You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A standalone example to detect Ignition process having terminated and relaunch Ignition.
In the example, show how to take optional arguments (can be dummies in the example) that the user can pass in to load file X and start process Y.
This is an extension to / intermediate step before #137 (a big ticket to finish) is implemented.
Motivation
Applications with long running-time, e.g. large-scale data collection, require simulation to be running for tens or hundreds of hours. Instability in GUI greatly discourages usage and inconveniences users, who may need to babysit the simulation and relaunch manually whenever simulation crashes.
Many machine learning libraries have the ability to save the intermediate model parameters. For repetitive training tasks, we don’t need #137. Users may be able to live with relaunching automatically and start a new training block from where the learning model was saved.
Even with #137, this will still help with automatic relaunching.
Alternatives considered
Currently, users either need to write this by themselves, or babysit the simulation manually, or consider an alternative with more stable GUI for large-scale training.
Implementation suggestion
Add the example script to ign-gazebo/examples.
Take optional arguments (see top).
Add a tutorial to ign-gazebo/tutorials.
You should test the rendered content (things render differently on GitHub and Doxygen generated pages).
To do that, open in a browser the generated page, which should have a path like file:///path/to/ign/build/ignition-gazebo6/doxygen/html/index.html
Check that images, code snippets, etc special content are rendered correctly.
Additional context
The text was updated successfully, but these errors were encountered:
Desired behavior
A standalone example to detect Ignition process having terminated and relaunch Ignition.
In the example, show how to take optional arguments (can be dummies in the example) that the user can pass in to load file X and start process Y.
This is an extension to / intermediate step before #137 (a big ticket to finish) is implemented.
Motivation
Applications with long running-time, e.g. large-scale data collection, require simulation to be running for tens or hundreds of hours. Instability in GUI greatly discourages usage and inconveniences users, who may need to babysit the simulation and relaunch manually whenever simulation crashes.
Many machine learning libraries have the ability to save the intermediate model parameters. For repetitive training tasks, we don’t need #137. Users may be able to live with relaunching automatically and start a new training block from where the learning model was saved.
Even with #137, this will still help with automatic relaunching.
Alternatives considered
Currently, users either need to write this by themselves, or babysit the simulation manually, or consider an alternative with more stable GUI for large-scale training.
Implementation suggestion
Add the example script to
ign-gazebo/examples
.Take optional arguments (see top).
Add a tutorial to
ign-gazebo/tutorials
.You should test the rendered content (things render differently on GitHub and Doxygen generated pages).
To do that, open in a browser the generated page, which should have a path like
file:///path/to/ign/build/ignition-gazebo6/doxygen/html/index.html
Check that images, code snippets, etc special content are rendered correctly.
Additional context
The text was updated successfully, but these errors were encountered: