Migrating from ESRI Attribute Rules

ESRI's Attribute Rules and GISPulse's triggers.yaml solve the same problem (declarative reactions on edits) with different vocabularies. This page walks you through the renames and the cases that need rethinking when you bring an Attribute Rules file across.

TL;DR

ESRI concept	GISPulse equivalent	Notes
type: constraint	kind: constraint (alias of validation)	use validate: rule with mode: tag to land in the row
type: calculation	kind: calculation (alias of trigger)	trigger with a set_field action
type: validation	kind: validation (native)	trigger with a validate: rule
triggeringEvents: ["Insert"]	when: [INSERT]	identical semantics
triggeringEvents: ["Update"]	when: [UPDATE] (catch-all)	use UPDATE_GEOM / UPDATE_ATTR for granular dispatch
triggeringEvents: ["Delete"]	when: [DELETE]	identical
evaluationOrder: 1	not yet — order = YAML order	tracked in v1.7+
Arcade expression	DSL geom functions + arithmetic	see DSL geom functions
subtype: 1 filter	predicate: "subtype = 1"	DSL predicate

kind: aliases

The kind: field on a trigger is optional — if omitted, the trigger defaults to the GISPulse-native trigger kind. ESRI users can keep their vocabulary by spelling out the alias:

triggers:
  - name: parcels_constraint_min_surface
    kind: constraint           # alias of "validation"
    table: parcels
    when: [INSERT, UPDATE_GEOM]
    actions: [...]

Alias	Maps to	When to use it
constraint	validation	The rule rejects or tags rows that fail a check (ESRI "constraint" type)
calculation	trigger	The rule writes a derived attribute (ESRI "calculation" type)
validation	validation	Native — same as ESRI
trigger	trigger	Native default for GISPulse

The alias is metadata only today: it does not change runtime behaviour. The GISPulse runtime decides what to do based on the trigger's actions: (and on a top-level validate: block); the kind: label is there to help operators read the YAML and to keep your migration diff small.

Triggering events: granular UPDATE

ESRI fires Update whenever any field changes. GISPulse v1.6.0 introduces UPDATE_GEOM and UPDATE_ATTR so you can react only to geometry edits or only to attribute edits:

when: [UPDATE_GEOM]   # only when the geometry actually changed
when: [UPDATE_ATTR]   # only when only attributes changed
when: [UPDATE]        # catch-all (= UPDATE_GEOM ∪ UPDATE_ATTR), v1.5.x-compatible

The runtime resolves a coarse UPDATE change-log row to one of the granular variants by reading the geom_changed flag the GeoPackage trigger captured at edit time. Existing v1.5.x configs keep working because UPDATE is still accepted and expanded internally.

BULK is a fourth value used by the watcher when many rows arrive at once (paste from QGIS, CSV import). Triggers that need to run a different code path on bulk events can declare when: [BULK].

Arcade vs the GISPulse DSL

The GISPulse DSL is intentionally smaller than Arcade. You get geometry helpers (geom_area_m2, geom_is_valid, …), arithmetic, and boolean/comparison operators inside validate: rules. You do not get IIf, lookups by OID, schema mutations, or arbitrary script execution — those land you in run_sql territory, where the YAML ships a hand-written SQL fragment that goes through your security review.

Cross-source lookups (layer_lookup, geom_within, geom_overlaps_any) are coming in the v1.6.x line; track #122 and #124 for status.

What does not translate

Feature	Status	Workaround
Subtypes	not modelled	use predicate: to filter rows
Domains	not modelled	use a validate: rule that compares against a list
Editor tracking fields	not auto-managed	add set_field actions for last_edited_user etc.
Attribute Rules error codes	not modelled	use validate: id: + message: + a tag column
Asynchronous batch evaluation	not the GISPulse model	rules fire immediately on the change-log tail

Checklist for a clean migration

1. Export your Attribute Rules to JSON via arcpy.management.ExportAttributeRules.
2. For each rule, decide whether it belongs in triggers: (mutates the row, ESRI calculation) or validate: (just checks, ESRI constraint or validation).
3. Translate the Arcade expression: simple math + geom checks map directly; anything Arcade-only goes into a run_sql action.
4. Set kind: to the ESRI vocabulary if it helps your team read the YAML — it is purely cosmetic.
5. Run gispulse triggers run --dry to see the compile output before wiring it to a live dataset.