reactronic
Version:
Reactronic - Transactional Reactive State Management
489 lines (399 loc) • 16.9 kB
Markdown
[](https://en.wikipedia.org/wiki/Software_release_life_cycle#Release_candidate)
[](https://github.com/nezaboodka/reactronic/blob/master/LICENSE)
[](https://www.npmjs.com/package/reactronic)
[](https://bundlephobia.com/result?p=reactronic)
[](https://gitlab.com/nezaboodka/nevod.website/-/blob/master/README.md)
# **Reactronic** - Transactional Reactive State Management
Reactronic is an experimental JavaScript library that provides
[transactional reactive](https://blog.nezaboodka.com/post/2019/593-modern-database-should-natively-support-transactionally-reactive-programming)
state management in a Web application.
Transactional reactivity means that state changes are
being made in an isolated data snapshot and then, once
atomically applied, are **consistently propagated** to
corresponding visual components for (re)rendering. All
that is done in automatic, seamless, and fine-grained
way. Reactronic **takes full care of tracking dependencies**
between visual components (reactive functions) and
application state (signalling objects).
Transactional reactivity is based on four fundamental
concepts:
- **Signalling Objects** - a set of objects that store
data of an application (state) and cause reactions
upon their changes;
- **Transactional Function** - a function that makes
changes in signalling objects in atomic way ("all
or nothing");
- **Reactive Function** - a function that is
(re-)executed in response to changes made in
signalling objects by transactional functions;
- **Cache Function** - a function which result is
remembered and, if becomes obsolete, causes
function to re-execute on-demand.
Demo application built with Reactronic: https://nevod.io/#/playground.
Source code of the demo: https://gitlab.com/nezaboodka/nevod.web.public/-/blob/master/README.md.
Quick introduction and detailed description is below.
## Quick Introduction
Here is an example of transactional reactive code:
``` typescript
class Demo extends SignallingObject {
name: string = 'Nezaboodka Software'
email: string = 'contact.com'
saveContact(name: string, email: string): void {
this.name = name
this.email = email
}
printContact(): void {
// depends on `name` and `email` and reacts to their changes
if (this.email.indexOf('@') >= 0)
throw new Error(`wrong email ${this.email}`)
console.log(this.name + ' <' + this.email + '>')
}
}
```
In the example above, `Demo` is a signalling object,
meaning that access to its fields are seamlessly tracked
to determine dependent reactive and cached functions.
Reactive function `printContact` reads `name` and `email`
fields, thus depends on them. It is executed automatically
in response to changes of these fields made by the
transactional function `saveContact`.
Here is an example of a cached result that is
(re-)computed on-demand:
``` typescript
class Demo extends SignallingObject {
name: string = 'Nezaboodka Software'
email: string = 'contact.com'
get contact(): string {
return this.name + ' <' + this.email + '>'
}
printContact(): void {
if (this.contact !== '')
Console.log(this.contact)
}
}
```
In the example above, the result of `contact` getter is
computed from source fields `name` and `email`. Once
computed, the result is cached and is reused until
source fields `name` and `email` are changed. Once
source fields changed, `contact` result becomes obsolete,
thus causing execution of depending reactive function
`printContact`. When function of reactive function
`printContact` runs it reads `contact` and causes its
re-computation.
## Signalling Objects
Signalling objects are aimed to store data of
an application. All such objects are transparently hooked
to track access to their properties, both on reads and
writes.
``` typescript
class MyModel extends SignallingObject {
url: string = "https://github.com/nezaboodka/reactronic"
content: string = "transactional reactive state management"
timestamp: Date = Date.now()
}
```
In the example above, the class `MyModel` is based on
Reactronic's `SignallingObject` class and all its
properties `url`, `content`, and `timestamp` are hooked.
## Transactional Function
Transactional function makes changes in signalling
objects in atomic way ("all or nothing"), thus provoking
execution of dependent reactive and cached functions.
Transactional function is instrumented with hooks to
provide transparent atomicity (by implicit context
switching and isolation).
``` typescript
class MyModel extends SignallingObject {
// ...
async load(url: string): Promise<void> {
this.url = url
this.content = await fetch(url)
this.timestamp = Date.now()
}
}
```
In the example above, the transactional function `load` makes
changes to `url`, `content` and `timestamp` properties.
While transactional function is running, the changes are visible
only inside the function itself. The new values become
atomically visible outside of the function only upon its
completion.
Atomicity is achieved by making changes in an isolated
data snapshot that is not visible outside of the running
function until it is fully finished and applied. Multiple
objects and their properties can be changed with full
respect to the all-or-nothing principle. To do so,
separate data snapshot is automatically maintained for
each transactional function. That is a logical snapshot
that does not create a full copy of all the data.
