
Security News
vlt Launches "reproduce": A New Tool Challenging the Limits of Package Provenance
vlt's new "reproduce" tool verifies npm packages against their source code, outperforming traditional provenance adoption in the JavaScript ecosystem.
@benev/slate
Advanced tools
@benev/slate
๐ง prerelease wip under constructon subject to change
you're not supposed to make your whole app out of web components.
they're too cumbersome โ web components are html-native, not typescript-native โ so they don't take typesafe props, they're referred to by html tag names with poor ide support, and juggling their dom registrations is a pain.
this is why views are important, and are a central feature of slate โ views are almost the same as components (they can use shadow-dom), except that views are ergonomically handled via javascript, they accept props, they don't need registration (they're simply imported) โ and views enjoy full typescript typings.
you want to think of web components as the tip of your iceberg โ they are the entrypoints to your ui โ they are the universal control surfaces to help html authors interact with your systems โ but below the surface, most of your internals can be made of easily-composable views.
npm i @benev/slate
nexus
import {Nexus, Context} from "@benev/slate"
export const nexus = new Nexus(
new class extends Context {
// this theme is applied to all your components and views
theme = css`
* {
margin: 0;
padding: 0;
box-sizing: border-box;
}
`
// add app-level stuff you'd like to make widely available
my_cool_thing = {my_awesome_data: 123}
}
)
lit
's templating functions, which directly implement signals
.import {html, css, svg} from "@benev/slate"
you can create custom html elements that work in plain html or any web framework.
nexus.shadow_component
export const MyShadowComponent = nexus.shadow_component(use => {
use.styles(css`span {color: yellow}`)
const count = use.signal(0)
const increment = () => count.value++
return html`
<span>${count}</span>
<button @click=${increment}>increment</button>
`
})
nexus.light_component
export const MyLightComponent = nexus.light_component(use => {
const count = use.signal(0)
const increment = () => count.value++
return html`
<span>${count}</span>
<button @click=${increment}>increment</button>
`
})
import {register_to_dom} from "@benev/slate"
register_to_dom({
MyShadowComponent,
MyLightComponent,
})
<section>
<my-shadow-component></my-shadow-component>
<my-light-component></my-light-component>
</section>
views are just like components, but are not registered to the dom as custom html elements.
instead, they are used via javascript.
you import them, and inject them into your lit-html templates.
they accept js parameters called props
, and are fully typescript-typed.
nexus.shadow_view
export const MyShadowView = nexus.shadow_view(use => (start: number) => {
use.name("my-shadow-view")
use.styles(css`span {color: yellow}`)
const count = use.signal(start)
const increment = () => count.value++
return html`
<span>${count}</span>
<button @click=${increment}>increment</button>
`
})
auto_exportparts
is enabled by default.
part
attribute, then it will automatically re-export all internal parts, using the part as a prefix.::part(search-input-icon)
.nexus.light_view
export const MyLightView = nexus.light_view(use => (start: number) => {
use.name("my-light-view")
const count = use.signal(start)
const increment = () => count.value++
return html`
<span>${count}</span>
<button @click=${increment}>increment</button>
`
})
html`<div>${MyShadowView([123])}</div>`
html`
<div>
${MyShadowView([123], {
content: html`<p>slotted content</p>`,
auto_exportparts: true,
attrs: {part: "cool", "data-whatever": true},
})}
</div>
`
html`<div>${MyLightView(123)}</div>`
<slate-view view="my-name">
componentuse
hooks โ for views and componentsslate's hooks have the same rules as any other framework's hooks: the order that hooks are executed in matters, so you must not call hooks under an if
statement or in any kind of for
loop or anything like that.
use.name("my-cool-view")
use.styles(css`span { color: yellow }`)
const [count, setCount] = use.state(0)
const increment = () => setCount(count + 1)
const random_number = use.once(() => Math.random())
use.mount(() => {
const interval = setInterval(increment, 1000)
return () => clearInterval(interval)
})
const scene = use.init(() => {
// called whenever dom is connected
const scene = setup_3d_scene_for_example()
return [
scene, // value returned
() => scene.cleanup(), // cleanup called on dom disconnect
]
})
use.defer(() => {
const div = document.querySelector("div")
const rect = div.getBoundingClientRect()
report_rect(rect)
})
const count = use.signal(0)
const increment = () => count.value++
you can directly inject the whole signal into html
html`<span>${count}</span>`
const count = use.signal(2)
const tripled = use.computed(() => count.value * 3)
console.log(tripled.value) //> 6
const count = use.op()
count.load(async() => fetchCount("/count"))
const state = use.flatstate({count: 0})
const increment = () => state.count++
watch.stateTree({})
.
const whatever = use.watch(() => use.context.state.whatever)
these are not hooks, just access to useful things you may need, so you're allowed to use them under if statements or whatever.
