Arcade is built to optimize queries to the server where the actual data is stored. So when you call FeaturesetByPortalItem(), the fs isn't actually loaded into your local RAM, it just creates a connection to the fs on the server. This is done to then optimize Filter(), Intersects(), First() and so on to send as few as possible query requests to the server.
This approach doesn't work with calling these functions in a loop. You're basically bombarding the server with hundreds to thousands filter requests (one for each feature in the fs), which slows the script down to a crawl. The developers are aware of this, but I wouldn't hold my breath waiting for a solution, it's basically a known limitation at this point. The intended approach is to use a joined layer.
But using a joined layer isn't always possible or sensible. For these cases, @jcarlson wrote a nifty function that actually loads the fs into your local RAM, where repeated Filter() calls are much faster: Improving Expression Performance: A Custom Function
In your case, we don't actually need that function, but the principle still applies: Copy the data from the server and do your filtering locally. In this case, we can use a simple dictionary (lines 21, 22, 34):
// only load the fields you need
var portal = Portal("https://www.arcgis.com/");
var polyfs = FeatureSetByPortalItem(
portal,
"4dbbad3d6f694e0ebc7c3b4132ea34df",
0,
["HydroID", "DPS_Region"],
false
);
var tablefs = FeatureSetByPortalItem(
portal,
"4dbbad3d6f694e0ebc7c3b4132ea34df",
6,
["FeatureID", "ModelID", "AddressCount", "MAX_TSTime"],
false
);
// repeatedly sending Filter requests to the server slows the script down to a crawl for larger Featuresets.
// It's best to create a copy in local RAM. In this case, we can just use a dictionary {HydroID: DPS_Region}.
// If you need Filter(), use the Memorize() function defined in the blog post above.
var dps_regions = Dictionary()
for(var p in polyfs) { dps_regions[Text(p.HydroID)] = p.DPS_Region }
// Create empty features array and feat object
var features = [];
var feat;
// Populate Feature Array
for (var t in tablefs) {
var tableID = Text(t["FeatureID"])
feat = {
attributes: {
FeatureID: tableID,
Name: dps_regions[tableID],
ModelID: t["ModelID"],
AddressCount: t["AddressCount"],
MAX_TSTime: t["MAX_TSTime"],
}
}
Push(features, feat)
}
var joinedDict = {
fields: [
{ name: "FeatureID", type: "esriFieldTypeString" },
{ name: "Name", type: "esriFieldTypeString" },
{ name: "ModelID", type: "esriFieldTypeInteger" },
{ name: "AddressCount", type: "esriFieldTypeInteger" },
{ name: "MAX_TSTime", type: "esriFieldTypeString" },
],
'geometryType': '',
'features':features
};
// Return dictionary cast as a feature set
return FeatureSet(joinedDict);
This script executes more than 2 times faster than your original script. I expect it will be even more noticeable for large Featuresets.
Take a look at these related Ideas and maybe lend your support. Hopefully, they will be implemented someday...
Arcade FeatureSet Functions should have the option to output to RAM
Arcade: Add a function that quickly converts a Featureset into an Array
Arcade: Add Function to Join Featuresets
Arcade: Add a function to merge multiple Featuresets
Have a great day!
Johannes
