owner stringclasses 14
values | repo stringclasses 14
values | id int64 116k 4.61B | issue_number int32 1 181k | author stringlengths 1 39 | body stringlengths 1 262k | created_at timestamp[us]date 2000-06-06 02:40:44 2026-06-02 20:16:18 | updated_at timestamp[us]date 2000-06-06 02:40:44 2026-06-02 20:22:56 | reactions unknown | author_association stringclasses 5
values |
|---|---|---|---|---|---|---|---|---|---|
vuejs | core | 645,238,217 | 1,380 | haishengXie0712 | @huangfengya
```
watch(
() => msg1.val,
(val, oldVal) => {
msg1.isOldEqualNew = val.val === oldVal.val
console.log(val === oldVal) // true?
console.log(`new val: ${val}, old val: ${oldVal}`)
}
)
``` | 2020-06-17T08:38:35 | 2020-06-17T08:38:35 | {} | NONE |
vuejs | core | 642,534,848 | 1,344 | rigor789 | Here's a list of the different builds you can use, and when to use which: https://github.com/vuejs/vue-next/tree/master/packages/vue#from-cdn-or-without-a-bundler | 2020-06-11T09:38:56 | 2020-06-11T09:38:56 | {} | CONTRIBUTOR |
vuejs | core | 642,543,090 | 1,342 | underfin | @lawvs .That scene is under`v-if` create comment.This issues really is inconsistent with 2.x. | 2020-06-11T09:56:20 | 2020-06-11T09:56:45 | {} | MEMBER |
vuejs | core | 642,758,618 | 1,262 | yyx990803 | I did a similar fix in https://github.com/vuejs/vue-next/commit/f3623e4d1ea83d552b5ab29955dead6c36a87723 but should be more efficient to avoid double filtering calls. | 2020-06-11T15:48:22 | 2020-06-11T15:48:22 | {} | MEMBER |
vuejs | core | 642,879,584 | 1,340 | yyx990803 | This is expected. `vnodeXXX` hooks is not equivalent to directives. In this case `vnodeXXX` hooks are **invoked on whatever vnode they are used on** - i.e. if it's used on a component, it will be invoked with the **component vnode**.
Directives, on the other hand, are always invoked on the final rendered DOM vnode. | 2020-06-11T19:17:59 | 2020-06-11T19:17:59 | {} | MEMBER |
vuejs | core | 643,849,613 | 1,367 | buns-li | hello, in you test case and the result I get .... :
| 2020-06-15T00:52:13 | 2020-06-15T00:52:13 | {} | NONE |
vuejs | core | 643,895,475 | 1,363 | cuixiaorui | > I think this change would affect the readability. It's not as clear as the original.
why？
This refactoring step comes from the book 《Refactoring》
There are clear reasons for this in Chapter 9.5 of The second edition of Refactoring
| 2020-06-15T04:28:51 | 2020-06-15T04:28:51 | {} | CONTRIBUTOR |
vuejs | core | 642,429,226 | 1,339 | haoqunjiang | It's a legit question. Some of the compatibility layers are not yet implemented in the current beta. | 2020-06-11T06:06:26 | 2020-06-11T06:06:26 | {} | MEMBER |
vuejs | core | 642,543,602 | 1,344 | macdevign | Hi rigor789,
thank for the link.
Just to clarify::
"From CDN or without a Bundler
vue(.runtime).global(.prod).js:
For direct use via <script src="..."> in the browser. Exposes the Vue global.
