Replies: 1 comment 1 reply
-
Thanks for posting this. The current platform depends on the app being launched first, but the team will review this. Are you referring to this Ray? Thanks. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi all,
I’m exploring ways to avoid relaunching the
AppLauncher
(IsaacSim GUI) repeatedly during development. Specifically, I’m trying to launch theAppLauncher
once and reuse it, even as I make changes to my code.To experiment with this, I set up a system using Ray, where I define an Actor that:
AppLauncher
env_cfg
runner_cfg
The nice part is that the Ray actor process persists across multiple runs of my
play.py
script. This allows me to restart the script with different tasks, and it properly loads the newenv_cfg
andrunner_cfg
. Any changes I make to AppLauncher internals are reflected without restarting the simulation.However, I'm running into a problem:
If I modify code (e.g., any cfgs), those changes do not reflect in the simulation. It seems the version of the module that was imported during the first actor initialization is cached. Even if I use
importlib.reload()
, the modified code is not reloaded afterAppLauncher
has been launched.If I don’t launch the
AppLauncher
in the Ray actor (i.e., only load the env cfg, etc), the reload works just fine. So this behavior seems tied to theAppLauncher
lifecycle.Has anyone found a workaround for this? Or does the
AppLauncher
somehow lock module states after initialization?Thanks.
Beta Was this translation helpful? Give feedback.
All reactions