// access your own things on the context
use.context.my_cool_thing
use.element.querySelector("p")
use.shadow.querySelector("slot")
const attrs = use.attrs({
start: Number,
label: String,
["data-active"]: Boolean,
})
set them like normal js properties
attrs.start = 123
attrs.label = "hello"
attrs["data-active"] = true
get them like normal js properties
console.log(attrs.start) // 123
console.log(attrs.label) // "hello"
console.log(attrs["data-active"]) // true
components rerender when any attributes change from outsidegold and silver are "plain" elements, which are alternatives to LitElement.
they're used as primitives underlying nexus components.
for most cases you probably want to stick with the nexus components, and only use gold/silver when you're doing some funky sorcery, or you yearn to go back to a simpler time without hooks.
consider these imports for the following examples:
import {GoldElement, SilverElement, attributes, flat} from "@benev/slate"
export class MyGold extends GoldElement {
static get styles() { return css`span {color: blue}` }
#attrs = attributes(this as GoldElement, {
label: String
})
#state = flat.state({
count: 0,
})
render() {
return html`
<span>${this.#state.count}</span>
<button @click=${() => this.#state.count++}>
${this.#attrs.label}
</button>
`
}
}
export class MySilver extends SilverElement {
#attrs = attributes(this as SilverElement, {
label: String
})
#state = flat.state({
count: 0,
})
render() {
return html`
<span>${this.#state.count}</span>
<button @click=${() => this.#state.count++}>
${this.#attrs.label}
</button>
`
}
}
if you want plain elements to have reactivity or have the context's css theme applied, you'll want to run them through nexus.components
before you register them:
register_to_dom({
...nexus.components({
MyGold,
MySilver,
}),
})
you can extend the context with anything you'd like to make easily available to your components and views:
export const nexus = new Nexus(
new class extends Context {
my_cool_thing = {my_awesome_data: 123}
}
)
but since your components are importing nexus
, the above example creates the context at import-time.
you may instead prefer to defer the creation of your context until later, at run-time:
// define your context class
export class MyContext extends Context {
my_cool_thing = {my_awesome_data: 123}
}
// create nexus *without yet* instancing the context
export const nexus = new Nexus<MyContext>()
//
// ... later in another file,
// maybe in your main.ts ...
//
// instance and assign your context, now, at runtime
nexus.context = new MyContext()
// just be sure to assign context *before* you register your components
register_to_dom(myComponents)
if you're using nexus components and views, you'll probably be using these utilities via the use
hooks, which will provide a better developer experience.
however, the following utilities are little libraries in their own right, and can be used in a standalone capacity.
signals are a simple form of state management.
this implementation is inspired by preact signals.
import {signal, signals} from "@benev/slate"
const count = signal(0)
const greeting = signal("hello")
count.value++
greeting.value = "bonjour"
console.log(count.value) //> 1
console.log(greeting.value) //> "bonjour"
signals.reaction(() => console.log("doubled", count.value * 2))
//> doubled 2
count.value = 2
//> doubled 4
html`<p>count is ${count}</p>`
const json = signals.op<MyJson>()
console.log(json.isLoading()) //> true
await json.load(async() => {
const data = await fetch_remote_data()
return JSON.parse(data)
})
console.log(json.isReady()) //> true
console.log(json.payload) //> {"your": "json data"}
count.value = 1
const tripled = signals.computed(() => count.value * 3)
console.log(tripled.value) //> 3
const tripled = signals.computed(() => count.value * 3)
console.log(tripled.value) //> 3
count.value = 10
console.log(tripled.value) //> 3 (too soon!)
await signals.wait
console.log(tripled.value) //> 30 (there we go)
import {SignalTower} from "@benev/slate"
const signals = new SignalTower()
signals
, but you can create your ownflatstate help you create state objects and react when properties change.
flatstate is inspired by mobx and snapstate, but designed to be simpler. flatstate only works on flat state objects. only the direct properties of state objects are tracked for reactivity. this simplicity helps us avoid weird edge-cases or unexpected footguns.
import {flat} from "@benev/slate"
const state = flat.state({count: 0})
flat.reaction(() => console.log(state.count))
flat.reaction(
// your "collector" function
() => ({count: state.count}),
// your "responder" function
({count}) => console.log(count),
)
const stop = flat.reaction(() => console.log(state.count))
stop() // end this particular reaction
const state = flat.state({amount: 100})
state.amount = 101
console.log(state.amount) //> 100 (old value)
await flat.wait
console.log(state.amount) //> 101 (now it's ready)
const state = flat.state({count: 0})
const rstate = Flat.readonly(state)
state.count = 1
await flat.wait
console.log(rstate.count) //> 1
rstate.count = 2 // !! ReadonlyError !!
const flat1 = new Flat()
const flat2 = new Flat()
import {apply} from "@benev/slate"
const MyElement2 = mixin.flat(flat)(MyElement)
// can also be a class decorator
const elements2 = apply.flat(flat)(elements)
create reactions that listen to both signals and flatstates at the same time.
signals and flat both share the same reaction syntax, but they are separate state management systems. reactor
lets you combine both.