Note that global builds are not UMD builds. They are built as IIFEs and is only meant for direct u... | 2020-06-11T09:57:32 | 2020-06-11T09:57:32 | {} | NONE |
vuejs | core | 642,921,188 | 1,195 | yyx990803 | Should be fixed after #1106. Thanks for the PR though! | 2020-06-11T20:49:43 | 2020-06-11T20:49:43 | {} | MEMBER |
vuejs | core | 643,467,908 | 1,197 | CyberAP | @yyx990803 there are two kinds of buffer right now: one for `renderToString` and another for `renderToStream`. The methods that renderers expose (`renderComponentSubtree` for example) are bound to buffer constructor and only `renderToString` methods are exposed. That means when runtime helpers are executed they will us... | 2020-06-12T20:23:02 | 2020-06-12T20:24:38 | {} | CONTRIBUTOR |
vuejs | core | 643,601,550 | 1,168 | 5onderling | Got fixed in beta 15. | 2020-06-13T10:07:17 | 2020-06-13T10:08:00 | {} | NONE |
vuejs | core | 644,133,598 | 1,370 | yyx990803 | I think Vue 3 exports the types you need, but note the types are not backwards compatible and may have different names. Please do read the `.d.ts` to make sure. | 2020-06-15T13:25:36 | 2020-06-15T13:25:36 | {} | MEMBER |
vuejs | core | 644,865,572 | 1,379 | yyx990803 | I happened to remove this in e68209bf - actually, based on the test it *should* be removed so that the update is indeed a self update in the child instead of a parent -> child update. | 2020-06-16T16:17:04 | 2020-06-16T16:17:04 | {} | MEMBER |
vuejs | core | 645,193,163 | 1,386 | posva | FYI the same happens with arrays that get pushed/pop/spliced and this is consistent with Vue 2 | 2020-06-17T07:05:31 | 2020-06-17T07:05:31 | {} | MEMBER |
vuejs | core | 645,363,518 | 1,240 | reaink | @yyx990803 @underfin @haishengXie0712 I want to use v-model on [web-components](https://developer.mozilla.org/en-US/docs/Web/Web_Components) Instead of vue-components E.g [here](https://blog.bitsrc.io/9-web-component-ui-libraries-you-should-know-in-2019-9d4476c3f103) Framework UI instead of vue-components, please h... | 2020-06-17T13:10:10 | 2020-06-17T13:10:10 | {} | NONE |
vuejs | core | 642,409,417 | 1,338 | Jinjiang | > The second argument provides a context object which exposes a selective list of properties that were previously exposed on `this` in 2.x APIs:
https://composition-api.vuejs.org/api.html#setup | 2020-06-11T05:06:06 | 2020-06-11T05:06:06 | {} | MEMBER |
vuejs | core | 642,453,505 | 1,339 | tangjinzhou | close it.
I can use beforeUnmount、unmounted replace all beforeDestroy、destroyed. | 2020-06-11T07:02:36 | 2020-06-11T07:02:36 | {} | CONTRIBUTOR |
vuejs | core | 642,542,981 | 1,345 | jacekkarczmarczyk | related? https://github.com/vuejs/vue-next/issues/574 | 2020-06-11T09:56:06 | 2020-06-11T09:56:06 | {} | CONTRIBUTOR |
vuejs | core | 643,358,384 | 1,320 | yyx990803 | We shouldn't have DOM-specific special cases in `runtime-core`. Actually, we just need to mount children before setting the props (which is Vue 2's behavior) | 2020-06-12T16:15:24 | 2020-06-12T16:15:24 | {} | MEMBER |
vuejs | core | 643,486,292 | 1,197 | yyx990803 | > always use a buffer that returns an array and await on deep promises at unroll stage, the same way it's done in renderToStream.
I think that is better - if both use the same buffer implementation then there's also no need to create bound versions of the render methods. | 2020-06-12T21:15:28 | 2020-06-12T21:15:28 | {} | MEMBER |
vuejs | core | 643,764,664 | 1,362 | yyx990803 | 并不觉得心智负担比 ref 少... 甚至可以说更多，因为 1. 有时候能当普通 string 用（+），有时候又不行（===），因为疏忽造成奇怪 bug 的概率大了很多。2. 对静态分析和类型推导不友好。 | 2020-06-14T13:10:24 | 2020-06-14T13:10:24 | {} | MEMBER |
vuejs | core | 643,924,425 | 1,311 | tangjinzhou | @underfin I had to reopen this issue. I updated the code, I think it is a bug of vue3. | 2020-06-15T06:11:55 | 2020-06-15T06:11:55 | {} | CONTRIBUTOR |
vuejs | core | 644,078,460 | 1,370 | underfin | https://github.com/vuejs/vue-next/blob/ee60a0967f56ebd1ff10572f5a506f9d82f4b14c/packages/runtime-core/src/index.ts#L186
Maybe it is helpful for you. | 2020-06-15T11:44:02 | 2020-06-15T11:44:02 | {
"+1": 1
} | MEMBER |
vuejs | core | 644,766,883 | 1,380 | yyx990803 | ^
This is expected behavior. | 2020-06-16T13:30:23 | 2020-06-16T13:30:56 | {} | MEMBER |
vuejs | core | 642,324,798 | 1,335 | mokeyish | @yyx990803
Here is my implemention of `angular like ioc` for vue 2.6x. My freinds and I think it can replace the `vuex`,It is more simplely for using.
