pbtsdb: PocketBase TanStack Database Integration

Type-safe PocketBase integration with TanStack Query and TanStack DB

A TypeScript library that seamlessly integrates PocketBase with TanStack Query and TanStack DB, providing:

• 🔥 Real-time subscriptions with automatic synchronization
• 🎯 Full TypeScript type safety for queries and relations
• ⚡ Reactive collections with TanStack DB
• 🔄 Automatic caching via TanStack Query
• ✨ Optimistic mutations with insert/update/delete support
• 🎨 React hooks for easy component integration
• 🔗 Type-safe joins and relation expansion

Table of Contents

• Installation
• Quick Start
• Core Concepts
• API Reference
  • createCollection()
  • React Integration
  • Subscriptions
• Usage Examples
  • Basic Queries
  • Filtering and Sorting
  • Relations and Joins
  • Includes (Nested Subqueries)
  • Real-time Updates
  • Mutations
• TypeScript
• Best Practices

Installation

npm install pbtsdb pocketbase @tanstack/react-query @tanstack/react-db @tanstack/query-db-collection

Peer Dependencies

• pocketbase >= 0.22.0
• @tanstack/react-query >= 5.0.0
• @tanstack/react-db >= 0.1.0
• @tanstack/query-db-collection >= 1.0.0
• react >= 18.0.0
• react-dom >= 18.0.0

All peer dependencies use minimum version constraints - newer versions should work.

Quick Start

Let's build a real-world blog with posts, authors, and comments using pbtsdb.

1. Define Your Schema

First, generate your types from PocketBase. Install pocketbase-schema-generator as a PocketBase hook to auto-generate types on schema changes.

// schema.ts - Auto-generated from PocketBase
interface Post {
    id: string;
    title: string;
    content: string;
    author: string;      // FK to users
    published: boolean;
    created: string;
    updated: string;
}

interface User {
    id: string;
    username: string;
    email: string;
    avatar?: string;
    created: string;
    updated: string;
}

interface Comment {
    id: string;
    post: string;        // FK to posts
    author: string;      // FK to users
    text: string;
    created: string;
    updated: string;
}

// Schema declaration for pbtsdb
type BlogSchema = {
    posts: {
        type: Post;
        relations: { author?: User };
    };
    users: {
        type: User;
        relations: {};
    };
    comments: {
        type: Comment;
        relations: {
            post?: Post;
            author?: User;
        };
    };
}

2. Set Up Your App

// app.tsx
import PocketBase from 'pocketbase';
import { QueryClient, QueryClientProvider } from '@tanstack/react-query';
import { createCollection, createReactProvider } from 'pbtsdb';

const pb = new PocketBase('http://localhost:8090');
const queryClient = new QueryClient({
    defaultOptions: {
        queries: { staleTime: 60_000 } // Cache for 1 minute
    }
});

// Create collections with automatic type inference
const c = createCollection<BlogSchema>(pb, queryClient);
export const { Provider, useStore } = createReactProvider({
    posts: c('posts', {
        omitOnInsert: ['created', 'updated'] as const
    }),
    users: c('users', {}),
    comments: c('comments', {
        omitOnInsert: ['created', 'updated'] as const
    })
});

export function App() {
    return (
        <QueryClientProvider client={queryClient}>
            <Provider>
                <BlogDashboard />
            </Provider>
        </QueryClientProvider>
    );
}

3. Build Your Components

// BlogDashboard.tsx
import { useLiveQuery } from '@tanstack/react-db';
import { useStore } from './app';

export function BlogDashboard() {
    const [posts] = useStore('posts');

    const { data: allPosts, isLoading } = useLiveQuery((q) =>
        q.from({ posts })
            .orderBy(({ posts }) => posts.created, 'desc')
    );

    if (isLoading) return <div>Loading posts...</div>;

    return (
        <div>
            <h1>Blog Posts</h1>
            {allPosts?.map(post => (
                <article key={post.id}>
                    <h2>{post.title}</h2>
                    <p>{post.content}</p>
                    {/* Expanded author is fully typed! */}
                    <small>By {post.expand?.author?.username}</small>
                </article>
            ))}
        </div>
    );
}

4. Add Real-time Comments

// PostWithComments.tsx
import { useLiveQuery } from '@tanstack/react-db';
import { eq } from '@tanstack/db';
import { useStore } from './app';
import { newRecordId } from 'pbtsdb';

export function PostWithComments({ postId }: { postId: string }) {
    const [comments, posts] = useStore('comments', 'posts');

