Skip to content

Version 26.05.1.1

What's new

Fact tables and relationships in the RMD

Fact tables and their links to dimension groups are now part of the RMD, so the whole structure of a project can be reviewed, exported and imported in one place. The RMD screen has two new tabs: Fact tables and Relationships. The Fact tables tab lists the fact tables of the current project version with their name and description, and shows the measures, dimensions, facts and related dimension groups added to each of them as sets of pills; every column can be sorted and searched. Row actions open a fact table for configuration, open the edit dialog for its name, description and base database table, or delete it, and a new fact table can be created straight from the tab. The Relationships tab lists the relationships defined between fact tables and dimension groups, with the fact table, the dimension group, the key in the fact table, the key in the dimension group, and the relationship type, which is currently always Many-to-one. Its actions open the Edit relationship dialog, where the foreign key configuration can be changed or left empty, and a confirmation dialog for deleting a relationship, which removes the foreign key configuration in the corresponding fact table. Creating, editing and deleting on both tabs is available to the Superadmin, Company admin and Developer roles; the other roles see the tabs without the action buttons.

Lineage of measures that use derived dimensions

When you open the lineage of a measure whose formula uses a derived dimension, the chain behind that dimension is now traceable to the end. The derived dimension is shown with its own formula and is linked to the elements it is calculated from, down to the fields in the database table they come from, so you can check where the value of such a measure actually originates without leaving the lineage view.

Automatic population of fact tables

A fact table can now be assembled from the RMD automatically, instead of picking every element by hand. The dialogs for adding and editing a fact table have a Base DB table field, where you select a table from one of the project's connections; the selected table is saved in the fact table settings. If a base table is selected when you create a fact table, the system offers to fill the fact table in — choose Populate automatically to let it run, or Create empty to start from an empty configuration. Population walks the dependencies in the RMD: it first adds the primary dimensions and primary facts whose connected sources point to fields of the base table, then recursively adds the derived dimensions and derived facts calculated from the elements already added — together with the primary and constant facts they need — and finally the measures that can be calculated from that set. Dimensions that belong to a dimension group are added together with their dimension group and stay in it. When population finishes, the measure blocks are laid out on the configuration canvas by dependency level: measures calculated directly from dimensions and facts first, then measures calculated from those, and so on. You can move any block afterwards and add or remove elements manually before saving.

An already saved fact table can be refilled with the Repopulate action, or by changing the base database table in its edit dialog; in both cases the system asks for confirmation before anything changes, and cancelling leaves the fact table in its last saved state. Repopulation removes the current measures, dimensions and facts and adds them again from the RMD, while previously added dimension groups, their foreign keys that belong to the selected base table, and the verification filters of the fact table are kept. A verification filter that ends up referencing an element which is no longer in the fact table becomes invalid instead of being deleted, and becomes valid again once the missing element is added back. If the formulas involved contain circular references, the operation is cancelled with a warning and nothing is changed; if nothing at all matches the selected base table, the current configuration is also left as it was.

Read connection configuration through the API

External tools and AI agents can now retrieve how a data warehouse is reached without anyone sending them the details by hand. The public API returns the connections of a project version — with optional filtering by database type (PostgreSQL, ClickHouse, MS SQL Server) and by connection status — the full parameters of a single connection, and its stored schema with the available tables and columns. Passwords and other credentials are never returned, and no call opens a connection to the target database or runs anything in it: the schema is the snapshot taken when the connection was set up or last refreshed, and the response states when that snapshot was made. Requests are authenticated with an API key, are limited to the projects of your own company that you have access to, and every attempt is written to the audit log.

Read data mart configuration through the API

The structure of assembled data marts is now available through the public API, so an agent or an external application can find out what a data mart contains and how to query it. You can list the data marts of a project version, filtered by data mart type and merge type or searched by name and description, read the complete configuration of a single data mart, and read the metadata of the physical object created for it in the database — its kind, database, schema, name, creation time and connection, or an explicit answer that no such object exists. A separate call returns the SQL SELECT statement generated for the data mart, with an optional row limit and offset. The API only builds and returns the statement: it is never executed, no data is previewed, and no physical object is created or changed. All project roles have read-only access to this data.

Read data model configuration through the API

The logical model of a project version can now be read through the public API alongside its measures, dimensions and facts. New endpoints return the dimension groups with their primary key, member dimensions and the fact tables that reference them; the fact tables with their measures, dimensions, facts, verification filters and linked dimension groups; and the relationships with the fact table, the dimension group and the keys mapped on each side. Each group has a list endpoint and a details endpoint, and a combined export returns the RMD content together with these data model entities in a single response with the time of the export. These endpoints also accept format=xlsx, which produces an Excel workbook and returns a signed link for downloading it. Access rules are the same as for the rest of the public API: an API key, read-only, restricted to the projects of your own company, and recorded in the audit log.

Unified lineage view

The product now shows one and the same lineage view for an element wherever you open it from, so there is no need to look for the screen with the right version of it. Lineage can be opened for measures of any type, for derived dimensions and for derived facts: on the RMD through the View lineage button on the control panel, which appears when a single element with an enabled formula cell is selected, and from the Details from RMD dialog on the data model, fact tables and data marts screens. In the view, derived dimensions and derived facts are shown in individual blocks with their formulas and are linked to the elements used in those formulas, primary facts are grouped in a single facts block where constant facts show their assigned values, and primary dimensions are grouped in a single dimensions block. With Select layers you can turn off the fact tables and dimension groups layer, which is shown by default between the elements and the database tables, and switch the database tables between the complete tables and a condensed form that contains only the fields used in calculations; facts and dimensions that belong to no fact table or dimension group are linked directly to their field in a database table. Clicking an element highlights its name in bold in every formula shown in the view, several elements can be highlighted at once, and clicking the background clears the highlighting.

The view is read-only: elements cannot be added or removed there, connecting lines cannot be drawn, and the system does not ask you to save it, because it is rebuilt automatically every time you open it. Anything you change through the Details from RMD or Edit measure dialogs appears in the lineage as soon as it is saved.

Improvements

  • Long values no longer stretch input fields and drop-down lists: in the dialogs for adding and editing verification filters inside a fact table, the field keeps its size when a long value is selected, and a value that does not fit is truncated with an ellipsis (IMP-2537)
  • The public API is now available only to companies with a valid license: a request from a company without one is rejected with an error stating that the API is not available without a valid license (IMP-2596)

Bug fixes

Manage fact table details

  • Fixed: changing the base database table of a fact table no longer clears the fact table configuration and the tables taken from the database (IMP-2676)

Manage product data import/export and API

  • Fixed: the order of the dimension columns returned by the API now matches the order of the columns in the application (IMP-2532)
  • Fixed: the last step of an import no longer ends with a server error (IMP-2666)

Manage project connections

  • Fixed: in a newly created project, the check boxes of the connected databases can now be selected while a fact table is being created (IMP-2516)

Manage RMD table content

  • Fixed: a double click on a column header no longer opens edit mode for the first cell of that column on the RMD tabs (IMP-2536)