ingestr v1.1.33 - OneLake Target Resolution Fixes

ingestr v1.1.33 was published on August 17, 2026, with a narrow OneLake correction: destination resolution now handles workspace names containing spaces and GUID item targets. For data engineers running OneLake loads, the important change is more reliable target selection rather than a new connector surface.

The full release notes and downloads are on the GitHub release page.

The first fix covers OneLake workspace names that contain spaces. This sounds small, but destination addressing is part of pipeline correctness. In this release, the significant input shape is a workspace name with embedded whitespace, and the connector now handles it correctly.

The release notes do not document the previous failure symptom. They do not say whether affected runs stopped during validation, failed during connection setup, or selected an unintended target. Operators should therefore avoid assuming a specific old behavior from the short changelog. The supported conclusion is narrower: v1.1.33 corrects handling for this workspace naming pattern.

This also defines the practical test case. Run a representative load against a workspace whose name contains at least one space. Confirm that the expected item receives the data, then compare row counts and schema with the source. A successful process exit is useful, but destination identity is the stronger check for a target resolution fix.

The same release fixes OneLake item targets expressed as GUIDs. This is a second addressing case, distinct from the human readable workspace name. The target value can be an identifier rather than a readable item name, and v1.1.33 corrects that path.

The notes do not name a CLI flag, environment variable, or connector configuration key. They also do not describe accepted GUID variants or any normalization rules. Keep existing configuration syntax and validate the exact identifier form already used by the pipeline. Changing unrelated settings during the same rollout would make a resolution failure harder to isolate.

For a useful smoke test, send a small and recognizable data set to a GUID target. Check the resolved destination before treating row delivery as proof. If the platform exposes both display names and identifiers, record both in the deployment evidence so a later incident can distinguish a data problem from an addressing problem.

The changelog lists only these OneLake target fixes. It does not announce schema changes, credential changes, new migration steps, or breaking behavior. v1.1.33 is also a regular release, not a beta, alpha, or release candidate. That makes a focused connector test more appropriate than a broad pipeline redesign.

Teams that use workspace names without spaces and do not address items by GUID have no operator visible change described in the notes. They should still run their normal regression checks, but there is no documented reason to alter schemas or orchestration settings for this tag.

For affected deployments, test both corrected cases independently. Use one run for a workspace name with spaces and another for a GUID item target. Independent runs make it clear which resolution path passed. Keep retries low enough that repeated attempts do not obscure the first error, and retain the resolved destination details with the run logs.

The v1.1.33 release page is intentionally brief. Operators should treat it as a targeted correctness patch and limit conclusions to the two named OneLake cases.