    // Real-time comments for this post
    const { data: postComments } = useLiveQuery((q) =>
        q.from({ comments })
            .where(({ comments }) => eq(comments.post, postId))
            .orderBy(({ comments }) => comments.created, 'desc')
    );

    const handleAddComment = (text: string, authorId: string) => {
        comments.insert({
            id: newRecordId(),
            post: postId,
            author: authorId,
            text
        });
        // Comment appears instantly (optimistic), syncs to PocketBase in background
    };

    return (
        <div>
            <h3>Comments ({postComments?.length || 0})</h3>
            {postComments?.map(comment => (
                <div key={comment.id}>
                    <strong>{comment.expand?.author?.username}:</strong>
                    <p>{comment.text}</p>
                </div>
            ))}
            <CommentForm onSubmit={handleAddComment} />
        </div>
    );
}

That's it! You now have a real-time blog with:

• ✅ Type-safe queries
• ✅ Automatic real-time updates
• ✅ Optimistic mutations
• ✅ Expanded relations

Core Concepts

Collections

Collections are reactive data stores that automatically sync with PocketBase:

// Create a collection using the curried API
const c = createCollection<MySchema>(pb, queryClient);
const booksCollection = c('books', {});

// Collections automatically:
// - Fetch data from PocketBase
// - Subscribe to real-time updates
// - Update React components when data changes
// - Cache data via TanStack Query

Real-time Subscriptions

Collections manage subscriptions automatically based on query lifecycle:

// Collections are lazy - no subscription until queried
const c = createCollection<MySchema>(pb, queryClient);
const booksCollection = c('books', {});

// Subscription starts automatically when query becomes active
const { data } = useLiveQuery((q) =>
    q.from({ books: booksCollection })
);
// ✅ Subscribed to changes while component is mounted
// ✅ Unsubscribes automatically when component unmounts

Subscription Lifecycle:

• Lazy: No subscription starts until the first useLiveQuery using the collection renders
• Automatic: Subscription starts when first subscriber mounts, stops when last subscriber unmounts
• Shared: Multiple components using the same collection share one subscription
• No manual control needed: The collection handles all subscription management internally

Mutations and Refetch

By default, pbtsdb does not refetch the entire collection after a successful insert, update, or delete. The realtime subscription delivers server-confirmed rows — including server-assigned fields like id, created, updated, and any values rewritten by PocketBase hooks — and TanStack DB's optimistic write keeps the UI consistent in the meantime. The post-mutation refetch is therefore redundant.

Set refetchOnMutation: true to opt back into the previous behavior — for example, when realtime is unreliable in your environment, or when a server-side hook produces a field you must read synchronously after the mutation resolves:

const collection = c('books', {
    refetchOnMutation: true,
});

The option only affects the built-in default handlers. If you supply your own onInsert, onUpdate, or onDelete, you control its return value yourself — return { refetch: false } (or omit a return) to skip the refetch, return { refetch: true } to force one.

Type Safety

Full TypeScript support with compile-time type checking:

const { data } = useLiveQuery((q) =>
    q.from({ books: booksCollection })
);

// TypeScript knows:
// - data[0].title is a string
// - data[0].genre is 'Fiction' | 'Non-Fiction' | 'Science Fiction'
// - data[0].author is a string (FK)

API Reference

createCollection()

The main function for creating type-safe collections. Uses a curried API for better type inference.

const c = createCollection<Schema>(
    pb: PocketBase,
    queryClient: QueryClient,
    factoryOptions?: CreateCollectionFactoryOptions
);
const collection = c(collectionName: string, options?: CreateCollectionOptions);

Parameters:

• pb - PocketBase instance
• queryClient - TanStack Query QueryClient instance
• factoryOptions - Optional configuration applied to every collection this factory builds (see Subscription Options)
• collectionName - Name of the PocketBase collection
• options - Optional configuration

Options:

• expand?: Record<string, Collection> - Relations to auto-expand and auto-upsert on every fetch
• omitOnInsert?: readonly string[] - Fields to make optional during insert (e.g., ['created', 'updated'] as const)
• syncMode?: 'eager' | 'on-demand' - Data fetching strategy (default: 'eager')
• onInsert?: InsertMutationFn | false - Custom insert handler or false to disable
• onUpdate?: UpdateMutationFn | false - Custom update handler or false to disable
• onDelete?: DeleteMutationFn | false - Custom delete handler or false to disable
• ignoreAutoCancellation?: boolean - Ignore PocketBase auto-cancellation errors (default: true)
• collectionOptions?: object - Additional TanStack DB collection options passed through directly (see Collection Options Passthrough)

