FeatureMesh FeatureQL Documentation

website·Indexed 30 days ago

https://docs.featuremesh.com

Documentation for FeatureQL, a functional, entity-centric DSL that transpiles to SQL (DuckDB, Trino, BigQuery). Includes a function reference, core concepts and patterns such as adjacency and hybrid queries, and a getting started guide.

Tokens: 10.5K
Snippets: 48
Records: 98
Agent score: 50%

What's inside FeatureMesh

Handle multi-currency consolidation and budget variance
Consolidation patterns in FeatureQL handle FX conversion and budget variance by converting each entity's local metric before summing them.
https://docs.featuremesh.com/llms/index_concepts_consolidation_patterns.mdwebsitellms
Handle many-to-many relationships
Many-to-many links in FeatureQL are implemented as their own entity, acting as a bridge or junction table.
https://docs.featuremesh.com/llms/many_to_many.mdwebsitellms
Implement array operations in FeatureQL
FeatureQL uses arrays of rows as the join surface inside an entity. Key array operations include MERGE, field access, UNWRAP, and ARRAY_AGG.
https://docs.featuremesh.com/llms/index_concepts_array_patterns.mdwebsitellms
Configure time-based sliding windows
Implement sliding windows using RANGE BETWEEN INTERVAL … PRECEDING AND CURRENT ROW on a timestamp column that is sorted via ORDER BY. This can be applied at the top level or inside a TRANSFORM block.
https://docs.featuremesh.com/llms/sliding_window.mdwebsitellms
Model bi-temporal data (Valid vs Recorded time)
FeatureQL supports a bi-temporal model that tracks two independent time axes: valid time (when the event occurred in the real world) and recorded time (when the event was entered into the system).
https://docs.featuremesh.com/llms/index_concepts_temporal_patterns.mdwebsitellms
Manage composite keys for cross-entity patterns
A composite key is used when a fact is identified by two entity IDs together, facilitating cross-entity data patterns.
https://docs.featuremesh.com/llms/index_concepts_cross_entity_patterns.mdwebsitellms
Create inline data tables with INLINE_COLUMNS
The INLINE_COLUMNS function allows developers to create small data tables (mock data or lookup tables) directly within FeatureQL without requiring an external SQL source.
https://docs.featuremesh.com/llms/inline_columns.mdwebsitellms
Chain TRANSFORM calls for nested aggregation
To perform successive aggregation (aggregating and then re-aggregating or filtering after grouping), chain two .TRANSFORM() calls together.
https://docs.featuremesh.com/llms/chained_transform.mdwebsitellms
Model bitemporal data in FeatureQL
A bitemporal model in FeatureQL tracks two independent time axes: valid time (when the event occurred in the real world) and transaction time (when it was recorded in the system). This is used for handling late corrections and semi-additive metrics.
https://docs.featuremesh.com/llms/bitemporal.mdwebsitellms
Create polymorphic custom functions (UDFs)
User Defined Functions (UDFs) extend FeatureQL with custom scalar functions that can accept multiple input types and emit results using dialect-specific SQL.
https://docs.featuremesh.com/llms/index_concepts_reuse_patterns.mdwebsitellms
Deduplicate rows within an array
Use the deduplication pattern to keep only one row per key inside an array of rows. Common implementations include first_last_per_group, top_n, or using row_number to keep the first or last record.
https://docs.featuremesh.com/llms/deduplication.mdwebsitellms
Perform non-equi joins with RELATED...ON
Use RELATED(<aggregate> ON <predicate>) to produce one value per outer entity when a standard key-only VIA relationship cannot express the required match (e.g., range joins, as-of joins, or theta joins).
https://docs.featuremesh.com/llms/related_on.mdwebsitellms