Release Notes — v1.6¶

Release date: March 2026

This release adds audit column injection — automatic governance and traceability columns appended to thread outputs via configuration, with additive cascade inheritance across loom, weave, and thread levels.

Audit Columns¶

Configuration¶

Define audit_columns as name-expression pairs at any config level. Expressions use Spark SQL syntax and support context variables.

Loom level (applied to all threads):

# my-project.loom
defaults:
  audit_columns:
    _weevr_loaded_at: "current_timestamp()"
    _weevr_run_id: "'${param.run_id}'"
    _weevr_loom: "'${loom.name}'"

Weave level (extends loom columns):

# staging.weave
defaults:
  audit_columns:
    _weevr_weave: "'${weave.name}'"

Thread level (extends loom + weave columns):

# stg_orders.thread
target:
  alias: silver.stg_orders
  audit_columns:
    _weevr_thread: "'${thread.name}'"
    _weevr_source: "'${thread.source}'"

Additive Cascade¶

Unlike standard config inheritance where child values replace parent values entirely, audit_columns uses additive merge:

Each level extends the column set from higher levels
Same-named columns at a lower level override the expression
Columns cannot be removed from the inherited set

In the example above, stg_orders would receive all seven audit columns from loom + weave + thread.

Context Variables¶

Audit column expressions can reference execution context:

Variable	Description
`${thread.name}`	Thread name
`${thread.qualified_key}`	Fully qualified key (e.g. `staging.stg_orders`)
`${thread.source}`	Primary source alias
`${thread.sources}`	JSON array of all sources with type metadata
`${weave.name}`	Weave name
`${loom.name}`	Loom name

Context variables are resolved before Spark SQL evaluation. Parameter variables (${param.*}) and runtime variables (${var.*}) are resolved during config loading, before audit column injection.

Write Mode Support¶

Audit columns are injected for all write modes: append, overwrite, merge, and CDC merge. They bypass target mapping mode — always included in the output regardless of whether mapping_mode is auto or explicit.

Column Conflict Validation¶

If an audit column name conflicts with an existing DataFrame column, the engine raises an ExecutionError before writing. Rename either the audit column or the business column to resolve the conflict.

Telemetry¶

ThreadTelemetry now includes an audit_columns_applied field listing the names of audit columns injected during execution.

New Public Exports¶

The following are now exported from weevr.operations:

AuditContext — execution context for audit column variable resolution
inject_audit_columns — resolve variables and inject audit columns into a DataFrame
resolve_audit_columns — merge audit column dicts additively across config levels
build_sources_json — build the ${thread.sources} JSON array from thread source definitions

Compatibility¶

No breaking changes. The previously unused Target.audit_template field has been replaced by Target.audit_columns. Since audit_template was never implemented, this is not a breaking change — no existing configurations referenced it.

Component	Version
Python	3.11
PySpark	3.5.x
Delta Lake	3.2.x
Microsoft Fabric Runtime	1.3