Compensating rollback operations are not needed in case
of a transactional function failure, because all the changes
made by transactional function in its logical snapshot are
simply discarded. In case a transaction function is
successfully applied, affected cached results are marked
as obsolete and corresponding caching functions are
re-executed in a proper order (but only when all the
data changes are fully applied).
Asynchronous operations (promises) are supported out of
the box during transactional function execution.
Transactional function may consist of a set of asynchronous
calls prolonging the function until completion of all of
them. An asynchronous call may spawn other asynchronous
calls, which prolong transactional execution until the whole
chain of asynchronous operations is fully completed.
## Reactive & Cached Functions
Reactive function is automatically and immediately called
in response to changes in signalling objects made by
transactional functions. Cached function is called on-demand
to renew the result if it was marked as obsolete due to
changes made by an transactional functions. Reactive and cached
functions are instrumented with hooks to seamless
listening of those signalling objects and other cached
functions (dependencies), which are used during their
execution.
``` tsx
class MyView extends Component<{model: MyModel}> {
render(): React.JSX.Element {
return (
<div>
<h1>{this.props.model.url}</h1>
<div>{this.props.model.content}</div>
</div>
)
} // render is listening to "url" and "content"
}
```
``` tsx
class Component<P> extends React.Component<P> {
render(): React.JSX.Element {
throw new Error('render method is undefined')
}
// called in response to changes
ensureUpToDate(): void {
if (this.shouldComponentUpdate()) {
// Ask React to re-render
Transaction.outside(() => this.setState({}))
}
} // EnsureUpToDate is listening to render
shouldComponentUpdate(): boolean {
const r = manageReaction(this.render)
return !r.isUpToDate
}
componentDidMount(): void {
// Run for the first time to start listening
this.ensureUpToDate()
}
componentWillUnmount(): void {
runTransactional(disposeSignallingObject, this)
}
}
```
In the example above, reactive function `refresh`
transparently listens to the cached function `render`.
In turn, cached function `render` listens to the
properties `url` and `content` of a corresponding `MyModel`
object. Once `url` or `content` values are changed, the
cached function `render` becomes obsolete and causes the
reactive function `refresh` to become obsolete and
re-executed. While being executed, the reactive function
`refresh` enqueues re-rendering request to React, which
calls cached function `render` causing it to renew its
cached value.
In general case, all reactive and cached functions
are automatically and immediately marked as obsolete
when changes are made in those signalling objects and
other cached results that were used during their
execution. And once marked, the functions are
automatically executed again, either immediately (for
reactive functions) or on-demand (for cached functions).
Reactronic takes full care of tracking dependencies
between all the signalling objects and reactive/cached
functions. With Reactronic, you no longer need to create
data change events in one set of objects, listen to
these events in other objects, and manually maintain
switching from the previous object version to a new one.
## Behavior Options
There are multiple options to configure behavior of
transactional reactivity.
**Order** options defines order of execution for
reactive functions:
- (TBD)
**Throttling** option defines how often reactive function
is executed in case of recurring changes:
- `(ms)` - minimal delay in milliseconds between executions;
- `-1` - execute immediately once transactional function changes are applied (synchronously);
- `0` - execute immediately via event loop (asynchronously with zero timeout);
- `>= Number.MAX_SAFE_INTEGER` - never execute (suspended reaction).
**Reentrance** option defines how to handle reentrant calls of transactional
and reactive functions:
- `preventWithError` - fail with error if there is an existing call in progress;
- `waitAndRestart` - wait for previous call to finish and then restart current one;
- `cancelPrevious` - cancel previous call in favor of recent one;
- `cancelAndWaitPrevious` - cancel previous call in favor of recent one (but wait until canceling is completed)
- `runSideBySide` - multiple simultaneous calls are allowed.
**Indicator** is an object that maintains status of running functions,
which it is attached to. A single indicator object can be shared between
multiple transactional, reactive, and cached functions, thus maintaining
consolidated status for all of them (busy, workers, etc).
## Notes
Inspired by: MobX, Nezaboodka, Excel.
Key Reactronic principles and differentiators:
- No compromises on consistency, clarity, and simplicity;
- Minimalism and zero boilerplating (it's not a framework bloating your code);
- Async, patches, undo/redo, conflict resolving are provided out of the box;
- Seamless integration with transactional reactive object-oriented databases like [Nezaboodka](https://nezaboodka.com/#products);
- Compact dependency-free implementation consisting of less than 2K lines of code.
