Robocorp Lab changelog and release notes

Robocorp Lab (4.22.8)

New features

• Dynamic Robot template updates
  • Added Playwright -template and more coming.
  • All templates also now visible in Portal
• Updated and curated Keyword Explorer content to match rpaframework v11.4.0
• RCC updated to v11.4.3 --> rcc configuration speedtest
  • In Lab, use File > Terminal and run rcc configuration speedtest command to test the performance of your system.
  • System requirements are notoriously hard to write up as in virtualized environments, we cannot trust just the hardware specs.
  • The speed test results are set so that zero is a "good enough" value, positive values are good and negative values indicate slower performance.
  • The results are split into operations over network and operations over filesystem
    • These are not just transfer speeds, as in virtualized systems, you can have blazing fast internet, but the CPU actions related to those might slow down the result.
    • Same goes for the filesystem measurement; the specs of the hardware mean nothing if the actual file storing and reading is slow due to some virtualization layer or a heavy virus scanner process.

Bug fixes

• Update to RCC v11.4.3
  • Added retry mechanisms to attempt retries earlier and in smaller chunks.
  • Added retries to removal processes as well (Windows file locks)
  • Dropped minimum worker count to 2 to ease the load on minimal hardware.
• Lab kernel switching now jumps directly to the new one without an extra step to "backup kernel"
• Dependency updates for most extensions
• In Python the problem with a={"b":"c"} format fixed
• Fixes to JupyterLab Tornado timing out, causing stuck states to Restarting or Connecting states.

Known issues

• Having `defaults '-channel in conda.yaml will cause errors; now is the time to get rid of that.
  • Parts of the Lab basement are no longer shipped via defaults -channel, causing weird errors.
  • Templates and robots in Robocorp Portal have been updated since February, but basically, you only need to remove that one line.
  • Simplest conda.yaml example is:

channels:
  - conda-forge
dependencies:
  - python=3.7.5
  - pip=20.1
  - pip:
      - rpaframework==11.4.0

• "Lab is unresponsive" on some machines.
  • Use rcc configuration speedtest to determine the cause and ping the result on Slack / Forum.
    • Disk speed seems to be the most common cause for slow environment setups
    • A Windows machine with 2 CPU cores does struggle quite a lot as Windows Defender (or anti-virus scanners in general) pretty much occupy half of the processing power trying to scan the incoming files.
  • Lab base dependencies are big, and we are working on getting a significant cut to those (via ipywidgets v8).

Robocorp Lab (4.21.1)

New features

• Recent Robots listing implementation changed not to show deleted items
• Set Global Variable & Set Suite Variable values are now held in cell-by-cell running
• Restart kernel -button added to the notebook button panel
  • All the kernel commands are still available behind Ctrl+Shift+C > type restart
  • Having a button there simplifies working with Python libraries.

Bug fixes

• Update to RCC v11.1.6
  • Reduced use of os.stat as they seem to affect Windows usage
  • V11 migration guide at: RCC changelog
• Lab kernel environments are now rotated to avoid file lock when changing conda.yaml
  • Note: If robot logic opens a process that freezes or does not shut down when Lab asks it to, you can still get file locks.

Known issues

• Having `defaults '-channel in conda.yaml will cause errors; now is the time to get rid of that.
  • Parts of the Lab basement are no longer shipped via defaults -channel, causing weird errors.
  • Templates and robots in Robocorp Portal have been updated since February, but basically, you only need to remove that one line.
  • Simplest conda.yaml example is:

channels:
  - conda-forge
dependencies:
  - python=3.7.5
  - pip=20.1
  - pip:
      - rpaframework==11.4.0

• "Lab is unresponsive" on some machines.
  • RCC v11.1.6 improvements to performance.
  • Kernel environments rotation now reduces the change of file locks which are compounding the issue.
  • Big dependencies still take time on the first run of a unique conda.yaml.
    • At the moment, pip for example, does not support any parallel actions, so machine utilization is really low during pip steps.
  • A Windows machine with 2 CPU cores does struggle quite a lot as Windows Defender (or anti-virus scanners in general) pretty much occupy half of the processing power trying to scan the incoming files.

