The Open Source
Geospatial Toolbox

In the frontend, we have lots of fun updates. There is a new Schedules node that allows users to add cron (schedules) to their workflows. There is also a new feature for Pinning data to a node, which means you can save data to a node and avoid running that node repeatably every time the workflow runs. In the backend, we overhauled the node orchestration logic. Shockingly, this meant removing code and removing complexity and sticking with a much more 'simple' orchestration layer.

• •
  Pin Node Data
  Users can now pin data to a node. This is particularly helpful for 'expensive' or resource intensive nodes. During testing (or even some production use cases) users can Pin (save) data to a node and the node will use that saved data INSTEAD of re-running that node. An example is running a node that takes 5 minutes to run but will always return the same data. You can Pin that data and bypass the 5 minute execution for testing.
• •
  New Schedules Node
  Our schedules node is now working! You can specify a specific schedule for your workflow to run on, such as every day at 8am, or every hour, or anything your heart desires. These schedules can be enabled/disabled and will run on the server regardless of whether you have Geoflow open.
• •
  Update UI for Input and Outputs Node Tab
  We've noticed it was pretty difficult to see all the data that a node has access to, especially if there is lots of nested data. It was squished into a small Popover and there was too little UI to really use it properly. Instead, we moved this into its own separate side Sheets where there is now the full height to display all the node data. In addition, a field can now be drag/dropped into an expression to use directly in a node. Most importantly, in the new data UI, you can now see the data keys AND their current values. Previously you could only see the keys, but now the UI displays their current values as well.

PDAL execution is now working! PDAL node executions happen in their own isolated "geoflow-worker" container. The high level idea is that each PDAL node is combined to form a pipeline. A simple workflow might look like this: Load File -> Ground Classify -> Thin, reproject, etc -> Save File These nodes would form a single PDAL pipeline and execute in a single transaction (we are generically referring to this as a 'chain'). If there are multiple chains, it processes them sequentially.

Well, despite David's best efforts to overcomplicate building the a website, we now officially have one. In the early stages of this project, it will serve as a launch pad for news, updates, and feature releases (such as this one.)