This is a reverse chronological list of the major TinyBase releases, with highlighted features.
We're excited to announce TinyBase 4.3, which provides an integration with PartyKit, a cloud-based collaboration provider.
This allows you to enjoy the benefits of both a "local-first" architecture and a "sharing-first" platform. You can have structured data on the client with fast, reactive user experiences, but also benefit from cloud-based persistence and room-based collaboration.
This release includes two new modules:
- The
persister-partykit-server
module provides a server class for coordinating clients and persistingStore
data to the PartyKit cloud. - The
persister-partykit-client
module provides the API to create connections to the server and a binding to yourStore
.
A TinyBase server implementation on PartyKit can be as simple as this:
import {TinyBasePartyKitServer} from 'tinybase/persisters/persister-partykit-server';
export default class extends TinyBasePartyKitServer {}
On the client, use the familiar Persister
API, passing in a reference to a PartyKit socket object that's been configured to connect to your server deployment and named room:
const persister = createPartyKitPersister(
store,
new PartySocket({
host: 'project-name.my-account.partykit.dev',
room: 'my-partykit-room',
}),
);
await persister.startAutoSave();
await persister.startAutoLoad();
The load method and (gracefully failing) save method on this Persister
use HTTPS to get or set full copies of the Store
to the cloud. However, the auto-save and auto-load modes use a websocket to transmit subsequent incremental changes in either direction, making for performant sharing of state between clients.
See and try out this new collaboration functionality in the Todo App v6 (collaboration) demo. This also emphasizes the few changes that need to be made to an existing app to make it instantly collaborative.
Also try out the tinybase-ts-react-partykit template that gets you up and running with a PartyKit-enabled TinyBase app extremely quickly.
PartyKit supports retries for clients that go offline, and so the disconnected user experience is solid, out of the box. Learn more about configuring this behavior here.
Note, however, that this release is not yet a full CRDT implementation: there is no clock synchronization and it is more 'every write wins' than 'last write wins'. However, since the transmitted updates are at single cell (or value) granularity, conflicts are minimized. More resilient replication is planned as this integration matures.
This release adds support for persisting TinyBase to a browser's IndexedDB storage. You'll need to import the new persister-indexed-db
module, and call the createIndexedDbPersister
function to create the IndexedDB Persister
.
The API is the same as for all the other Persister
APIs:
const store = createStore()
.setTable('pets', {fido: {species: 'dog'}})
.setTable('species', {dog: {price: 5}})
.setValues({open: true});
const indexedDbPersister = createIndexedDbPersister(store, 'petStore');
await indexedDbPersister.save();
// IndexedDB ->
// database petStore:
// objectStore t:
// object 0:
// k: "pets"
// v: {fido: {species: dog}}
// object 1:
// k: "species"
// v: {dog: {price: 5}}
// objectStore v:
// object 0:
// k: "open"
// v: true
indexedDbPersister.destroy();
Note that it is not possible to reactively detect changes to a browser's IndexedDB storage. A polling technique is used to load underlying changes if you choose to 'autoLoad' your data into TinyBase.
This release also upgrades Prettier to v3.0 which has a peer-dependency impact on the tools
module. Please report any issues!
This release introduces the new ui-react-dom
module. This provides pre-built components for tabular display of your data in a web application.
