GitHub
08/25/2026, 12:51 AMworkspace.yml into the workspace you deploy to: a property you set is updated, one you omit is left unchanged, the workspace name comes from the file's name property, and app_properties are recomputed so removed keys are dropped.
• Deploy workspace artifacts — every other artifact backs a resource matched by name (alias for a dataset) derived from the artifact's filepath, minus the extension. It covers create, update and delete, and the update semantics: a property is applied as a replacement (an omitted property resets to its default), while a schemaless object such as pipeline properties is merged so only the provided keys change. Worked PipelineML examples show the replacement and merge cases, including an explicit null to unset a key.
This also corrects the PipelineML reference, which named the custom-script key inline_script instead of script.
Checklist
• I have read the contribution guide
Was generative AI tooling used to co-author this PR?
• Yes (Claude Code)
Generated-by: Claude Code following the agentic coding guidelines
Related Issues
None.
---
🤖 Generated with Claude Code
Summary by Sourcery
Document DataML workspace deployment behavior and correct the PipelineML custom-script field name.
Enhancements:
• Document how deploying DataML artifacts creates, updates, merges, and deletes their corresponding workspace resources, including replacement and schemaless-property merge semantics with examples.
• Document WorkspaceML deployment behavior, including workspace property merging and app property recomputation.
Documentation:
• Correct the PipelineML reference to use the script key for custom scripts.
meltano/meltanoGitHub
08/25/2026, 9:36 PM