https://github.com/mokeyish/vue-ioc-di
How can I implement it in vue3.0? vue 3.0 create a new intance internal. vue3.0 `provide/inject` implemention is too sim... | 2020-06-10T23:50:46 | 2020-06-10T23:50:46 | {} | NONE |
vuejs | core | 642,392,141 | 1,102 | milewski | ... I have added `"skipLibCheck": true,` temporarily to my tsconfig.json until this issue is sorted out | 2020-06-11T04:01:54 | 2020-06-11T04:01:54 | {} | NONE |
vuejs | core | 642,450,863 | 1,339 | tangjinzhou | I can use beforeMount、mounted、updated ...... but beforeDestroy、destroyed ....
Is there a plan to do compatibility ? @sodatea | 2020-06-11T06:56:14 | 2020-06-11T06:56:14 | {} | CONTRIBUTOR |
vuejs | core | 642,576,951 | 1,344 | rigor789 | @macdevign these files are published to npm - and you can already find them on cdn's like [jsdelivr](https://cdn.jsdelivr.net/npm/vue@3.0.0-beta.14/dist/) and [unpkg](https://unpkg.com/browse/vue@3.0.0-beta.14/dist/) | 2020-06-11T11:14:10 | 2020-06-11T11:14:10 | {} | CONTRIBUTOR |
vuejs | core | 642,685,687 | 1,342 | underfin | I track this with 2.x.Find some inconsistent behavior with the blew.It's has a bit of weird.
```
// ""
return this.$createElement("div", [true]);
// "true"
return this.$createElement("div", true);
``` | 2020-06-11T14:16:41 | 2020-06-11T14:16:41 | {} | MEMBER |
vuejs | core | 643,221,319 | 1,350 | posva | this one should be changed because `spellcheck` is not a boolean attribute (https://developer.mozilla.org/en-US/docs/Web/HTML/Global_attributes/spellcheck)
It does work properly on Vue 2 by leaving the value to `false`. The relevant code is at https://github.com/vuejs/vue/blob/dev/src/platforms/web/util/attrs.js#L20 | 2020-06-12T11:28:25 | 2020-06-12T11:32:16 | {
"+1": 4
} | MEMBER |
vuejs | core | 643,258,946 | 1,342 | yyx990803 | Your user doesn't inspect your DOM. | 2020-06-12T13:03:47 | 2020-06-12T13:03:47 | {
"+1": 4,
"-1": 2,
"confused": 1,
"laugh": 4
} | MEMBER |
vuejs | core | 643,336,770 | 1,351 | yyx990803 | When you directly watch a reactive object, it is implicitly deep because it would be pointless if it's not a deep watch. | 2020-06-12T15:31:21 | 2020-06-12T15:31:21 | {
"+1": 2
} | MEMBER |
vuejs | core | 643,392,631 | 1,235 | ozguruysal | I see, thanks for the tip @yyx990803, I got it now and thanks for fixing the issue @underfin. | 2020-06-12T17:19:03 | 2020-06-12T17:19:03 | {} | NONE |
vuejs | core | 643,593,126 | 1,352 | underfin | Look like it is intentional. I fixed other warn for invalid expression with `v-else` on beta.15. | 2020-06-13T08:48:34 | 2020-06-13T08:48:34 | {} | MEMBER |
vuejs | core | 643,865,132 | 1,367 | shengxinjing | forget it . It's not important logic. | 2020-06-15T02:10:26 | 2020-06-15T02:10:26 | {} | CONTRIBUTOR |
vuejs | core | 644,527,185 | 1,380 | HcySunYang | This is consistent with the behavior of Vue2, because the old and new values have the same reference. | 2020-06-16T04:42:01 | 2020-06-16T04:42:01 | {} | MEMBER |
vuejs | core | 644,607,588 | 1,366 | pikax | That's something I do quite a lot, as you said it can be done quite easily on userland.