The following is the list of all the components released in v4.1:
Component | Purpose | |
---|---|---|
ValuesInHtmlTable | Renders Values . | demo |
TableInHtmlTable | Renders a Table . | demo |
SortedTableInHtmlTable | Renders a sorted Table , with optional interactivity. | demo |
SliceInHtmlTable | Renders a Slice from an Index . | demo |
RelationshipInHtmlTable | Renders the local and remote Tables of a relationship | demo |
ResultTableInHtmlTable | Renders a ResultTable . | demo |
ResultSortedTableInHtmlTable | Renders a sorted ResultTable , with optional interactivity. | demo |
EditableCellView | Renders a Cell and lets you change its type and value. | demo |
EditableValueView | Renders a Value and lets you change its type and value. | demo |
These pre-built components are showcased in the UI Components demos. Using them should be very familiar if you have used the more abstract ui-react
module:
const App = ({store}) => (
<SortedTableInHtmlTable tableId="pets" cellId="species" store={store} />
);
store.setTables({
pets: {
fido: {species: 'dog'},
felix: {species: 'cat'},
},
});
const app = document.createElement('div');
const root = ReactDOMClient.createRoot(app);
root.render(<App store={store} />);
console.log(app.innerHTML);
// ->
`
<table>
<thead>
<tr><th>Id</th><th class="sorted ascending">↑ species</th></tr>
</thead>
<tbody>
<tr><th>felix</th><td>cat</td></tr>
<tr><th>fido</th><td>dog</td></tr>
</tbody>
</table>
`;
root.unmount();
The EditableCellView
component and EditableValueView
component are interactive input controls for updating Cell
and Value
content respectively. You can generally use them across your table views by adding the editable
prop to your table component.
The new StoreInspector
component allows you to view, understand, and edit the content of a Store
in a debug web environment. Try it out in most of the demos on the site, including the Movie Database demo, pictured. This requires a debug build of the new ui-react-dom
module, which is now also included in the UMD distribution.
Also in this release, the getResultTableCellIds
method and addResultTableCellIdsListener
method have been added to the Queries
object. The equivalent useResultTableCellIds
hook and useResultTableCellIdsListener
hook have also been added to ui-react
module. A number of other minor React hooks have been added to support the components above.
Demos have been updated to demonstrate the ui-react-dom
module and the StoreInspector
component where appropriate.
This major release provides Persister
modules that connect TinyBase to SQLite databases (in both browser and server contexts), and CRDT frameworks that can provide synchronization and local-first reconciliation:
Module | Function | Storage |
---|---|---|
persister-sqlite3 | createSqlite3Persister | SQLite in Node, via sqlite3 |
persister-sqlite-wasm | createSqliteWasmPersister | SQLite in a browser, via sqlite-wasm |
persister-cr-sqlite-wasm | createCrSqliteWasmPersister | SQLite CRDTs, via cr-sqlite-wasm |
persister-yjs | createYjsPersister | Yjs CRDTs, via yjs |
persister-automerge | createSqliteWasmPersister | Automerge CRDTs, via automerge-repo |
See the Database Persistence guide for details on how to work with SQLite databases, and the Synchronizing Data guide for more complex synchronization with the CRDT frameworks.
Take a look at the vite-tinybase-ts-react-crsqlite template, for example, which demonstrates Vulcan's cr-sqlite to provide persistence and synchronization via this technique.
You can persist Store
data to a database with either a JSON serialization or tabular mapping. (See the DatabasePersisterConfig
documentation for more details).
For example, this creates a Persister
object and saves and loads the Store
to and from a local SQLite database. It uses an explicit tabular one-to-one mapping for the 'pets' table:
const sqlite3 = await sqlite3InitModule();
const db = new sqlite3.oo1.DB(':memory:', 'c');
store.setTables({pets: {fido: {species: 'dog'}}});
const sqlitePersister = createSqliteWasmPersister(store, sqlite3, db, {
mode: 'tabular',
tables: {load: {pets: 'pets'}, save: {pets: 'pets'}},
});
await sqlitePersister.save();
console.log(db.exec('SELECT * FROM pets;', {rowMode: 'object'}));
// -> [{_id: 'fido', species: 'dog'}]
db.exec(`INSERT INTO pets (_id, species) VALUES ('felix', 'cat')`);
await sqlitePersister.load();
console.log(store.getTables());
// -> {pets: {fido: {species: 'dog'}, felix: {species: 'cat'}}}
sqlitePersister.destroy();
CRDTs allow complex reconciliation and synchronization between clients. Yjs and Automerge are two popular examples. The API should be familiar! The following will persist a TinyBase Store
to a Yjs document:
store.setTables({pets: {fido: {species: 'dog'}}});
const doc = new Y.Doc();
const yJsPersister = createYjsPersister(store, doc);
await yJsPersister.save();
// Store will be saved to the document.