Returns: Fully-typed Collection instance with subscription capabilities

Examples:

Basic collection (lazy, subscribes automatically on first query):

const c = createCollection<MySchema>(pb, queryClient);
const booksCollection = c('books', {});

With auto-expand relations:

const c = createCollection<MySchema>(pb, queryClient);
const authorsCollection = c('authors', {});
const booksCollection = c('books', {
    expand: {
        author: authorsCollection  // Auto-expand and auto-upsert
    }
});

// Expand is automatic on every fetch
const { data } = useLiveQuery((q) => q.from({ books: booksCollection }));

// Expanded records auto-inserted into authorsCollection

Collection Options Passthrough

Pass any TanStack DB BaseCollectionConfig option directly via collectionOptions. This is useful for configuring indexing, garbage collection, and other collection-level settings:

import { BasicIndex } from 'pbtsdb';

const c = createCollection<MySchema>(pb, queryClient);
const booksCollection = c('books', {
    collectionOptions: {
        autoIndex: 'eager',
        defaultIndexType: BasicIndex,
        gcTime: 60000,       // 1 minute GC
        startSync: true,     // Start syncing immediately
    }
});

The following fields are managed by pbtsdb and excluded from collectionOptions: getKey, syncMode, onInsert, onUpdate, onDelete, schema.

React Integration

createReactProvider()

Creates a React Provider and useStore hook from a collections map.

const { Provider, useStore } = createReactProvider(collections: CollectionsMap);

Parameters:

• collections - Object mapping keys to Collection instances

Returns:

• Provider - React Context Provider component
• useStore - Hook to access collections (variadic args, returns typed tuple)

Example:

import { createCollection, createReactProvider } from 'pbtsdb';

const c = createCollection<MySchema>(pb, queryClient);
const collections = {
    authors: c('authors', {}),
    books: c('books', {
        omitOnInsert: ['created', 'updated'] as const
    }),
};

const { Provider, useStore } = createReactProvider(collections);

// Wrap your app
<Provider>
    <App />
</Provider>

With custom collection key:

const collections = {
    myBooks: c('books', {})  // Key 'myBooks', PocketBase collection 'books'
};

const { Provider, useStore } = createReactProvider(collections);

// Access via custom key
const [myBooks] = useStore('myBooks');

useStore()

Access collections from the provider. Uses variadic arguments and returns a typed tuple.

Single collection:

const [collection] = useStore('key')

Multiple collections:

const [col1, col2, col3] = useStore('key1', 'key2', 'key3')

Examples:

function BooksList() {
    const [books] = useStore('books');  // ✅ Typed automatically!

    const { data } = useLiveQuery((q) =>
        q.from({ books })
    );

    return <div>{/* ... */}</div>;
}

function BooksWithAuthors() {
    const [books, authors] = useStore('books', 'authors');  // ✅ Variadic!

    const { data } = useLiveQuery((q) =>
        q.from({ book: books })
            .join(
                { author: authors },
                ({ book, author }) => eq(book.author, author.id),
                'left'
            )
    );

    return <div>{/* ... */}</div>;
}

Subscriptions

Collections manage real-time subscriptions to PocketBase automatically. No manual subscription management is needed for normal usage.

Automatic Subscription Lifecycle

// Subscriptions start automatically when useLiveQuery renders
function MyComponent() {
    const [books] = useStore('books');
    const { data } = useLiveQuery((q) => q.from({ books }));
    // ✅ Subscription active while this component is mounted
    // ✅ Automatically stops when component unmounts
}

isSubscribed()

Check if a collection has an active subscription.

const isSubbed = collection.isSubscribed(); // boolean

waitForSubscription()

Wait for subscription to be established (useful in tests).

await collection.waitForSubscription(); // Wait with default 5s timeout
await collection.waitForSubscription(10000); // Wait with custom timeout (ms)

Subscription Options

Pass subscribeOptions as the third createCollection argument to attach extra options — headers, filter, expand, fields — to every real-time subscription the factory creates.

const c = createCollection<Schema>(pb, queryClient, {
    subscribeOptions: () => {
        const token = getShareToken();
        return token ? { headers: { 'X-Share-Token': token } } : undefined;
    },
});

It is a getter, not a static object, and that matters. Subscriptions restart on reconnect and whenever the subscriber count rises from zero, so a value read once would go stale exactly when it counts — for example when a visitor authenticates mid-session and should stop presenting an anonymous token.