My only concern here is the naming `RValue`, I don't think `RValue` is good enough name to be exported from the library. | 2020-06-16T08:10:47 | 2020-06-16T08:10:47 | {} | MEMBER |
vuejs | core | 644,608,537 | 1,359 | pikax | This will add quite more complexity to the template code, the solution for your problem would be better type inference in the template, with tools such as `vetur` and others.
| 2020-06-16T08:12:42 | 2020-06-16T08:12:42 | {
"confused": 2
} | MEMBER |
vuejs | core | 428,390,278 | 4 | yyx990803 | @chrisvfritz static class fields are not in the spec yet. Currently in plain ES you can only do static getters.
> Is there a way we can keep everything top-level?
Maybe. Having the options as an object is easier for compatibility reasons. Making them flat involves extracting these flat properties back into an obj... | 2018-10-09T23:45:03 | 2018-10-09T23:45:03 | {} | MEMBER |
vuejs | core | 428,653,477 | 4 | chrisvfritz | > static class fields are not in the spec yet. Currently in plain ES you can only do static getters.
@yyx990803 I might have misunderstood, but I thought we were waiting to release Vue 3 until class features had stabilized. [Field declarations](https://github.com/tc39/proposal-class-fields) and [static class fields]... | 2018-10-10T17:08:57 | 2018-10-10T17:08:57 | {} | CONTRIBUTOR |
vuejs | core | 429,067,454 | 5 | johnleider | I like all of these changes. I felt `nativeOn` made things confusing for users. The fall-through will make it easier to construct components. | 2018-10-11T18:25:34 | 2018-10-11T18:25:34 | {
"hooray": 1
} | NONE |
vuejs | core | 429,649,294 | 7 | yyx990803 | It looks quite fragile/hacky to me, and is essentially rendering the whole app twice... which means it decreases performance significantly... not sure if it's worth it at all.
That said it's totally possible in v3. | 2018-10-14T18:19:15 | 2018-10-14T18:20:31 | {} | MEMBER |
vuejs | core | 429,974,290 | 9 | yyx990803 | I just edited my proposal a bit. When used without an argument the generated props should not care about what key it is.
When used without an argument:
``` html
<input v-model="foo">
```
``` js
h('input', {
modelValue: this.foo,
'onUpdate:modelValue': value => {
this.foo = value
}
})
```
... | 2018-10-15T19:00:02 | 2018-10-15T19:08:33 | {} | MEMBER |
vuejs | core | 432,002,469 | 5 | blake-newman | In that case I think we should mimic React keeping class name and style merging but not passing through attributes on spread to element nodes. This means that you get best of both worlds.
Easy merging of classes and styles and explicit passing of props. | 2018-10-22T21:50:45 | 2018-10-22T21:50:45 | {} | MEMBER |
vuejs | core | 434,883,034 | 13 | LinusBorg | Great, that's what I hoped for 😁 | 2018-10-31T23:35:00 | 2018-10-31T23:35:00 | {} | MEMBER |
vuejs | core | 437,508,687 | 16 | yyx990803 | No, definitely not. Shared reactive state should be using the explicit `observable` API. | 2018-11-09T22:01:41 | 2018-11-09T22:01:41 | {} | MEMBER |
vuejs | core | 429,180,038 | 5 | yyx990803 | Oh, that's only needed if you have your own data that you want to merge into. | 2018-10-12T02:07:58 | 2018-10-12T02:07:58 | {
"+1": 1
} | MEMBER |
vuejs | core | 429,590,565 | 9 | yyx990803 | Actually with v3's flat vnode data implementation, declaring props is unnecessary. When you pass data down it doesn't matter if it's a prop or not.
So the wrapper can simply do
``` vue
<input v-bind="$attrs">
```
Regarding the input event re-emit, I kinda want to avoid this event vs. value mismatch altogethe... | 2018-10-14T02:17:41 | 2018-10-14T02:21:50 | {} | MEMBER |
vuejs | core | 429,887,073 | 9 | yyx990803 | @chrisvfritz yeah, I agree regarding emitting `update:value` instead. #8 also makes sense.
