Exporting

tql export writes rows as re-importable SQL — insert statements you can read. For a whole database, schema and all, and fast enough for tables in the tens of gigabytes, use Dump, load and sync instead.

From the command line

This is the scriptable way:

tql export                           # asks which connection, then which table
tql export prod orders --limit=1000 --sql=./orders.sql
tql export prod --sql=./prod.sql     # every table, one file
tql export prod --list               # what tables are there
tql export prod --database=shop      # pick the database on a server connection

A connection that names a server rather than a database asks which one first, or takes --database=; without either it stops and says so rather than exporting every schema on the box.

Name a connection and it goes straight through; name nothing and it asks three questions: which connection (most recently used first), which table (the whole database is the first answer), and where to save it, offering the auto-named file in the export directory. Answer that last one with a folder and it names the file for you; --sql skips the question. Piped or with --no-interaction it never asks. It says what it needed and stops with a non-zero exit.

--sql takes a file or a directory; omit it and the file is named automatically in the export directory, after the database, the table and the time (karlm_v2-users-20260922-231302.sql, or karlm_v2-20260922-231302.sql for the whole thing), since the connection's own name says nothing about what is in the file. --limit caps rows per table, which is how you pull a slice of production rather than all of it.

Save one somewhere else, ~/Downloads say, and that becomes the folder the next export is offered in. ui.export_path in the config overrides it, and if the remembered folder is gone tql falls back to its own.

From inside the interface

:export writes what you are looking at to a .sql file of insert statements. On a table that is every row, read in chunks so a large table does not go through memory at once; after a query it is the rows you have loaded.

:export database writes every table in the current database to one .sql file.

Files land in ~/.config/tql/exports (override with ui.export_path), named connection-table-YYYYMMDD-HHMMSS.sql. The status line reports the row count, file size and path.

What it is not

Data only, no schema. Your migrations own the schema; this is for pulling rows from one database into another. A .sql file from tql export can be replayed with tql load, all or nothing, into the same kind of database it came from.