fix: skip on-demand loading when no index exists for orderBy + limit queries

.changeset/fix-orderby-limit-no-index.md

@@ -0,0 +1,7 @@
+---
+'@tanstack/db': patch
+---
+
+fix: orderBy + limit queries crash when no index exists
+
+When auto-indexing is disabled (the default), queries with `orderBy` and `limit` where the limit exceeds the available data would crash with "Ordered snapshot was requested but no index was found". The on-demand loader now correctly skips cursor-based loading when no index is available.

packages/db/src/query/compiler/joins.ts

@@ -312,6 +312,14 @@ function processJoin(
 
           if (!loaded) {
             // Snapshot wasn't sent because it could not be loaded from the indexes
+            const collectionId = followRefCollection.id
+            const fieldPath = followRefResult.path.join(`.`)
+            console.warn(
+              `[TanStack DB]${collectionId ? ` [${collectionId}]` : ``} Join requires an index on "${fieldPath}" for efficient loading. ` +
+                `Falling back to loading all data. ` +
+                `Consider creating an index on the collection with collection.createIndex((row) => row.${fieldPath}) ` +
+                `or enable auto-indexing with autoIndex: 'eager' and a defaultIndexType.`,
+            )
             lazySourceSubscription.requestSnapshot()
           }
         }),

packages/db/src/query/compiler/order-by.ts

@@ -191,6 +191,17 @@ export function processOrderBy(
           index = undefined
         }
 
+        if (!index) {
+          const collectionId = followRefCollection.id
+          const fieldPath = followRefResult.path.join(`.`)
+          console.warn(
+            `[TanStack DB]${collectionId ? ` [${collectionId}]` : ``} orderBy with limit requires an index on "${fieldPath}" for efficient lazy loading. ` +
+              `Falling back to loading all data. ` +
+              `Consider creating an index on the collection with collection.createIndex((row) => row.${fieldPath}) ` +
+              `or enable auto-indexing with autoIndex: 'eager' and a defaultIndexType.`,
+          )
+        }
+
         orderByAlias =
           firstOrderByExpression.path.length > 1
             ? String(firstOrderByExpression.path[0])
@@ -292,12 +303,16 @@ export function processOrderBy(
 
       // Set up lazy loading callback to track how much more data is needed
       // This is used by loadMoreIfNeeded to determine if more data should be loaded
-      setSizeCallback = (getSize: () => number) => {
-        optimizableOrderByCollections[targetCollectionId]![`dataNeeded`] =
-          () => {
-            const size = getSize()
-            return Math.max(0, orderByOptimizationInfo!.limit - size)
-          }
+      // Only enable when an index exists — without an index, lazy loading can't work
+      // and all data is loaded eagerly via requestSnapshot instead.
+      if (index) {
+        setSizeCallback = (getSize: () => number) => {
+          optimizableOrderByCollections[targetCollectionId]![`dataNeeded`] =
+            () => {
+              const size = getSize()
+              return Math.max(0, orderByOptimizationInfo!.limit - size)
+            }
+        }
       }
     }
   }

packages/db/src/query/effect.ts

@@ -883,7 +883,7 @@ class EffectPipelineRunner<TRow extends object, TKey extends string | number> {
     for (const [, orderByInfo] of Object.entries(
       this.optimizableOrderByCollections,
     )) {
-      if (!orderByInfo.dataNeeded) continue
+      if (!orderByInfo.dataNeeded || !orderByInfo.index) continue
 
       if (this.pendingOrderedLoadPromise) {
         // Wait for in-flight loads to complete before requesting more

packages/db/src/query/live/collection-subscriber.ts

@@ -332,12 +332,12 @@ export class CollectionSubscriber<
       return true
     }
 
-    const { dataNeeded } = orderByInfo
+    const { dataNeeded, index } = orderByInfo
 
-    if (!dataNeeded) {
-      // dataNeeded is not set when there's no index (e.g., non-ref expression).
-      // In this case, we've already loaded all data via requestSnapshot
-      // and don't need to lazily load more.
+    if (!dataNeeded || !index) {
+      // dataNeeded is not set when there's no index (e.g., non-ref expression
+      // or auto-indexing is disabled). Without an index, lazy loading can't work —
+      // all data was already loaded eagerly via requestSnapshot.
       return true
     }
 

packages/db/tests/query/order-by.test.ts

@@ -561,6 +561,65 @@ function createOrderByTests(autoIndex: `off` | `eager`): void {
         ])
       })
 
+      it(`works with orderBy + limit when limit exceeds available data and no index exists`, async () => {
+        // When limit > number of rows, the topK operator is not full after
+        // the initial snapshot. The on-demand loader must not attempt
+        // cursor-based loading (requestLimitedSnapshot) when there is no index.
+        const collection = createLiveQueryCollection((q) =>
+          q
+            .from({ employees: employeesCollection })
+            .orderBy(({ employees }) => employees.salary, `desc`)
+            .limit(20) // Much larger than the 5 employees
+            .select(({ employees }) => ({
+              id: employees.id,
+              name: employees.name,
+              salary: employees.salary,
+            })),
+        )
+        await collection.preload()
+
+        const results = Array.from(collection.values())
+        expect(results).toHaveLength(5)
+        expect(results.map((r) => r.salary)).toEqual([
+          65_000, 60_000, 55_000, 52_000, 50_000,
+        ])
+      })
+
+      it(`handles delete from topK when limit exceeds available data and no index exists`, async () => {
+        // After a delete, the topK becomes even less full. The on-demand loader
+        // must gracefully handle this without attempting cursor-based loading.
+        const collection = createLiveQueryCollection((q) =>
+          q
+            .from({ employees: employeesCollection })
+            .orderBy(({ employees }) => employees.salary, `desc`)
+            .limit(20)
+            .select(({ employees }) => ({
+              id: employees.id,
+              name: employees.name,
+              salary: employees.salary,
+            })),
+        )
+        await collection.preload()
+
+        const results = Array.from(collection.values())
+        expect(results).toHaveLength(5)
+
+        // Delete Diana (highest salary) — topK shrinks, triggering loadMoreIfNeeded
+        const dianaData = employeeData.find((e) => e.id === 4)!
+        employeesCollection.utils.begin()
+        employeesCollection.utils.write({
+          type: `delete`,
+          value: dianaData,
+        })
+        employeesCollection.utils.commit()
+
+        const newResults = Array.from(collection.values())
+        expect(newResults).toHaveLength(4)
+        expect(newResults.map((r) => r.salary)).toEqual([
+          60_000, 55_000, 52_000, 50_000,
+        ])
+      })
+
       itWhenAutoIndexEager(
         `applies incremental insert of a new row inside the topK but after max sent value correctly`,
         async () => {