Making `v-model` smart can be tricky, because we can't know at compile time what element the `$attrs` will be spread on in the child component.
One possible implementation: `v-model` generates code like this:
``` html
<... | 2018-10-15T14:56:15 | 2018-10-15T19:07:21 | {} | MEMBER |
vuejs | core | 431,997,144 | 5 | yyx990803 | Right, but I think in those cases you are probably better off declaring the props explicitly. | 2018-10-22T21:35:31 | 2018-10-22T21:35:31 | {} | MEMBER |
vuejs | core | 432,662,268 | 5 | blake-newman | Yes i agree, that is why you must still explicitly ask for `$attrs` to fall through for style and class merging.
This is what I propose which follows React following 2.x API as i'm unsure on the structure for 3.x:
This should be consistent for component fallthrough and element fallthrough
## Case 1
JS:
```j... | 2018-10-24T13:45:23 | 2018-10-24T13:49:33 | {} | MEMBER |
vuejs | core | 442,398,028 | 8 | Jinjiang | I have another idea. Not quite sure it's good enough, but just an idea. 😅
``` js
class Foo extends Vue {
props,
data,
model() {
return {
value, ...others
}
},
...
}
```
In another word, because the update event must be defined in the component to make sense. So maybe it's good ... | 2018-11-28T10:29:22 | 2018-11-28T10:29:22 | {} | MEMBER |
vuejs | core | 443,073,835 | 9 | Jinjiang | Sorry for my bad description. 😅
As we discussed above, now
``` vue
<input v-model="foo">
```
will equal to:
``` js
h('input', {
modelValue: this.foo,
'onUpdate:modelValue': value => {
this.foo = value
}
})
```
not:
``` js
h('input', {
value: this.foo,
'onUpdate:value': value... | 2018-11-30T02:57:54 | 2018-11-30T02:58:25 | {} | MEMBER |
vuejs | core | 429,063,166 | 4 | phanan | > do we actually need a special API for non-reactive properties? When you don't want a property to be reactive, couldn't you just avoid initializing it in data and do so in created/mounted instead?
That's a very good point. IMHO reactivity should be opted-in instead of opted-out. | 2018-10-11T18:14:05 | 2018-10-11T18:14:05 | {} | MEMBER |
vuejs | core | 429,649,551 | 7 | yyx990803 | Also, it seems it cannot have nested components with queries. The walker stops when any component has async data. Still quite a limitation IMO. | 2018-10-14T18:22:42 | 2018-10-14T18:22:42 | {} | MEMBER |
vuejs | core | 431,500,753 | 5 | yyx990803 | @blake-newman not entirely sure if I understand your point... but is the concern about spreading `$attrs` would overwrite existing class and style? In response to that:
- `$attrs` does contain `class` and `style`
- When you do the following, `class` and `style` are automatically merged (same for `v-bind`):
`... | 2018-10-19T21:13:36 | 2018-10-19T21:14:31 | {} | MEMBER |
vuejs | core | 442,395,119 | 8 | Jinjiang | > Should we consider renaming the `input` event to `update:value` when no argument is provided? It would be another breaking change
So how could we bind an event named `update:value` in template without `v-model`? Is `<MyComponent @update:value="xxx" />` or `<MyComponent v-on:update:value="xxx" />` (two colons?) a v... | 2018-11-28T10:19:44 | 2018-11-28T10:19:44 | {} | MEMBER |
vuejs | core | 428,636,935 | 2 | chrisvfritz | > I'm not sure if there will be such use cases that the author want to hand-pick all allowed attributes and declare them inside props so nothing in $attrs will be touched. Technically it is a valid usage but users may come across the warning.
@Justineo Beyond the unnecessary clutter this would add to the component d... | 2018-10-10T16:20:07 | 2018-10-10T16:20:07 | {
"+1": 1
} | CONTRIBUTOR |
vuejs | core | 429,979,827 | 8 | Jinjiang | May I have a question about `v-model` and `.sync` here? 😅
IMO, both `.sync` and `v-model` are a little limited when use case becomes complicated. Because it often require you to do some validation or conversion during the two ways.
