← Back to all sparks
R

rolap

ANALYTICS
Velocity0.0

Obtaining Star Databases from Flat Tables

A star-schema modelling package grew a query language, a deployment path, and then a map layer.

olapstar-schemadata-warehousingspatial-datar-package
Current state
rolap builds dimensional models — star databases and constellations — from flat tables inside R. Over 2023 it acquired a multidimensional query interface, the ability to deploy models into relational databases, incremental refresh, and geographic layers exportable as GeoPackage. Since early 2024 it has been quiet, with the only 2025 release removing a test that clashed with another package.
Where it's heading
The 2023 releases trace a deliberate progression from modelling to operating: first a common data model and flat table class, then role-playing dimensions, then incremental refresh, then querying and deployment, then geography, then slowly changing dimensions. That is essentially the feature checklist of a data warehouse, assembled in about six months and documented with a vignette at each step. The pace since has dropped to almost nothing, which reads as a project that reached its intended scope rather than one that stalled.
Prediction
Given eighteen months in which the only release was a test removal, the next release is more likely to be maintenance than another warehouse feature — though the entries give no clear signal either way.

Recent moves

  1. 1y ago

    Conflicting test temporarily removed

    A single test removed because of a conflict with another package. The only release in eighteen months, and it says nothing about the package's direction.

    View source ↗
  2. 2y ago

    Slowly changing dimension structure added

    Adds a slowly changing dimension structure to the star database, filling in one of the last standard warehouse concepts the package was missing. It closes out the intense 2023 build-out rather than opening a new direction.

    View source ↗
  3. 2y ago

    Geographic layers and GeoPackage export

    Brings geography into the dimensional model with geoattribute definitions, layer queries and GeoPackage export. It lets a dimension carry real geometry, so aggregated results can leave the package as a spatial layer rather than a table.

    View source ↗
  4. 2y ago

    rolap 2.4.0

    ⚡ SPARK

    The release that turns a modelling tool into something you can query and run: star_query() with fact and dimension selection, plus deploy() and load_star_database() for pushing models into relational databases. Every later release extends a system whose operating surface was defined here.

    View source ↗
  5. 2y ago

    Incremental refresh for star databases

    Adds incremental refresh for star databases, with functions to derive new dimension instances and replay transformation code. Refreshing without rebuilding is what makes a model usable against data that keeps arriving.

    View source ↗
  6. 2y ago

    Flat table class and a shared data model

    Defines the flat_table class and a common data model shared by star databases and constellations, along with functions to query and modify dimension instances. The shared model is the foundation the querying and deployment work later builds on.

    View source ↗