import {reactor, flatstate, signal} from "@benev/slate"
const state = state({count: 0})
const count = signal(0)
// use the reactor to setup a reaction
reactor.reaction(() => console.log(`
flat count is ${state.count},
signal count is ${count.value}
`))
reactor.reaction(
() => ({a: state.count, b: count.value}),
results => console.log(results),
)
const stop = reactor.reaction(
() => console.log(state.count)
)
// end this reaction
stop()
await reactor.wait
utility for ui loading/error/ready states.
useful for implementing async operations that involve loading indicators.
you get a better dev-experience if you use ops via signals, but here is the documentation for plain ops on their own, without signals.
import {Op} from "@benev/slate"
Op.loading()
//= {status: "loading"}
Op.error("a fail occurred")
//= {status: "error", reason: "a fail occurred"}
Op.ready(123)
//= {status: "ready", payload: 123}
Op.is.loading(op)
//= false
Op.is.error(op)
//= false
Op.is.ready(op)
//= true
const count = Op.ready(123)
const loadingCount = Op.loading()
Op.payload(count)
//= 123
Op.payload(loadingCount)
//= undefined
let my_op = Op.loading()
await Op.load(
// your setter designates which op to overwrite
op => my_op = op,
// your async function which returns the ready payload
async() => {
await nap(1000)
return 123
}
)
signals.op()
or use.op()
, the OpSignal these return has nicer ergonomics
const count = signals.op()
// run an async operation
await count.load(async() => {
await sleep(1000)
return 123
})
// check the status of this OpSignal
count.isLoading() //= false
count.isError() //= false
count.isReady() //= true
// grab the payload (undefined when not ready)
count.payload //= 123
// directly assign the op signal
count.setLoading()
count.setError("big fail")
count.setReady(123)
// bad
register_to_dom(
apply.signals(signals)(
apply.flat(flat)(
apply.css(theme)(
requirement.provide(context)(elements)
)
)
)
)
import {Pipe} from "@benev/slate"
// good
Pipe.with(elements)
.to(requirement.provide(context))
.to(apply.css(theme))
.to(apply.flat(flat))
.to(apply.signals(signals))
.to(register_to_dom)
.done()
when you want to return the resultno time to document these fully, but they're there
debounce
โ is a pretty good debouncerdeep
โ utilities for data structures like 'equal' and 'freeze'is
โ proper type guardsexplode_promise
โ make an inside-out promisegenerate_id
โ generate a crypto-random hexadecimal id stringpub
โ easy pub/sub toolrequirement
โ pass required data to a group of thingsv0.1.0
nexus rewrite
Slate
renamed to Nexus
ObsidianRenderer
to ShadowViewRenderer
CarbonRenderer
to ShadowComponentRenderer
QuartzRenderer
to LightViewRenderer
OxygenRenderer
to LightComponentRenderer
shadow views and components
slate.shadow_view({name: "coolview", styles}, use => () => {
return html`hi`
})
slate.shadow_view(use => () => {
use.name("coolview")
use.styles(styles)
return html`hi`
})
use
use.setup
renamed to use.mount
use.prepare
renamed to use.once
SetupFn
to Mount
Setdown
to Unmount
InitFn
to Init
setupFn
to mountFn
use.effect(fn, dependencies)
(all views and components)use.defer(fn)
(all views and components)use.name(name)
(all views)use.styles(styles)
(shadow_view and shadow_component)views
<obsidian-view>
renamed to <slate-view>
<slate-view>
ops (and OpSignal)
Op.run
to Op.load
OpSignal->run
to OpSignal->load
Op.load
and OpSignal->load
now throws errors
Op.Mode
to Op.Status
op.mode
to op.status
on your op objectsop.loading
to op.isLoading()
op.error
to op.isError()
op.ready
to op.isReady()
Op.is.loading
to have proper ts type guardstools
MapSubset
type to MapBase
maptool(map).grab(..)
to maptool(map).guarantee(..)
mapGuarantee(map, k, v)
function, technically more efficient than maptool(map).guarantee(k, v)
ob
tool syntax
ob.map(object, transform)
becomes ob(object).map(transform)
ob.filter(object, predicate)
becomes ob(object).filter(predicate)
deepEqual
and deepFreeze
into new deep
tool
deepEqual
becomes deep.equal
deepFreeze
becomes deep.freeze
is
tool
is.object(x)
is.array(x)
is.defined(x)
watch
watch.track
now returns an unsubscribe function, instead of the collector's data
Historian
and AppCore
for creating apps with undo/redo capabilitiesFAQs
frontend web stuff
The npm package @benev/slate receives a total of 185 weekly downloads. As such, @benev/slate popularity was classified as not popular.
We found that @benev/slate demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago.ย It has 0 open source maintainers collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
vlt's new "reproduce" tool verifies npm packages against their source code, outperforming traditional provenance adoption in the JavaScript ecosystem.
Research
Security News
Socket researchers uncovered a malicious PyPI package exploiting Deezerโs API to enable coordinated music piracy through API abuse and C2 server control.
Research
The Socket Research Team discovered a malicious npm package, '@ton-wallet/create', stealing cryptocurrency wallet keys from developers and users in the TON ecosystem.