When the value is a primitive type, that's OK. But when it's an array like multiple... | 2018-10-15T19:17:55 | 2018-10-15T19:17:55 | {} | MEMBER |
vuejs | core | 433,405,444 | 7 | Akryum | BTW react-apollo switched to rendering the app multiple times: https://github.com/apollographql/react-apollo/pull/2533 | 2018-10-26T13:22:19 | 2018-10-26T13:22:19 | {} | MEMBER |
vuejs | core | 433,910,830 | 12 | yyx990803 | @HerringtonDarkholme Hooks are stateful under the hood. A functional component with hooks is no longer a stateless functional component, it is internally represented as a stateful instance. It actually works fine in a class. On the other hand, not making it available in the idiomatic API makes it useless to a large gro... | 2018-10-29T13:32:21 | 2018-10-29T13:33:28 | {
"+1": 1
} | MEMBER |
vuejs | core | 434,928,471 | 13 | KaelWD | :+1: we use provide/inject pretty much everywhere in vuetify, so keeping the 2.x API (or something like it) is vital for us to be able to update to 3 without a massive rewrite. Also like @znck, most of our usage is in computed properties. | 2018-11-01T04:39:21 | 2018-11-01T04:40:36 | {} | CONTRIBUTOR |
vuejs | core | 435,482,658 | 12 | chrisvfritz | > Could there be a setter for `useComputed`?
@Jinjiang I think it would make sense for `useComputed` to work the same as current computed properties, where they can accept a function or object with `get` and `set` methods:
```js
const double = useComputed({
get: () => data.count * 2
set: newValue => { data... | 2018-11-02T19:22:24 | 2018-11-02T19:22:24 | {} | CONTRIBUTOR |
vuejs | core | 440,409,354 | 17 | posva | I also think the term doesn't completely fit those components.
Regarding removing render, I think, it's almost never done except for components having side effects, so it's better to not have it by default. Usually people writing these kind of components do have a very advanced knowledge on Vue | 2018-11-20T19:56:34 | 2018-11-20T19:56:34 | {} | MEMBER |
vuejs | core | 435,555,190 | 8 | Justineo | > So I'm still in favor of option 2. What do you think?
Not quite sure about it…it also disables usage like `v-bind.sync="obj"`. | 2018-11-03T02:57:38 | 2018-11-25T01:48:19 | {} | MEMBER |
vuejs | core | 428,413,462 | 2 | Jinjiang | > I think the solution is making it a best practice to always spread `$attrs` if you intend to ship your component as a library.
I guess there would be quite amount of mis-operations when:
* write a component which is designed to accept `style` but forget `$attrs`, or
* use a `class` to a fragment component or "... | 2018-10-10T02:08:27 | 2018-10-10T02:20:43 | {} | MEMBER |
vuejs | core | 437,121,993 | 11 | Akryum | PR is up: https://github.com/vuejs/vue/pull/9017 | 2018-11-08T19:17:00 | 2018-11-08T19:17:00 | {} | MEMBER |
vuejs | core | 428,397,155 | 4 | yyx990803 | **Edit: changed options to be flat static properties on the class** | 2018-10-10T00:26:37 | 2018-10-10T00:26:37 | {} | MEMBER |
vuejs | core | 428,818,561 | 4 | chrisvfritz | @yyx990803 As I understand it, one of the main reasons we're defaulting to a class-based API is to prevent the schism where TypeScript users have to use a significantly different interface from everyone else. I agree that's a problem worth solving.
However, if we're unable to sell the benefits of the class-based API... | 2018-10-11T04:41:22 | 2018-10-11T04:41:22 | {
"+1": 2
} | CONTRIBUTOR |
vuejs | core | 429,599,488 | 9 | chrisvfritz | > Actually with v3's flat vnode data implementation, declaring props is unnecessary. When you pass data down it doesn't matter if it's a prop or not.
Yay! 😄
> Regarding the input event re-emit, I kinda want to avoid this event vs. value mismatch altogether. Maybe the code `v-model` generates should just handle ... | 2018-10-14T06:08:35 | 2018-10-14T06:08:35 | {} | CONTRIBUTOR |
vuejs | core | 430,015,272 | 9 | chrisvfritz | > I don't think it makes sense to use v-model with arguments on a plain element, it seems the use case is really for two-way binding on component only. So the runtime special handling will only apply to modelValue and onUpdate:modelValue.