Returning undefined subscribes with no extra options, identical to omitting the option entirely.

The common use is authorizing anonymous access. PocketBase snakecases header names the same way for real-time as for REST, so X-Share-Token is readable in a collection rule as @request.headers.x_share_token — a single rule covers both transports:

@request.auth.id != "" || @request.headers.x_share_token = "..."

Setting the matching header on REST requests remains the application's job, via pb.beforeSend. Requires pocketbase >= 0.22.0.

Utility Functions

newRecordId()

Generate a PocketBase-compatible record ID (15-character alphanumeric string).

import { newRecordId } from 'pbtsdb';

const id = newRecordId(); // "a1b2c3d4e5f6g7h"

// Use when creating records
const newBook = {
    id: newRecordId(),
    title: 'New Book',
    // ... other fields
};

booksCollection.insert(newBook);

Returns: string - 15-character lowercase alphanumeric ID

Re-exported TanStack DB Utilities

pbtsdb re-exports commonly used TanStack DB utilities so you don't need to depend on @tanstack/db directly:

import {
    // Includes helpers
    toArray,
    createEffect,

    // Index types (for collectionOptions.defaultIndexType)
    BasicIndex,
    BTreeIndex,
    ReverseIndex,

    // Types
    type DeltaEvent,
    type DeltaType,
    type EffectConfig,
    type EffectContext,
    type IndexConstructor,
} from 'pbtsdb';

Usage Examples

Example 1: Task Manager with Filtering

// TaskBoard.tsx
import { useLiveQuery } from '@tanstack/react-db';
import { eq, and } from '@tanstack/db';
import { useStore } from './app';

export function TaskBoard({ userId }: { userId: string }) {
    const [tasks] = useStore('tasks');

    // Filter tasks by assignee and status - updates in real-time
    const { data: myTasks } = useLiveQuery((q) =>
        q.from({ tasks })
            .where(({ tasks }) => and(eq(tasks.assignee, userId), eq(tasks.status, 'in_progress')))
            .orderBy(({ tasks }) => tasks.due_date, 'asc')
    );

    const handleComplete = (taskId: string) => {
        tasks.update(taskId, (draft) => { draft.status = 'done'; });
    };

    return (
        <div>
            <h2>My Tasks ({myTasks?.length || 0})</h2>
            {myTasks?.map(task => (
                <div key={task.id}>
                    {task.title}
                    <button onClick={() => handleComplete(task.id)}>Complete</button>
                </div>
            ))}
        </div>
    );
}

Example 2: E-commerce Product Catalog with Filtering

// ProductCatalog.tsx
import { useLiveQuery } from '@tanstack/react-db';
import { and, gte, lte } from '@tanstack/db';
import { useStore } from './app';

export function ProductCatalog() {
    const [products] = useStore('products');
    const [category, setCategory] = useState<string | null>(null);
    const [maxPrice, setMaxPrice] = useState(1000);

    // Dynamic filtering - updates reactively
    const { data: filteredProducts } = useLiveQuery((q) => {
        let query = q.from({ products })
            .where(({ products }) => and(
                products.in_stock === true,
                lte(products.price, maxPrice)
            ));

        if (category) {
            query = query.where(({ products }) => products.category === category);
        }

        return query.orderBy(({ products }) => products.rating, 'desc');
    });

    return (
        <div>
            <select onChange={(e) => setCategory(e.target.value || null)}>
                <option value="">All Categories</option>
                <option value="electronics">Electronics</option>
            </select>
            <input type="range" max="1000" value={maxPrice}
                onChange={(e) => setMaxPrice(+e.target.value)} />

            {filteredProducts?.map(product => (
                <ProductCard key={product.id} product={product} />
            ))}
        </div>
    );
}

