Keeping data current

Kadaster publishes a new national extract on the 8th of every month, linked from its download page. bagkit does not apply changes to an existing output; it builds a new one from the full extract. See ADR-0001 for why.

curl -O https://service.pdok.nl/kadaster/adressen/atom/v1_0/downloads/lvbag-extract-nl.zip
bagkit convert lvbag-extract-nl.zip --output bag.sqlite --force

How replacing works

• The new output is built beside the destination, so you need disk space for both.
• A file replaces the old one in one atomic rename once it is complete and synced.
• A directory replaces the old one by moving it aside first and removing it afterwards, so for a moment there is no destination, and nobody ever reads a half-written one.
• If the conversion fails, the previous output stays as it was.

Running servers

Stop bagkit serve before replacing its database, and start it again afterwards. On Windows, a database that is open can’t be replaced. On macOS and Linux the replacement succeeds, but the running server keeps reading the old database until it restarts.

SQLite databases with open writers can’t be replaced: close them and checkpoint any -wal or -journal files first.

Which date to use

By default, a conversion selects the voorkomens current today. Pass --date to match the extract’s own date, which the metadata records as extract_date, or to see the BAG as it will stand on a future date that the extract already knows about.