Occurrences

The durable timeline of every occurrence edge, observed by a run's judge or by the outside-run Always watcher, with the frozen condition behind each one.

View as Markdown

The Occurrences tab of Events (/events?view=occurrences) is the durable timeline of what the things under test did: one row per occurrence edge, whether a run's judge observed it or the outside-run Always watcher did. Seeing it takes View events; exporting takes Export/import histories. Occurrences are defined per model under Process > Setup > Occurrences; this tab reads their history and changes nothing.

It follows the shared conduct of the three timelines on Events: 500 rows a page with a pager under them, a date range, a debounced search, live refresh, one detail panel, CSV and JSON export of the whole period.

What one row is

An occurrence is a Boolean condition over a model's channels with an on-delay and an off-delay. A row is one of four transitions of one observation session:

Transition	Meaning
Activated	The condition rose and held for the on-delay.
Cleared	The condition fell and held for the off-delay.
Unavailable	One or more channels the condition references stopped delivering a value. The judge holds its last confirmed condition and never fabricates a clear.
Restored	Every referenced channel delivers again; the condition is evaluated afresh and its delay restarts.

Each row also says in which scope it was observed and why its session began:

Field	Values	Meaning
Scope	Run, Outside run	Run: a run's own judge, armed by the procedure; the edge is also written into that run's file. Outside run: the Always watcher that reads the unit's bindings between runs.
Reason	Run started, Run resumed, Watcher started, Outside-run observation resumed, Definition changed	Why a fresh observation session began: a run started; a run was resumed after the app restarted; the Always watcher started for this occurrence and unit; the unit returned to outside-run observation after a run; or the condition, delays, referenced channels, enabled state or armed mode were edited.

A new session starts from nothing: no condition, availability or debounce state is carried across an unobserved gap, so an already-true condition can produce another Activated when a run starts or the app comes back, and there is no synthetic "session started" row in this timeline. The run's own file records the session boundary even when no transition follows; Histories and reports use it to classify an older open activation as Interrupted. That marker stays out of this timeline and raises no notification. Display-only edits (name, message, severity) keep the session and only change later snapshots.

Search and filter

Control	Values	What it does
Search	Placeholder "Search name, message, model, unit or procedure"	Contains-match on the occurrence's name, its message, the model name, the unit name and the procedure name, as snapshotted at the time.
Severity	All severities (default), Critical, Warning, Info	Keeps only transitions of occurrences that carried that severity when they happened.
Refresh		Re-reads the journal; tooltip "Refresh occurrences".
Export	Export as CSV…, Export as JSON…	Writes the rows currently listed. Drawn only with Export/import histories; greyed while the list is empty.

The rows

Part of the row	Content
Time	Date and time the judge observed the edge (for a run, the judged sample's own timestamp).
Dot	The severity's tone: critical for Critical, caution for Warning, neutral for Info.
Name	The occurrence's name at the time.
Summary	"Transition · obs N (Reason) · Model / Unit · context", where N is the edge's sequence within its session and the context is "Run · " followed by the procedure's name (or just "Run" when no procedure name was recorded) or "Outside run". Example: "Activated · obs 1 (Run started) · Furnace / Line 2 · Run · Bake 180".
Level	The severity word.

The message a row carries is the occurrence's own message for Activated and Cleared (empty when the definition has none, shown as "No message captured." in the panel), and for the availability edges "Signal unavailable:" or "Signal restored:" followed by the names of the affected channels, separated by commas.

The detail panel

Selecting a row opens the panel with the severity badge and the name, the full timestamp, the summary line, the message, and four cards:

Card	Field	Content
EVENT	Level	The severity.
EVENT	Event ID	The journal's identity of this edge. For a run observation it is the same id the run file holds.
OCCURRENCE SNAPSHOT	Transition	Activated, Cleared, Unavailable or Restored.
OCCURRENCE SNAPSHOT	Model	The model's name at the time.
OCCURRENCE SNAPSHOT	Unit	The unit's name at the time.
OCCURRENCE SNAPSHOT	Scope	Run or Outside run.
OCCURRENCE SNAPSHOT	Procedure	Only for the Run scope: the procedure's name, or a dash when none was recorded.
OCCURRENCE SNAPSHOT	Run ID	Only for the Run scope: the run's id, the same one Histories shows.
OCCURRENCE SNAPSHOT	Model ID, Unit ID, Occurrence ID	The stable ids behind the names, unchanged by a rename.
OBSERVATION SESSION	Journal sequence	The row's insertion number in the journal, which orders rows that share a timestamp.
OBSERVATION SESSION	Observation ID	The identity of the session that observed the edge; every edge of one session shares it.
OBSERVATION SESSION	Transition sequence	The edge's one-based order within the session. A gap means an edge the journal could not persist.
OBSERVATION SESSION	Started	When the session began, as a full timestamp.
OBSERVATION SESSION	Reason	Run started, Run resumed, Watcher started, Outside-run observation resumed or Definition changed.
FROZEN CONFIGURATION	Condition	The JavaScript condition the session actually judged.
FROZEN CONFIGURATION	On / off delay	"x ms / y ms", the debounce the session used.
FROZEN CONFIGURATION	Referenced channel IDs	The ids of the channels that gated the condition, separated by semicolons.

Export columns

CSV and JSON carry the same fields, in this order: Timestamp (the CSV header adds the time reference, "Timestamp (UTC-03:00)"), JournalSequence, Severity, Name, Transition, Scope, Model, ModelId, Unit, UnitId, Procedure, RunId, Message, EventId, OccurrenceId, ObservationId, ObservationSequence, ObservationStartedAt, ObservationReason, Condition, OnDelayMs, OffDelayMs and ReferencedChannelIds (semicolon-separated in the CSV, an array in the JSON). The JSON is an envelope with TimeZone and Events. File names are occurrence-history-<yyyyMMdd-HHmmss>.csv and .json. The CSV follows the station's decimal-separator preference like every other export: with point decimals the list separator is ,, with comma decimals it is ;. A cell beginning with =, +, - or @ is prefixed with an apostrophe so a spreadsheet reads it as text.

Where else an occurrence is recorded

A run observation is written to the run's own file first, then copied here with the same event id and the run context; if the copy fails, the run file still holds the edge and Console carries an Error line saying so. An outside-run observation exists only here. Every edge also leaves a Console line under the Process area: an activation at the level of its severity (Critical as an error, Warning as a warning, Info as information), a signal loss at that same level, a clear and a restoration as information. An activation additionally raises an in-window notification titled "Occurrence · " followed by the unit's name and, for a Warning or Critical activation observed during a run, an OS notification while the window is hidden or minimized.

Empty states and the status line

"No occurrences yet" with "Occurrence transitions from runs and outside-run Always monitoring will appear here."
"No occurrences match these filters" with "Adjust the severity or search terms to widen this history."
The status line reads "N shown · page 1", then " · filtered" when a search, a severity or a date range is in force; on the right, "1 selected" and "Updated hh:mm:ss".

What the Occurrences tab does not do

It does not define, arm or edit an occurrence (Process > Setup > Occurrences).
It does not pair edges into episodes with durations: that derived view, with its Interrupted and Incomplete classifications, is in Histories for a run and on the unit's own panels.
It has no unit, run, scope or date filter; search by the unit, model or procedure name, or open the run in Histories.
It shows no live state: whether an occurrence is active right now is read on the unit's panel, not here.