events#
One row per event: a marked moment or interval, such as a manoeuvre, an anomaly, or a phase boundary.
Metadata table, so no dataset_rid predicate is required.
For what an event marks, see Events in the data model.
Columns#
Column |
Type |
Nullable |
Notes |
|---|---|---|---|
|
|
No |
Resource ID. |
|
|
No |
Event name. |
|
|
No |
Free-text description. |
|
|
No |
Event type. |
|
|
No |
When the event began. |
|
|
No |
Length of the event. Zero for an instantaneous event. |
|
|
Yes |
Who created the event. Null for events created automatically. |
|
|
No |
Labels applied to the event. |
|
|
No |
Key-value properties applied to the event. Read a value with |
|
|
Yes |
Data reviews associated with the event. |
Note that events carry duration_seconds rather than an end timestamp, so add the duration to start_time when you need the closing bound.
Warning
The Events table does not filter by dataset. Be cautious when querying this table, as results span your entire workspace and this may result in high query times. A query that reads too many event rows fails, so filter by time or type and add a LIMIT.
Examples#
Most recent events:
SELECT event_rid, name, type, start_time, duration_seconds
FROM events
ORDER BY start_time DESC
LIMIT 20
Instantaneous events only:
SELECT name, type, start_time
FROM events
WHERE duration_seconds = 0
ORDER BY start_time DESC
Events of one type in a window:
SELECT name, start_time, duration_seconds
FROM events
WHERE type = '<event type>'
AND start_time > TIMESTAMP '2026-07-17 00:00:00'
AND start_time < TIMESTAMP '2026-07-18 00:00:00'
ORDER BY start_time
Automatically created events, which have no author:
SELECT name, type, start_time
FROM events
WHERE created_by IS NULL
ORDER BY start_time DESC
LIMIT 20
Next steps#
runs: the periods events sit within.
Writing queries: scoping and query patterns.