Skip to main content

Release notes 26.3.3

· 3 min read

Dataedo 26.3.3 enables profiling of JSON columns and makes Data Profiling banners more accurate in Portal. It also fixes object dependencies on PostgreSQL repositories, along with several import issues in BigQuery, Databricks, Paste Metadata, PostgreSQL and SQL Server.

Portal:

  • Data Profiling:
    • JSON columns can now be properly profiled
    • The Details column in the Columns whose run failed section of the table Profiling tab now correctly shows why each column failed.
  • Ui changes:
    • The Show more button in grid views, no longer overlaps with neighboring column headers.
    • Banners in Data Profiling now display correct names depending on viewed object type (table for tables, View for views etc.)
    • Data Profiling banners now dynamically react to changes in publication policy — eg an Error banner will disappear once a column is unpublished, since that information is now obsolete
    • Banners showing whether a column has been excluded from Data Profiling now remain visible, even if other columns from the same table are published
    • Error banners are now correctly displayed if a column threw an error during profiling

Repository

  • When importing datasources to Dataedo repositories hosted on PostgreSQL, object dependencies are now stored correctly

Connectors:

  • Bugfixes
    • BigQuery: partition columns are now included in the metadata import
    • Snowflake: reworked the logic of simultaneous execution of multiple Data Quality rule instances. Column checks should not disconnect randomly now
    • Databricks: resolved an issue where building lineage for large datasources could result in Databricks API errors
    • Paste Metadata: views and functions removed from the pasted file are now marked as deleted on the next import, the same as tables. Views and Functions are also available in the Types list in Advanced filters; in Desktop the first pasted row is no longer skipped when one of its cells matches a template column name (e.g. a table named Description).
    • Paste Metadata and Interface Tables (including Reporting): values mapped to Title on the Extended properties tab are now imported to objects. Titles entered manually in Dataedo are not overwritten
    • PostgreSQL — removed a bug that caused temporary tables non-existent in the original datasource to appear after an import
    • SQL Server: when Build execution flows is enabled in an import task but Parse SQL is disabled or canceled, the import no longer fails with an Execution flows: ... could not be generated message. It now finishes successfully, with a warning instead
    • SQL Server and Azure SQL: data-based Classification no longer crashes the database engine when sampling temporal tables with hidden period columns (e.g. ValidFrom, ValidTo).
caution

If your Paste Metadata connection uses an Advanced filters rule that includes only Tables and Procedures, all views and functions will be marked as deleted on the first import after the upgrade. To restore them, add Views and Functions to the rule and import again