Skip to main content
A Batch Calculation rule checks the values a model worked out for a batch. It reads the node values the calculation already stored and never recalculates the model, so whatever the batch shows on its Calculations tab is what the rule judges. To write one, see Create a data rule and the {{node.<name>}} variables in Rule Expressions.

​
When the rule runs

A Batch Calculation rule runs automatically each time a batch’s calculation finishes with every output node computed, which the batch shows as Updated. That happens when you generate a batch, and again when an edit to a data point it used recalculates it. A rule’s correction does not recalculate batches already generated; see Correcting many values at once.
  • Incomplete and Error batches are skipped by the automatic run. Saving the rule, or Evaluate a period on its Runs tab, checks them too.
  • Results arrive within seconds of the calculation finishing. The alert on the batch page can take a few minutes longer.
  • The rule’s effective period is matched against the batch’s start date, in your account’s time zone.
  • A rule set to Run only on demand never runs automatically. It runs only over a period you start.

​
A node with no value passes

A comparison against a node that has no value is not true, so the rule records a pass. {{node.moisture_deduction}} > 0.2 stays silent on a batch where moisture_deduction never computed. To catch an empty node, test for it:
On a child node this fires automatically, as long as the batch’s output nodes still computed. A batch whose output node is empty is Incomplete, which the automatic run skips, so IS MISSING on an output node fires only when you save the rule or evaluate a period. A calculated node combines the child nodes that have values and skips the ones that don’t. A total can therefore come out lower than expected without being empty, and IS MISSING on the total will not catch it. Test the child node itself.

​
Totals belong in the model

A Batch Calculation rule cannot use a WITHIN window, a capture-rate function, or the substitute action. Those are for Data Points rules. To check a monthly or cumulative figure, build it into the model as a node and reference that node.

​
Model versions

A Batch Calculation rule follows its model across versions. Activating a new version needs no change to the rule: it keeps checking batches from every version of the model, and its results list shows them all.
  • The rule editor, autocomplete and the checks on save read the model’s active version. A rule naming a node the active version no longer has is rejected when you next save it.
  • On batches calculated with an older version that lacks a node the rule names, the node has no value, so the rule passes.
  • Deleting the version the rule was created against does not delete the rule, or its results on other versions’ batches.

​
How alerts reach a batch

A batch carries alerts from its own calculation and from the upstream batches and data points that fed it, so one failing result upstream shows on every batch built from it, and dismissing it clears it from all of them. A banner on the batch page counts them, N alerts triggered on this batch. See Alerts on a batch for which tab lists each kind.

​
Reported batches are not re-checked

Saving a rule, editing it, and Evaluate a period all skip batches included in a report that is Completed or Verified, so a change to a rule never re-judges figures you have already filed. A batch page shows alerts only from a rule’s current version, so after an edit, a reported batch no longer shows that rule’s earlier alerts.

​
Corrections waiting for approval

A batch’s calculation cannot use a data point whose correction is waiting for approval. See Approve corrections.