console.log(doc.toJSON());
// -> {tinybase: {t: {pets: {fido: {species: 'dog'}}}, v: {}}}
yJsPersister.destroy();
The following is the equivalent for an Automerge document that will sync over the broadcast channel:
const docHandler = new AutomergeRepo.Repo({
network: [new BroadcastChannelNetworkAdapter()],
}).create();
const automergePersister = createAutomergePersister(store, docHandler);
await automergePersister.save();
// Store will be saved to the document.
console.log(docHandler.doc);
// -> {tinybase: {t: {pets: {fido: {species: 'dog'}}}, v: {}}}
automergePersister.destroy();
store.delTables();
There are three new methods on the Store
object. The getContent
method lets you get the Store
's Tables
and Values
in one call. The corresponding setContent
method lets you set them simultaneously.
The new setTransactionChanges
method lets you replay TransactionChanges
(received at the end of a transaction via listeners) into a Store
, allowing you to take changes from one Store
and apply them to another.
Persisters now provide a schedule method that lets you queue up asynchronous tasks, such as when persisting data that requires complex sequences of actions.
The way that data is provided to the DoRollback
and TransactionListener
callbacks at the end of a transaction has changed. Where
previously they directly received content about changed Cell
and Value
content, they now receive functions that they can choose to call to receive that same data. This has a performance improvement, and your callback or listener can choose between concise TransactionChanges
or more verbose TransactionLog
structures for that data.
If you have build a custom persister, you will need to update your implementation. Most notably, the setPersisted
function parameter is provided with a getContent
function to get the content from the Store
itself, rather than being passed pre-serialized JSON. It also receives information about the changes made during a transaction. The getPersisted
function must return the content (or nothing) rather than JSON. startListeningToPersisted
has been renamed addPersisterListener
, and stopListeningToPersisted
has been renamed delPersisterListener
.
This release allows you to track the Cell
Ids
used across a whole Table
, regardless of which Row
they are in.
In a Table
(particularly in a Store
without a TablesSchema
), different Rows can use different Cells. Consider this Store
, where each pet has a different set of Cell
Ids
:
store.setTable('pets', {
fido: {species: 'dog'},
felix: {species: 'cat', friendly: true},
cujo: {legs: 4},
});
Prior to v3.3, you could only get the Cell
Ids
used in each Row
at a time (with the getCellIds
method). But you can now use the getTableCellIds
method to get the union of all the Cell
Ids
used across the Table
:
console.log(store.getCellIds('pets', 'fido')); // previously available
// -> ['species']
console.log(store.getTableCellIds('pets')); // new in v3.3
// -> ['species', 'friendly', 'legs']
You can register a listener to track the Cell
Ids
used across a Table
with the new addTableCellIdsListener
method. Use cases for this might include knowing which headers to render when displaying a sparse Table
in a user interface, or synchronizing data with relational or column-oriented database system.
There is also a corresponding useTableCellIds
hook in the optional ui-react
module for accessing these Ids
reactively, and a useTableCellIdsListener
hook for more advanced purposes.
Note that the bookkeeping behind these new accessors and listeners is efficient and should not be slowed by the number of Rows in the Table
.
This release also passes a getIdChanges function to every Id
-related listener that, when called, returns information about the Id
changes, both additions and removals, during a transaction. See the TableIdsListener
type, for example.
let listenerId = store.addRowIdsListener(
'pets',
(store, tableId, getIdChanges) => {
console.log(getIdChanges());
},
);
store.setRow('pets', 'lowly', {species: 'worm'});
// -> {lowly: 1}
store.delRow('pets', 'felix');
// -> {felix: -1}
store.delListener(listenerId).delTables();
This release lets you add a listener to the start of a transaction, and detect that a set of changes are about to be made to a Store
.
To use this, call the addStartTransactionListener
method on your Store
. The listener you add can itself mutate the data in the Store
.
