Skip to content

Release Notes

Polars Cloud 0.0.7 (2025-04-11)

Enhancements

  • We now default to the streaming engine for distributed workloads A big benefit is that query results are now streamed directly to final S3/external storage instead of being collected into memory first. This enables handling much larger results without causing out-of-memory errors.
  • Polars Cloud now supports organizations. An organization can contain multiple workspaces and will eventually serve as the central entity for managing user roles and billing. During migration, an organization was automatically created for each of your workspaces. If you have multiple workspaces with the same name, you can now select one by specifying the organization: 
    workspace = Workspace("my_workspace", organization="my_organization")
context = ComputeContext(workspace=workspace)
  • Polars Cloud login tokens are now automatically refreshed for up to 8 hours after the initial login. No more need to login every 15 minutes.
  • A clear error message is now shown when connecting to Polars Cloud with an out-of-date client version.
  • The portal now has a feedback button in the bottom right corner so you can easily report feedback to us. We greatly appreciate it, so please use it as much as you would like.
  • Distributed queries now display the physical plan in addition to the logical plan. The logical plan is the optimized query and how it would be run on a single node. The physical plan represents how we chop this up into stages and tasks so that it can be run on multiple nodes in a distributed manner.
  • The workspace compute listing page has been reworked for improved clarity and shows more information.

Bug fixes

  • Improved out-of-memory reporting
  • Ctrl-C now works to exit during login
  • Workspace members can now be re-invited if they were previously removed

Polars Cloud 0.0.6 (2025-03-31)

This version of Polars Cloud is only compatible with the 1.26.x releases of Polars Open Source.

Enhancements

  • Added support for Polars IO plugins.
  • These allow you to register different file formats as sources to the Polars engines which allows you to benefit from optimizations like projection pushdown, predicate pushdown, early stopping and support of our streaming engine.
  • Added .show() for remote queries
  • You can now call .remote().show() instead of remote().limit().collect().collect()
  • All Polars features that rely on external Python dependencies are now available in Polars Cloud such as:
  • scan_iceberg()
  • scan_delta()
  • read_excel()
  • read_database()

Bug fixes

  • Fixed an issue where specifying CPU/memory minimums could select an incompatible legacy EC2 instance.
  • The API now returns clear errors when trying to start a query on an uninitialized workspace.
  • Fixed a 404 error when opening a query overview page in a workspace different from the currently selected workspace.
  • Viewing another workspace members compute detail page no longer shows a blank screen.