https://github.com/nerd-toolkit/nerd/blob/master/nerd/simulator/simulator/Physics/SimBody.cpp#L156) that removed the CollisionObjects, but it has been commented out and stated the owners should take care of that. Either that or the code is de-commented, I guess.

— Reply to this email directly or view it on GitHub.