Image-based workflow improvements and support for macOS & Ubuntu
The image-based selector feature now also supports macOS and Linux.
Added confidence slider, which can used to fine-tune the hits you get for your image.
Helps out for example in Windows where the contrast changes when the application is focused vs. unfocused.
Multi-monitor snipping is still experimental. Feedback is welcome from all platforms.
Keyword explorer documentation formatting
Keyword explorer now out of beta with improved documentation formatting.
Improved documentation formatting also visible in the documentation site: see an example
Robot run supports Python robots and output improved
Support for pure Python robots added.
More visibility to the output feed coming from the robot run.
Fixes
Cloud linking now works directly from Lab view: Upload to Robocorp Cloud directs to linking without having to exit your robot.
Loading a shared robot from robocorp.com/robots while in Lab view notifies you to save and exit the current robot to import the new one.
Known issues
Opening a robot with a conda.yaml that is not found in the cache takes a minute or two (Conda package loading). Closing the Lab view for the robot does not stop the cache set up, so the user needs to wait for it to complete.
JupyterLab v2 forces extensions build to happen on target machines, which makes the Lab initialization error-prone. We are pushing JupyterLab v3 upgrade hard to get rid of the issues on startups.
Miniconda, Chrome web drivers, and temp folder handling are causing file lock states on Windows machines.