Release Notes — v1.8¶

Release date: March 2026

This release adds dictionary rename — named column sets that source bulk column mappings from Delta tables, YAML files, or runtime parameters — and naming enhancements including kebab-case, collision dedup, and reserved word protection.

Dictionary Rename (Column Sets)¶

Configuration¶

Declare column_sets at the weave or loom level. Each column set maps raw column names to friendly names from an external source.

column_sets:
  sap_dictionary:
    source:
      type: delta
      alias: ref.column_dictionary
      from_column: raw_name
      to_column: friendly_name
      filter: "system = 'SAP'"
    on_unmapped: pass_through
    on_extra: ignore
    on_failure: abort

  runtime_overrides:
    param: column_overrides

Reference a column set in a rename step:

steps:
  - rename:
      column_set: sap_dictionary
      columns:
        MANUAL_COL: manual_override  # static wins

Source Types¶

Source	Configuration	Use case
Delta table	`type: delta`, `alias` or `path`	Shared dictionary in lakehouse
YAML file	`type: yaml`, `path`	Version-controlled mappings
Runtime parameter	`param: <name>`	Orchestration-driven mappings

Behavior Controls¶

Setting	Values	Default	Description
`on_unmapped`	`pass_through`, `error`	`pass_through`	DataFrame columns with no mapping entry
`on_extra`	`ignore`, `warn`, `error`	`ignore`	Mapping entries for non-existent columns
`on_failure`	`abort`, `warn`, `skip`	`abort`	Source read failure or empty result

Cascade¶

Column sets cascade from loom → weave. Weave-level definitions with the same name override loom-level definitions.

Naming Enhancements¶

kebab-case Pattern¶

kebab-case is now a supported naming pattern:

naming:
  columns: kebab-case

Backtick-quoting required

Hyphenated column names require backtick-quoting in Spark SQL expressions. A validation warning is emitted when selected.

Collision Dedup¶

When normalization produces duplicate column names, the new on_collision setting controls the behavior:

naming:
  columns: snake_case
  on_collision: suffix  # _2, _3 suffixes in source order

Value	Behavior
`error` (default)	Raise `ConfigError` listing collisions
`suffix`	Append `_2`, `_3` in source column order

Reserved Word Protection¶

Prevent SQL-reserved column names from breaking downstream queries:

naming:
  columns: snake_case
  reserved_words:
    strategy: prefix
    prefix: "_"
    extend: [timestamp, value]
    exclude: [to, by]

Strategy	Behavior
`quote` (default)	No-op — Spark backtick-quotes automatically
`prefix`	Prepend configured prefix string
`error`	Raise `ConfigError` listing conflicts

The default reserved word list contains 82 ANSI SQL:2016 keywords. Use extend and exclude to customize.

Pipeline Order¶

The full naming pipeline executes in a fixed order:

Rename — column set + static mappings applied
Normalize — pattern transformation (snake_case, etc.)
Dedup — collision resolution (suffix or error)
Reserved words — protection applied (prefix, quote, error)

Observability¶

ColumnSetResult captures per-column-set resolution outcome (name, source_type, mappings_loaded, skipped)
WeaveTelemetry.column_set_results populates structured results from materialization
DEBUG-level per-mapping logs for rename, normalize, and reserved word operations

New Public Exports¶

From weevr.model:

ColumnSet — column set domain model
ColumnSetSource — source configuration for column sets
ReservedWordConfig — reserved word protection settings

From weevr.telemetry:

ColumnSetResult — per-column-set resolution result

Compatibility¶

No breaking changes. All new fields default to None or safe defaults — existing configurations work without modification.

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