Example 3: Social Media Feed with Likes

// SocialFeed.tsx
import { useLiveQuery } from '@tanstack/react-db';
import { eq } from '@tanstack/db';
import { useStore } from './app';
import { newRecordId } from 'pbtsdb';

export function SocialFeed({ currentUserId }: { currentUserId: string }) {
    const [posts, likes] = useStore('posts', 'likes');

    const { data: feedPosts } = useLiveQuery((q) =>
        q.from({ posts }).orderBy(({ posts }) => posts.created, 'desc')
    );

    const { data: userLikes } = useLiveQuery((q) =>
        q.from({ likes }).where(({ likes }) => eq(likes.user, currentUserId))
    );

    const likedPostIds = new Set(userLikes?.map(l => l.post) || []);

    const handleLike = (postId: string) => {
        if (likedPostIds.has(postId)) {
            const like = userLikes?.find(l => l.post === postId);
            if (like) likes.delete(like.id);
        } else {
            likes.insert({ id: newRecordId(), post: postId, user: currentUserId });
        }
    };

    return (
        <div>
            {feedPosts?.map(post => (
                <div key={post.id}>
                    <strong>{post.expand?.author?.username}</strong>
                    <p>{post.content}</p>
                    <button onClick={() => handleLike(post.id)}>
                        {likedPostIds.has(post.id) ? '❤️' : '🤍'} {post.likes_count}
                    </button>
                </div>
            ))}
        </div>
    );
}

Example 4: Real-time Collaborative Todo List

// CollaborativeTodoList.tsx
import { useLiveQuery } from '@tanstack/react-db';
import { eq } from '@tanstack/db';
import { useStore } from './app';
import { newRecordId } from 'pbtsdb';

export function CollaborativeTodoList({ listId, userId }: { listId: string; userId: string }) {
    const [todos] = useStore('todos');
    const [newText, setNewText] = useState('');

    // Real-time todos - updates when any user adds/edits
    const { data: allTodos } = useLiveQuery((q) =>
        q.from({ todos })
            .where(({ todos }) => eq(todos.list_id, listId))
            .orderBy(({ todos }) => todos.created, 'asc')
    );

    const handleAdd = () => {
        if (!newText.trim()) return;
        todos.insert({ id: newRecordId(), text: newText, completed: false, list_id: listId, created_by: userId });
        setNewText('');
    };

    return (
        <div>
            <input value={newText} onChange={(e) => setNewText(e.target.value)}
                onKeyPress={(e) => e.key === 'Enter' && handleAdd()} />
            <ul>
                {allTodos?.map(todo => (
                    <li key={todo.id}>
                        <input type="checkbox" checked={todo.completed}
                            onChange={() => todos.update(todo.id, d => { d.completed = !d.completed; })} />
                        {todo.text}
                        <button onClick={() => todos.delete(todo.id)}>×</button>
                    </li>
                ))}
            </ul>
        </div>
    );
}

Real-time collaboration works automatically - when User A adds/edits a todo, User B sees it instantly.

Example 5: Form with Optimistic Updates and Error Handling

// CreateBookForm.tsx
import { useStore } from './app';
import { newRecordId } from 'pbtsdb';

export function CreateBookForm() {
    const [books] = useStore('books');
    const [title, setTitle] = useState('');
    const [error, setError] = useState<string | null>(null);

    const handleSubmit = async (e: React.FormEvent) => {
        e.preventDefault();
        setError(null);

        try {
            // Optimistic insert - appears instantly
            const tx = books.insert({ id: newRecordId(), title, author: 'author_id' });
            await tx.isPersisted.promise;

            if (tx.state === 'completed') setTitle('');
            else setError('Failed to create book');
        } catch (err: any) {
            setError(err.data ? Object.values(err.data).join(', ') : err.message);
        }
    };

    return (
        <form onSubmit={handleSubmit}>
            {error && <div className="error">{error}</div>}
            <input value={title} onChange={(e) => setTitle(e.target.value)} required />
            <button type="submit">Add Book</button>
        </form>
    );
}

Optimistic updates show changes instantly; automatic rollback on server errors.

