Memory Management Questions
I'm having a problem with my simulations crashing when they run out of memory, and at this point I'm pretty sure it's not caused by a memory leak, but here are some questions about memory management just to be sure:
Does a class that inherits directly from Abstract need to call Abstract's destroy method?
Do lists containing only primitive types (ints) need to be freed in the destroy method?
If objects in a list have references to each other, but are not referenced by anything else, is freeing the list
enough to free all references? (the objects in the list don't free objects within their destroy methods because they
assume the objects will be freed by freeing the list)
I'm pretty sure I'm deleting everything that needs to be gotten rid of. My problem is that I have a neural network representation that supports arbitrary connectivity, and therefore makes use of node objects to define its structure. I use mutations to add nodes and connections between nodes, but there's no bound on the number of nodes a network can have. So even though I'm not leaking memory, the amount of memory I need for nodes is unbounded. The amount increases with more and more mutations.
-Jacob
Arbitrary Neural Space
Write a plugin in c/c++ or python then you can use the native memory management of those languages. I wrote a (pretty much) ansi c plugin for neural networks that can be grown up to a predefined size but that probably isn't what your looking for.
I'll give it a shot. By the
I'll give it a shot. By the way, although I'm currently writing simulations in steve, I know that they can be written in python as well. Is writing the simulation in python the same thing as using a python plugin? Or are plugins completely separate from other code, even if I happen to use python to write it?
Then again, no matter what the answer is, I'd assume that C++ is the better option for speed and memory management. What's your opinion?
Memory Management and Plugins?
ckowall,
How do you build a separate interface library or plug-in for breve with only python? Could you point me in the direction of some sample code, or documentation?
Also, with ai a fail close must be a prime directive. Breve many times fails open with out warning. It's only ethical for robot, say part falls off, no good if he try to run off, and bump into other robot.
Python example would be large help.
Respectfully yours,
Doug