From this release onwards, listeners added with the existing addWillFinishTransactionListener
method are also able to mutate data. Transactions added with the existing addDidFinishTransactionListener
method cannot mutate data.
const startListenerId = store.addStartTransactionListener(() => {
console.log('Start transaction');
console.log(store.getTables());
// Can mutate data
});
const willFinishListenerId = store.addWillFinishTransactionListener(() => {
console.log('Will finish transaction');
console.log(store.getTables());
// Can mutate data
});
const didFinishListenerId = store.addDidFinishTransactionListener(() => {
console.log('Did finish transaction');
console.log(store.getTables());
// Cannot mutate data
});
store.setTable('pets', {fido: {species: 'dog'}});
// -> 'Start transaction'
// -> {}
// -> 'Will finish transaction'
// -> {pets: {fido: {species: 'dog'}}}
// -> 'Did finish transaction'
// -> {pets: {fido: {species: 'dog'}}}
store
.delListener(startListenerId)
.delListener(willFinishListenerId)
.delListener(didFinishListenerId);
store.delTables();
This release also fixes a bug where using the explicit startTransaction
method inside another listener could create infinite recursion.
This new release adds a powerful schema-based type system to TinyBase.
If you define the shape and structure of your data with a TablesSchema
or ValuesSchema
, you can benefit from an enhanced developer experience when operating on it. For example:
// Import the 'with-schemas' definition:
import {createStore} from 'tinybase/with-schemas';
// Set a schema for a new Store:
const store = createStore().setValuesSchema({
employees: {type: 'number'},
open: {type: 'boolean', default: false},
});
// Benefit from inline TypeScript errors.
store.setValues({employees: 3}); // OK
store.setValues({employees: true}); // TypeScript error
store.setValues({employees: 3, website: 'pets.com'}); // TypeScript error
The schema-based typing is used comprehensively throughout every module - from the core Store
interface all the way through to the ui-react
module. See the new Schema-Based Typing guide for instructions on how to use it.
This now means that there are three progressive ways to use TypeScript with TinyBase:
- Basic Type Support (since v1.0)
- Schema-based Typing (since v3.1)
- ORM-like type definitions (since v2.2)
These are each described in the new TinyBase and TypeScript guide.
Also in v3.1, the ORM-like type definition generation in the tools
module has been extended to emit ui-react
module definitions.
Finally, v3.1.1 adds a reuseRowIds
parameter to the addRow
method and the useAddRowCallback
hook. It defaults to true
, for backwards compatibility, but if set to false
, new Row
Ids
will not be reused unless the whole Table
is deleted.
This major new release adds key/value store functionality to TinyBase. Alongside existing tabular data, it allows you to get, set, and listen to, individual Value
items, each with a unique Id
.
store.setValues({employees: 3, open: true});
console.log(store.getValues());
// -> {employees: 3, open: true}
listenerId = store.addValueListener(
null,
(store, valueId, newValue, oldValue) => {
console.log(`Value '${valueId}' changed from ${oldValue} to ${newValue}`);
},
);
store.setValue('employees', 4);
// -> "Value 'employees' changed from 3 to 4"
store.delListener(listenerId).delValues();
Guides and documentation have been fully updated, and certain demos - such as the Todo App v2 (indexes) demo, and the Countries demo - have been updated to use this new functionality.
If you use the optional ui-react
module with TinyBase, v3.0 now uses and expects React v18.
In terms of core API changes in v3.0, there are some minor breaking changes (see below), but the majority of the alterations are additions.
The Store
object gains the following:
- The
setValues
method,setPartialValues
method, andsetValue
method, to set keyed value data into theStore
. - The
getValues
method,getValueIds
method, andgetValue
method, to get keyed value data out of theStore
. - The
delValues
method anddelValue
method for removing keyed value data. - The
addValuesListener
method,addValueIdsListener
method, addValueListener method, andaddInvalidValueListener
method, for listening to changes to keyed value data. - The
hasValues
method,hasValue
method, andforEachValue
method, for existence and enumeration purposes. - The
getTablesJson
method,getValuesJson
method,setTablesJson
method, andsetValuesJson
method, for reading and writing tabular and keyed value data to and from a JSON string. Also see below. - The
getTablesSchemaJson
method,getValuesSchemaJson
method, setTablesSchema method,setValuesSchema
method,delTablesSchema
method, and delValuesSchema method, for reading and writing tabular and keyed value schemas for theStore
. Also see below.