Example 6: Dashboard with Multiple Collections and Joins

// ProjectDashboard.tsx
import { useLiveQuery } from '@tanstack/react-db';
import { eq } from '@tanstack/db';
import { useStore } from './app';

export function ProjectDashboard({ projectId }: { projectId: string }) {
    const [projects, tasks, teamMembers, users] = useStore('projects', 'tasks', 'team_members', 'users');

    const { data: projectList } = useLiveQuery((q) =>
        q.from({ projects }).where(({ projects }) => eq(projects.id, projectId))
    );

    const { data: projectTasks } = useLiveQuery((q) =>
        q.from({ tasks }).where(({ tasks }) => eq(tasks.project, projectId))
    );

    // Join team members with users
    const { data: team } = useLiveQuery((q) =>
        q.from({ member: teamMembers })
            .where(({ member }) => eq(member.project, projectId))
            .join({ user: users }, ({ member, user }) => eq(member.user, user.id), 'left')
            .select(({ member, user }) => ({ id: member.id, role: member.role, name: user?.name }))
    );

    const completed = projectTasks?.filter(t => t.completed).length || 0;
    const total = projectTasks?.length || 0;

    return (
        <div>
            <h1>{projectList?.[0]?.name}</h1>
            <p>Progress: {completed}/{total} tasks</p>
            <p>Team: {team?.map(m => m.name).join(', ')}</p>
        </div>
    );
}

Demonstrates variadic useStore(), client-side aggregations, and TanStack DB joins.

Includes (Nested Subqueries)

TanStack DB 0.6.0 introduces includes — nested subqueries within select() that project normalized data into hierarchical shapes. This is useful when you want to compose related data from multiple collections reactively.

Single relation with findOne()

import { useLiveQuery, eq } from '@tanstack/react-db';

const { data: booksWithAuthors } = useLiveQuery((q) =>
    q.from({ b: booksCollection }).select(({ b }) => ({
        id: b.id,
        title: b.title,
        author: q
            .from({ a: authorsCollection })
            .where(({ a }) => eq(a.id, b.author))
            .select(({ a }) => ({ id: a.id, name: a.name }))
            .findOne(),
    }))
);

Many relation with toArray()

import { useLiveQuery, eq, toArray } from '@tanstack/react-db';
// Or: import { toArray } from 'pbtsdb';

const { data: booksWithTags } = useLiveQuery((q) =>
    q.from({ b: booksCollection }).select(({ b }) => ({
        id: b.id,
        title: b.title,
        tags: toArray(
            q.from({ bt: bookTagsCollection })
                .where(({ bt }) => eq(bt.book, b.id))
                .join({ t: tagsCollection }, ({ bt, t }) => eq(bt.tag, t.id))
                .select(({ t }) => ({ id: t.id, name: t.name }))
        ),
    }))
);

Combining expand with includes

Use PocketBase's expand to auto-populate a related collection, then use includes to query from it:

const authorsCollection = c('authors', { syncMode: 'on-demand' });
const booksCollection = c('books', {
    syncMode: 'on-demand',
    expand: { author: authorsCollection },  // Auto-populates authorsCollection
});

const { data } = useLiveQuery((q) =>
    q.from({ b: booksCollection }).select(({ b }) => ({
        id: b.id,
        title: b.title,
        // Query from the expand-populated authorsCollection
        author: q.from({ a: authorsCollection })
            .where(({ a }) => eq(a.id, b.author))
            .select(({ a }) => ({ id: a.id, name: a.name }))
            .findOne(),
        tags: toArray(
            q.from({ bt: bookTagsCollection })
                .where(({ bt }) => eq(bt.book, b.id))
                .join({ t: tagsCollection }, ({ bt, t }) => eq(bt.tag, t.id))
                .select(({ t }) => ({ id: t.id, name: t.name }))
        ),
    }))
);

TypeScript

pbtsdb is fully type-safe. Here's what you need to know:

Define Your Schema

Use the simple schema format shown in the Quick Start:

type MySchema = {
    collection_name: {
        type: RecordInterface;    // Your record type
        relations: {
            field_name: RelatedType;  // Related record types
        };
    };
}

Pro tip: Use pocketbase-schema-generator to auto-generate types from your PocketBase database.

