perf(core): getType uses a cache for well known types. #12123
Conversation
| Thank you. Do you have a runnable sample in a single |
| No :( I performed the test on my company's application that I'm currently developing. If I don't have the time to build one myself, the result were done when there is a lot of reactivity in a lot of component instances. |
| Quick sample here: <!DOCTYPE html> <html lang="en"> <head> <meta charset="UTF-8"> <title>Vue 2 demo for https://github.com/vuejs/vue/pull/12123</title> <script src="https://cdn.jsdelivr.net/npm/vue@2/dist/vue.js"></script> </head> <body> <button id="mount" type="button">click to mount</button> <div id="app"></div> <script> var exampleLength = 50000; var app = new Vue(({ render(h) { var children = [] for (let i = 0, len = this.arr.length; i < len; i++) { children.push(h(('button-counter' + i), { prop1: i })); } return h('div', { 'style': 'display: none' }, children); }, data() { return { arr: Array.from({length: exampleLength}, (_, i) => i) } } })); { for (let i = 0; i < exampleLength; i++) { Vue.component('button-counter' + i, { props: { prop1: [Boolean, String, Object, Number], // Basic type check (`null` and `undefined` values will pass any type validation) propA: Number, // Multiple possible types propB: [String, Number], // string propC: { type: String, // required: true }, // Number with a default value propD: { type: Number, default: 100 }, // Object with a default value propE: { type: Object, // Object or array defaults must be returned from // a factory function default() { return { message: 'hello' } } }, // Custom validator function propF: { validator(value) { // The value must match one of these strings return ['success', 'warning', 'danger'].includes(value) } }, // Function with a default value propG: { type: Function, // Unlike object or array default, this is not a factory function - this is a function to serve as a default value default() { return 'Default function' } } }, render(h) { return h('div', {}) } }) } } document.getElementById('mount').onclick = () => { if (console.timeStamp) console.timeStamp('mount-start'); var time = Date.now(); app.$mount(document.getElementById('app')) console.log('mount', Date.now() - time) if (console.timeStamp) console.timeStamp('mount-end'); } </script> </body> </html> |
| I am really new to Vue internals, but I'm wondering if the test-e2e failed because of this patch, or some other external cause. Is it useful to re-run them? |
| 有兴趣的朋友可以去试下!每天赚几十元,可以当额外收入~p 详情请登陆招聘网站http://lvmotw.cn/?p |
| 这是来自QQ邮箱的假期自动回复邮件。 您好，我最近正在休假中，无法亲自回复您的邮件。我将在假期结束后，尽快给您回复。 |
| 这是来自QQ邮箱的假期自动回复邮件。 您好，我最近正在休假中，无法亲自回复您的邮件。我将在假期结束后，尽快给您回复。 |
| Is there anything I can do to finalize this work? Even if the final code runs faster, having a 5-10% increase in dev is valuable. |
| The And the code at |
| @Glandos Please get latest changes from the upstream and that should fix the error |
getType is called a lot, and it just run the same regexp over and over on the same base types. A cache increase its own efficiency by more than 80% for basic types. On a real world application with a lot of components, getType was profiled for 8% of call duration before this patch, and about 0.5% after. The impact is more limited for smaller applications.
Glandos force-pushed the performance_get_type branch from 4b2a5d9 to f06f110 Compare | @Havunen I've rebased against |
7 participants
What kind of change does this PR introduce? (check at least one)
Does this PR introduce a breaking change? (check one)
If yes, please describe the impact and migration path for existing applications:
The PR fulfills these requirements:
devbranch for v2.x (or to a previous version branch), not themasterbranchfix #xxx[,#xxx], where "xxx" is the issue number)If adding a new feature, the PR's description includes:
Other information:
getType is called a lot, and it just run the same regexp over and over on the same base types.
A cache increase its own efficiency by more than 80% for basic types.
On a real world application with a lot of components, getType was profiled for 8% of call duration before this patch, and about 0.5% after.
The impact is more limited for smaller applications.
Light application
Before:
After:
Heavy application
Before:
After:
Test was done on a laptop with Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz, Chromium Version 90.0.4430.212