The following types have been added to the store
module:
Values
,Value
, andValueOrUndefined
, representing keyed value data in aStore
.ValueListener
andInvalidValueListener
, to describe functions used to listen to (valid or invalid) changes to aValue
.ValuesSchema
andValueSchema
, to describe the keyedValues
that can be set in aStore
and their types.ValueCallback
,MapValue
,ChangedValues
, andInvalidValues
, which also correspond to their 'Cell
' equivalents.
Additionally:
- The persisters' load method and startAutoLoad method take an optional
initialValues
parameter for settingValues
when a persistedStore
is bootstrapped. - The
Checkpoints
module will undo and redo changes to keyed values in the same way they do for tabular data. - The
tools
module provides agetStoreValuesSchema
method for inferring value-based schemas. ThegetStoreApi
method andgetPrettyStoreApi
method now also provides an ORM-like code-generated API for schematized key values.
All attempts have been made to provide backwards compatibility and/or easy upgrade paths.
In previous versions, getJson
method would get a JSON serialization of the Store
's tabular data. That functionality is now provided by the getTablesJson
method, and the getJson
method instead now returns a two-part array containing the tabular data and the keyed value data.
Similarly, the getSchemaJson
method used to return the tabular schema, now provided by the getTablesSchemaJson
method. The getSchemaJson
method instead now returns a two-part array of tabular schema and the keyed value schema.
The setJson
method used to take a serialization of just the tabular data object. That's now provided by the setTablesJson
method, and the setJson
method instead expects a two-part array containing the tabular data and the keyed value data (as emitted by the getJson
method). However, for backwards compatibility, if the setJson
method is passed an object, it will set the tabular data, as it did prior to v3.0.
Along similar lines, the setSchema
method's previous behavior is now provided by the setTablesSchema
method. The setSchema
method now takes two arguments, the second of which is optional, also aiding backward compatibility. The delSchema
method removes both types of schema.
This release includes a new tools
module. These tools are not intended for production use, but are instead to be used as part of your engineering workflow to perform tasks like generating APIs from schemas, or schemas from data. For example:
store.setTable('pets', {
fido: {species: 'dog'},
felix: {species: 'cat'},
cujo: {species: 'dog'},
});
const tools = createTools(store);
const [dTs, ts] = tools.getStoreApi('shop');
This will generate two files:
// -- shop.d.ts --
/* Represents the 'pets' Table. */
export type PetsTable = {[rowId: Id]: PetsRow};
/* Represents a Row when getting the content of the 'pets' Table. */
export type PetsRow = {species: string};
//...
// -- shop.ts --
export const createShop: typeof createShopDecl = () => {
//...
};
This release includes a new tinybase
CLI tool which allows you to generate Typescript definition and implementation files direct from a schema file:
npx tinybase getStoreApi schema.json shop api
Definition: [...]/api/shop.d.ts
Implementation: [...]/api/shop.ts
Finally, the tools
module also provides ways to track the overall size and structure of a Store
for use while debugging.
This release allows you to create indexes where a single Row
Id
can exist in multiple slices. You can utilize this to build simple keyword searches, for example.
Simply provide a custom getSliceIdOrIds function in the setIndexDefinition
method that returns an array of Slice
Ids
, rather than a single Id
:
store.setTable('pets', {
fido: {species: 'dog'},
felix: {species: 'cat'},
rex: {species: 'dog'},
});
const indexes = createIndexes(store);
indexes.setIndexDefinition('containsLetter', 'pets', (_, rowId) =>
rowId.split(''),
);
console.log(indexes.getSliceIds('containsLetter'));
// -> ['f', 'i', 'd', 'o', 'e', 'l', 'x', 'r']
console.log(indexes.getSliceRowIds('containsLetter', 'i'));
// -> ['fido', 'felix']
console.log(indexes.getSliceRowIds('containsLetter', 'x'));
// -> ['felix', 'rex']
This functionality is showcased in the Word Frequencies demo if you would like to see it in action.