Roadmap:
- Patches and conflict resolution API (partially done)
- History/undo/redo API and implementation (partially done)
- Sync API and implementation (not implemented yet)
## Installation
NPM: `npm install reactronic`
## API (TypeScript)
```typescript
// Classes
class TxObject { } // transactional object
class SxObject { } // signalling object
// Decorators & Operators
function signal(boolean) // field only
function signal(proto, prop) // field only
function transaction(proto, prop, pd) // method only
function reaction(proto, prop, pd) // method only
function cache(proto, prop, pd) // method only
function options(value: Partial<ReactivityOptions>): F<any>
function runNonReactive<T>(func: F<T>, ...args: any[]): T
function runSensitive<T>(sensitivity: Sensitivity, func: F<T>, ...args: any[]): T
// SnapshotOptions, ReactivityOptions, Kind, Reentrance, Indicator, LoggingOptions, ProfilingOptions
export type SnapshotOptions = {
readonly hint?: string
readonly isolation?: Isolation
readonly journal?: Journal
readonly logging?: Partial<LoggingOptions>
readonly token?: any
}
type ReactivityOptions = {
readonly kind: Kind
readonly isolation: Isolation
readonly order: number
readonly noSideEffects: boolean
readonly signalArgs: boolean
readonly throttling: number // milliseconds, -1 is immediately, Number.MAX_SAFE_INTEGER is never
readonly reentrance: Reentrance
readonly journal: Journal | undefined
readonly indicator: Indicator | null
readonly logging?: Partial<LoggingOptions>
}
enum Kind {
plain = 0,
transaction = 1,
reaction = 2,
cache = 3
}
enum Reentrance {
preventWithError = 1, // fail with error if there is an existing call in progress (default)
waitAndRestart = 0, // wait for existing call to finish and then restart current one
cancelPrevious = -1, // cancel previous call in favor of recent one
cancelAndWaitPrevious = -2, // cancel previous call in favor of recent one (but wait until canceling is completed)
overwritePrevious = -2, // allow previous to complete, but overwrite it with ignoring any conflicts
runSideBySide = -3 // multiple simultaneous calls are allowed
}
class Indicator {
readonly isBusy: boolean
readonly counter: number
readonly workers: ReadonlySet<Worker>
readonly busyDuration: number
abstract whenBusy(): Promise<void>
abstract whenIdle(): Promise<void>
static create(hint: string, activationDelay: number, deactivationDelay: number): Indicator
}
type Worker = {
readonly id: number
readonly hint: string
isCanceled: boolean
isFinished: boolean
cancel(error?: Error, retryAfter?: Transaction): this
whenFinished(): Promise<void>
}
type LoggingOptions = {
readonly off: boolean
readonly transaction: boolean
readonly operation: boolean
readonly step: boolean
readonly indicator: boolean
readonly read: boolean
readonly write: boolean
readonly change: boolean
readonly obsolete: boolean
readonly error: boolean
readonly warning: boolean
readonly gc: boolean
}
type ProfilingOptions = {
repetitiveUsageWarningThreshold: number // default: 10 times
mainThreadBlockingWarningThreshold: number // default: 16.6 ms
asyncActionDurationWarningThreshold: number // default: 150 ms
}
// Transaction
type F<T> = (...args: any[]) => T
class Transaction implements Worker {
static readonly current: Transaction
readonly id: number
readonly hint: string
run<T>(func: F<T>, ...args: any[]): T
wrap<T>(func: F<T>): F<T>
apply(): void
seal(): this // a1.seal().whenFinished().then(fulfill, reject)
cancel(error?: Error, retryAfter?: Transaction): this
isCanceled: boolean
isFinished: boolean
whenFinished(): Promise<void>
join<T>(p: Promise<T>): Promise<T>
static create(options: SnapshotOptions | null): Transaction
static run<T>(options: SnapshotOptions | null, func: F<T>, ...args: any[]): T
static off<T>(func: F<T>, ...args: any[]): T
static isFrameOver(everyN: number, timeLimit: number): boolean
static requestNextFrame(sleepTime: number): Promise<void>
static isCanceled: boolean
}
// Reaction
abstract class Reaction<T> {
readonly options: Options
readonly args: ReadonlyArray<any>
readonly value: T
readonly error: any
readonly stamp: number
readonly isReusable: boolean
configure(options: Partial<Options>): Options
markObsolete(): boolean
pullLastResult(args?: any[]): T | undefined
}
// ReactiveSystem
class ReactiveSystem {
static why(short: boolean = false): string
static getMethodCache<T>(method: F<T>): Cache<T>
static configureCurrentReaction(options: Partial<Options>): Options
static getRevisionOf(obj: any): number
static takeSnapshot<T>(obj: T): T
static dispose(obj: any): void
static reactivityAutoStartDisabled: boolean
static readonly isLogging: boolean
static readonly loggingOptions: LoggingOptions
static setLoggingMode(isOn: boolean, options?: LoggingOptions)
static setLoggingHint<T extends object>(obj: T, name: string | undefined): void
static getLoggingHint<T extends object>(obj: T): string | undefined
static setProfilingMode(isOn: boolean, options?: Partial<ProfilingOptions>): void
}
```
## Contribution
By contributing, you agree that your contributions will be
automatically licensed under the Apache 2.0 license (see LICENSE file).