feature request: structs instead of global workspace
ghorn opened this issue · comments
Greg Horn commented
The generated code uses a global workspace, including the third-party solver's workspace. I'm running into some drawbacks from this such as inability to simulate more than one actor simultaneously, inability to run multi-threaded, etc.
Was this done for simplicity, or is there a performance goal? Would you be open to PRs that encapsulate the workspace in a struct?
Greg Horn commented