My first post about omni-sql covered the SQL editor, five database engines, and contextual autocomplete. Since v0.2.5, I've added workflows that let me do more with the data after running a query. The current published release is v0.5.1.
Analyze data locally with DuckDB
I can send a query from the main editor to Analyze locally, check the SQL before importing it, and load either the complete result or an explicit sample. I can also start with a CSV or Parquet file. Each source becomes a dataset in a local DuckDB workspace, where I can join data from different connections and files with SQL.
The analysis editor completes dataset and column names and can suggest joins from imported foreign-key metadata. Datasets can be added, renamed, and removed without starting over. Local datasets are kept across app restarts, so an analysis can continue later.
The result grid is only a preview of the analysis. I can page through the result and export the complete query result as CSV or Parquet. Exports also include a .omni.json file with the datasets, source coverage, and ingestion times. That makes it easier to tell later whether an export came from full sources or samples.
Query S3 data in the same workspace
S3 connections now appear in the main IDE. I can browse buckets and supported CSV and Parquet objects, and query them with the SQL editor. Delta and Iceberg tables are supported when the required DuckDB extensions are available. S3 table names and columns also appear in editor completion.
I can import a bounded result from another configured database into a local dataset and join it with S3 data. For S3 setups that use DuckLake, the object browser can discover catalogs backed by a saved PostgreSQL connection or a SQLite/DuckDB file.
A few changes in the day-to-day SQL workflow
- In a
JOIN ... ONclause, autocomplete can suggest predicates based on foreign keys and the aliases in the query. - From a foreign-key value in query results, I can navigate to its referenced row.
-
/catalogopens a searchable list of SQL commands and inserts a template into the editor. - Generic ODBC connections are available alongside the existing adapters; they require a compatible driver installed on the machine.
- PostgreSQL and MySQL tables can show catalog-backed DDL. Other engines show a labeled partial definition built from cached metadata.
- On Windows, signed updates can be installed from inside the app.
omni-sql remains an open-source desktop project. If you work across databases or query files in S3, I'd like to hear which of these workflows would save you time and where they still fall short.
Top comments (0)