Type-Safe Collections

Always create collections with proper type parameters:

// ✅ Good - full type safety
const c = createCollection<MySchema>(pb, queryClient);
const books = c('books', {
    omitOnInsert: ['created', 'updated'] as const
});

// ✅ Good - with auto-expand relations
const authors = c('authors', {});
const books = c('books', {
    expand: {
        author: authors
    }
});

Best Practices

1. Define Collections Centrally

Define all collections once at app initialization:

// ✅ Do this - centralized, type-safe
const c = createCollection<MySchema>(pb, queryClient);

export const { Provider, useStore } = createReactProvider({
    posts: c('posts', { omitOnInsert: ['created', 'updated'] as const }),
    users: c('users', {}),
    comments: c('comments', { omitOnInsert: ['created', 'updated'] as const })
});

2. Create Dependencies Before Dependents

When using expand collections, create the target collection first:

// ✅ Good - authors exists before books references it
const c = createCollection<MySchema>(pb, queryClient);
const authors = c('authors', {});
const books = c('books', {
    expand: {
        author: authors  // authors is already created
    }
});

// ❌ Bad - can't reference what doesn't exist yet
const books = c('books', {
    expand: {
        author: ???  // Where is authors?
    }
});

3. Subscriptions are Automatic

Don't manually subscribe - just use useLiveQuery:

// ✅ Do this
const { data } = useLiveQuery((q) => q.from({ posts }));

// ❌ Don't do this
useEffect(() => {
    posts.subscribe();
    return () => posts.unsubscribe();
}, []);

4. Handle Loading States

Always check loading and error states:

const { data, isLoading, error } = useLiveQuery((q) => q.from({ posts }));

if (isLoading) return <div>Loading...</div>;
if (error) return <div>Error: {error.message}</div>;
if (!data?.length) return <div>No posts found</div>;

return <PostsList posts={data} />;

5. Use Expand for Performance

Use PocketBase's expand feature for better performance:

// ✅ Fast - single query with server-side expand
const c = createCollection<MySchema>(pb, queryClient);
const authors = c('authors', {});
const posts = c('posts', {
    expand: {
        author: authors  // Auto-expand on every fetch
    }
});

const { data } = useLiveQuery((q) => q.from({ posts }));

// ⚠️ Slower - multiple queries + client-side join
// Only use TanStack DB joins for inner/right/full join behavior

6. Configure QueryClient Defaults

const queryClient = new QueryClient({
    defaultOptions: {
        queries: {
            staleTime: 60_000,      // 1 minute
            gcTime: 300_000,        // 5 minutes
            refetchOnWindowFocus: false
        }
    }
});

Configuration

Custom Logger Integration

By default, pbtsdb logs debug messages to the console in development mode. You can integrate with your own logging service (Sentry, LogRocket, etc.) using setLogger:

import { setLogger } from 'pbtsdb';

// Example: Send errors to Sentry
setLogger({
    debug: (msg, context) => {
        // Custom debug handling (e.g., only log in dev)
        if (process.env.NODE_ENV === 'development') {
            console.debug('[pbtsdb]', msg, context);
        }
    },
    warn: (msg, context) => {
        console.warn('[pbtsdb]', msg, context);
        // Optional: Send to monitoring service
        myMonitoringService.warn(msg, context);
    },
    error: (msg, context) => {
        console.error('[pbtsdb]', msg, context);
        // Send errors to error tracking service
        Sentry.captureMessage(msg, {
            level: 'error',
            extra: context,
        });
    },
});

Disable logging completely:

import { setLogger } from 'pbtsdb';

setLogger({
    debug: () => {},
    warn: () => {},
    error: () => {},
});

Reset to default logger:

import { resetLogger } from 'pbtsdb';

resetLogger();

License

MIT

Contributing

Contributions welcome! Please open an issue or PR.

Development Setup

Prerequisites:

• Node.js 18+
• Git

Clone and Install:

git clone https://github.com/yourusername/pbtsdb
cd pbtsdb
npm install

Running Tests

Tests use a real PocketBase instance with fully automated infrastructure:

npm test  # Auto-resets DB → Starts server → Runs tests → Stops server

The npm test command automatically:

1. Resets the test database to a clean state
2. Applies migrations and creates test collections
3. Starts PocketBase server on port 8210
4. Runs all Vitest tests
5. Stops the server when complete

No manual server setup required! All test infrastructure is automated.

Advanced (for watch mode or debugging):

# Start test server manually
npm run test:server

# Run tests against running server (in another terminal)
npm run test:run

# Just reset database without starting server
npm run db:reset

Code Quality

npm run checks      # Run TypeScript type checking and linting
npm run lint:fix    # Auto-fix linting issues
npm run typecheck   # TypeScript only

Documentation

• See AGENTS.md for comprehensive development guidelines
• See test/README.md for detailed testing documentation

---

Built with:

• PocketBase - Backend-as-a-Service
• TanStack Query - Powerful data fetching
• TanStack DB - Reactive database
• TypeScript - Type safety