Announcing the next major version of TinyBase 2.0! This is an exciting release that evolves TinyBase towards becoming a reactive, relational data store, complete with querying, sorting, and pagination. Here are a few of the highlights...
The flagship feature of this release is the new queries
module. This allows you to build expressive queries against your data with a SQL-adjacent API that we've cheekily called TinyQL. The query engine lets you select, join, filter, group, sort and paginate data. And of course, it's all reactive!
The best way to see the power of this new engine is with the two new demos we've included this release:
The Car Analysis demo showcases the analytical query capabilities of TinyBase v2.0, grouping and sorting dimensional data for lightweight analytical usage, graphing, and tabular display. Try this demo here.
The Movie Database demo showcases the relational query capabilities of TinyBase v2.0, joining together information about movies, directors, and actors from across multiple source tables. Try this demo here.
To complement the query engine, you can now sort and paginate Row
Ids
. This makes it very easy to build grid-like user interfaces (also shown in the demos above). To achieve this, the Store
now includes the getSortedRowIds
method (and the addSortedRowIdsListener
method for reactivity), and the Queries
object includes the equivalent getResultSortedRowIds
method and addResultSortedRowIdsListener
method.
These are also exposed in the optional ui-react
module via the useSortedRowIds
hook, the useResultSortedRowIds
hook, the SortedTableView
component and the ResultSortedTableView
component, and so on.
The v2.0 query functionality is fully supported by the ui-react
module (to match support for Store
, Metrics
, Indexes
, and Relationship
objects). The useCreateQueries
hook memoizes the creation of app- or component-wide Query objects; and the useResultTable
hook, useResultRow
hook, useResultCell
hook (and so on) let you bind you component to the results of a query.
This is, of course, supplemented with higher-level components: the ResultTableView
component, the ResultRowView
component, the ResultCellView
component, and so on. See the Building A UI With Queries guide for more details.
Thank you for all your support as we brought this important new release to life, and we hope you enjoy using it as much as we did building it. Please provide feedback via Github and Twitter!
Adds support for explicit transaction start and finish methods, as well as listeners for transactions finishing.
The startTransaction
method and finishTransaction
method allow you to explicitly enclose a transaction that will make multiple mutations to the Store
, buffering all calls to the relevant listeners until it completes when you call the finishTransaction
method.
Unlike the transaction
method, this approach is useful when you have a more 'open-ended' transaction, such as one containing mutations triggered from other events that are asynchronous or not occurring inline to your code. You must remember to also call the finishTransaction
method explicitly when the transaction is started with the startTransaction
method, of course.
store.setTables({pets: {fido: {species: 'dog'}}});
store.addRowListener('pets', 'fido', () => console.log('Fido changed'));
store.startTransaction();
store.setCell('pets', 'fido', 'color', 'brown');
store.setCell('pets', 'fido', 'sold', true);
store.finishTransaction();
// -> 'Fido changed'
In addition, see the addWillFinishTransactionListener
method and the addDidFinishTransactionListener
method for details around listening to transactions completing.
Together, this release allows stores to couple their transaction life-cycles together, which we need for the query engine.
Note: this API was updated to be more comprehensive in v4.0.
This adds a way to revert transactions if they have not met certain conditions.
When using the transaction
method, you can provide an optional doRollback
callback which should return true if you want to revert the whole transaction at its conclusion.
The callback is provided with two objects, changedCells
and invalidCells
, which list all the net changes and invalid attempts at changes that were made during the transaction. You will most likely use the contents of those objects to decide whether the transaction should be rolled back.
Note: this API was updated to be more comprehensive in v4.0.
This release allows you to listen to invalid data being added to a Store
, allowing you to gracefully handle errors, rather than them failing silently.
There is a new listener type InvalidCellListener
and a addInvalidCellListener
method in the Store
interface.
These allow you to keep track of failed attempts to update the Store
with invalid Cell
data. These listeners can also be mutators, allowing you to address any failed writes programmatically.
For more information, please see the addInvalidCellListener
method documentation. In particular, this explains how this listener behaves for a Store
with a TablesSchema
.