Change log

What changed in each release, and why it mattered.

Every release supplies a machine-readable diff alongside the data files, keyed on address_id, so you can reconcile against your own copy without comparing full extracts.

2026 Q3 — launch release

First public release. Address points and derived postal code points, national coverage from the Statistics Canada National Address Register, delivered in GeoPackage and CSV, WGS84.

Nothing to diff against yet. Subsequent releases will list added, retired, moved and attribute-changed records here.

How to read a diff

The diff file is CSV, one row per changed record: address_id, change_flag, previous_release, and the changed columns with before and after values. Records absent from the diff are unchanged and can be skipped entirely, which for most municipalities is the large majority of rows.

Four flags:

added — a new address appeared in the source
retired — an address was removed by its source
moved — the position changed; the address did not
attribute_change — a component or administrative field changed

Why moved is not added

As better municipal sources are layered onto the NAR baseline, existing addresses will shift position, sometimes by tens of metres, without being new. They are flagged moved.

If your workflow caches coordinates rather than joining on address_id, those releases will look like churn. They are not. Anything counting growth should filter to added only.

Postal point changes

Postal points are recomputed each release rather than diffed. A point moves when the addresses under it change, which happens routinely as the address base improves. Compare address_count between releases rather than position — a code whose count jumps is a code whose coverage improved.