Robocorp Lab (4.20.18)

Bug fixes

• Update to RCC v11.1.3
  • Reduces the number of parallel workers to reduce the file I/O load that seems to affect some Windows machines.
  • Now using Copy and rename pattern to block any termination in the middle of file writing from causing broken files.
    • Should fix the INIT_3_10 and INIT_4_10 errors that some users have been hitting hard.
  • v11 is now cleared from all old-style cache commands and logic
    • rcc env ... -commands removed and replaces by either rcc holotree ... or rcc config ... commands
    • Breaking command changes highlighted in RCC changelog
• Lab startup checks improved > removed redundant initial checks

Known issues

• Having `defaults '-channel in conda.yaml will cause errors; now is the time to get rid of that.
  • Parts of the Lab basement are no longer shipped via defaults -channel, causing weird errors.
  • Templates and robots in Robocorp Portal have been updated since February, but basically, you only need to remove that one line.
  • Simplest conda.yaml example is:

channels:
  - conda-forge
dependencies:
  - python=3.7.5
  - pip=20.1
  - pip:
      - rpaframework==11.4.0

• "Lab is unresponsive" on some machines.
  • While we still have not been able to repro this on our side, we got some profiler data that helped us get further, thanks to some accommodating community members.
  • Patch v4.20.18 now avoids most of the failure cases we have seen in a way that breaks the failure cycles and resumes better.
  • Patch v4.20.18 also reduces the number of parallel file actions, which seems to be the triggering aspect.
  • Big dependencies still take time, and at the moment, pip for example does not support any parallel actions, so machine utilization is really low during pip steps.
  • We have a couple more ideas in the works still on this matter.

Robocorp Lab (4.20.15)

Bug fixes

• JupyterLab now cleans up all processes left behind > unclosed browser instances were still causing a problem
• On macOS, the image-based locators now work.
• On macOS, crashes caused by File > Shutdown fixed
• Attaching to opened browser in cell-by-cell running back to working order
• If the environment building fails, there is now a Details-button to guide on further steps
  • Shows diagnostics run for the robot structure that highlights potential problems in .yaml and .json files
  • Shows the raw RCC log to see the pip and micromamba outputs to find the potential problem.

Known issues

• Having `defaults '-channel in conda.yaml will cause errors; now is the time to get rid of that.
  • Parts of the Lab basement are no longer shipped via defaults -channel, causing weird errors.
  • Templates and robots in Robocorp Portal have been updated since February, but basically, you only need to remove that one line.
  • Simplest conda.yaml example is:

channels:
  - conda-forge
dependencies:
  - python=3.7.5
  - pip=20.1
  - pip:
      - rpaframework==11.4.0

• "Lab is unresponsive" on some machines.
  • Patches .12 and .15 most likely will affect the frequency of this issue, but as we have not been able to reproduce this, so the hunt is still on.
  • Happens after using Lab and manifests as the whole UI slowing down
  • Happens mainly on Windows and Linux
  • We have not been able to reproduce this, so issue reports with additional details are more than welcome:
    • When this happens, what processes are taking CPU time?
    • Does this happen with specific types of robots? Interested in conda.yaml content.
    • Does this happen when changing from robot to robot?

Robocorp Lab (4.20.12)

Bug fixes

• JupyterLab no longer leaves behind orphan python processes
  • This has been causing environment update failure on Windows machines
• Special character handling in env.json
• Blocked opening multiple Scratchpads at the same time
• JupyterLab zmq error fix

Known issues

• Significant slow-down in Lab responsiveness
  • v4.20.10 fixed the issue for some user but not all, so the hunt is still on.
  • Happens after using Lab and manifests as the whole UI slowing down
  • Happens mainly on Windows and Linux
  • We have not been able to reproduce this, so issue reports with additional details are more than welcome:
    • When this happens, what processes are taking CPU time?
    • Does this happen with specific types of robots? Interested in conda.yaml content.
    • Does this happen when changing from robot to robot?
• On macOS, the image-based locators fail to start.
  • We are hunting for the reason. It seems to have something to do with mac signing.