ArchitectureA supervised JVM-class runtime — OLTP on seven engines, OLAP on three. AI-native, MCP-native, observable as plain SQL.Read the architecture
Está viendo la edición Perú. Está viendo la edición Colombia. You're viewing the Pakistan edition. Cambiar a la edición global →Cambiar a la edición global →Switch to the global edition →

Exports come out flat: every machine format undoes a report's grouping

Excel, CSV, SQL, JSON, XML and Parquet writers restore repeated values and drop subtotal rows by default, so a machine format is data rather than a layout. Parquet joins the export menu with its codec and column choice.

A grouped report clears a value that repeats the row above and inserts a subtotal after each group. On screen that is a layout. In a CSV it is corrupt data, and until this release only Excel knew the difference.

Flat by default

  • Every writer Excel, CSV, SQL, JSON and XML undo the grouping unless asked to keep it: subtotal rows left out, a cleared cell carrying the value it keeps.
  • Parquet too The Parquet writer honours the same switch, and the MIME detector recognises a Parquet file.
  • In the export dialog An ungroup toggle on every format a program reads, on by default; PDF, HTML and text keep their groups.

Parquet in the menu

  • Between JSON and text, gated by access right, with a page codec of SNAPPY, GZIP or none and columns to exclude, always flat. Labels in eight locales.
  • From scripts The result-set writer and reader export and import Parquet to a blob or a file, tested against files written by pyarrow.
  • Unload with a ceiling A table unload stops at a row count, applied while reading, and the capped file is still a complete document.

The rule is one sentence: a format a program reads gets data, a format a person reads gets the layout.