Robocorp Lab (4.20.10)
The big changes caused some problems, so patches are in place.
- Significant slow-down in Lab responsiveness
- We have not been able to reproduce this, but we have found the probable culprit thanks to the submitted issues and the logs therein.
- In the case where the environment was not ready, Lab was spamming the code-completion extension, which generated many actions that eventually failed.
- macOS able to launch multiple Lab instances, causing failures
- When closing Lab via mac red-cross and then spamming the dock icon, had a bug that allowed multiple windows starts
Kernel errorwhen opening .robot file right after opening a robot
- Related to the slow-down issues, now the pre-built kernel is spun up immediately, avoiding the case where no kernel exists.
- Lab initialization behind slow network starting over from scratch on any failures
- Labs cleanup method now update to use RCC's new cleanup that leaves the mamba cache in place
- We can now do this as micromamba has the logic to check and validate its cache.
- Smaller fixes:
- External links fixed
- Terminology now inline to start using
Robocorp Control Room
- JupyterLab leaving behind orphan python processes
- When opening and closing robots in Lab, we noticed that JupyterLab is leaving behind orphan Python processes.
- This does not create any immediate problems for Lab users, but over time especially on Windows, this could lead to file locks in the environments left open.
- Workaround: Unfortunately, the simplest way to get rid of the processes is to reboot the machine.
- We are pushing for a fix to JupyterLab and looking for a solution in Lab.
- On macOS the image based locators fails to start.
- We are hunting for the reason, seems to have something to do with mac signing.
September 1, 2021