@yyx990803 I agree it doesn't make sense to use `v-model` with arguments on pl... | 2018-10-15T21:13:46 | 2018-10-15T21:13:46 | {} | CONTRIBUTOR |
vuejs | core | 433,404,246 | 7 | Akryum | Maybe we could make `renderToString` asynchronous with ways to hook into it and do data prefetching and eventual secondary re-rerenders of components that prefetched some data? | 2018-10-26T13:18:09 | 2018-10-26T13:18:09 | {} | MEMBER |
vuejs | core | 433,436,205 | 5 | yyx990803 | @blake-newman I'm not sure if I can understand your proposal.
A few notes regarding v3:
Because VNode data is now flat, there's no longer the difference between `data`, `props` and `attrs` when you look at a VNode.
- On a plain element: `class` and `style` has special treatments. Fields that start with `on` ar... | 2018-10-26T14:55:45 | 2018-10-26T20:50:04 | {} | MEMBER |
vuejs | core | 433,780,251 | 12 | HerringtonDarkholme | Even as a traditional `class` fan, I believe hooks are superior and will prevail frontend development.
Great to see it integrated in Vue-next!
My only concern is that should we mix hook with normal class? They are two different mental model and it might be difficult for users to switch between... | 2018-10-29T03:46:59 | 2018-10-29T03:46:59 | {} | MEMBER |
vuejs | core | 440,388,968 | 17 | LinusBorg | I agree with Evan here, I know the term referring to components that render a scoped slot, and usually no markup of their own. They only provide functionality through the slot props and don't render any markup of their own.
The term is not perfect and I think it was coined by the React community when using renderPro... | 2018-11-20T18:51:29 | 2018-11-20T18:51:29 | {
"+1": 1
} | MEMBER |
vuejs | core | 428,632,607 | 2 | Justineo | I'm not sure if there will be such use cases that the author want to hand-pick all allowed attributes and declare them inside `props` so nothing in `$attrs` will be touched. Technically it is a valid usage but users may come across the warning. | 2018-10-10T16:07:38 | 2018-10-10T16:07:38 | {} | MEMBER |
vuejs | core | 428,687,388 | 4 | yyx990803 | @chrisvfritz
Ideally, we want to wait until class fields become part of the spec before releasing 3.0. I'm thinking somewhere around mid 2019, and with them being currently stage 3, it's unlikely for them to make it into ES2019. However, it's hopeful that they are stage 4 around that point, and that would be safe e... | 2018-10-10T18:46:55 | 2018-10-10T19:04:45 | {} | MEMBER |
vuejs | core | 428,885,960 | 4 | Jinjiang | How about removing `data()` and move the job into `constructor` function?
now:
``` js
class App extends Component {
// data
data() {
return { count: 0 }
}
}
```
new:
``` js
class App extends Component {
constructor() {
// data
this.count = 0;
}
}
```
Does it do the sam... | 2018-10-11T09:26:47 | 2018-10-11T09:26:47 | {
"+1": 1
} | MEMBER |
vuejs | core | 428,916,028 | 4 | HerringtonDarkholme | > Constructors can, and often, do more than just initializing the data.
Constructors in general can do a lot of things. But component class isn't general purpose, it is for binding view with data. For component constructors, most of them should do few things (or be dumb). React advises users only initialize state ... | 2018-10-11T11:12:40 | 2018-10-11T11:13:40 | {} | MEMBER |
vuejs | core | 429,053,832 | 4 | yyx990803 | Ok I've managed to make both `data` AND initializers work:
``` ts
// plain ES2015, use `data()` for more concise code
class Foo extends Component {
data() {
return {
count: 1
}
}
}
// TS or with Babel, use initializers for the nice syntax
class Foo extends Component {
count: number =... | 2018-10-11T17:49:23 | 2018-10-11T17:53:43 | {
"+1": 1,
"hooray": 1
} | MEMBER |
vuejs | core | 430,113,944 | 4 | DanielRosenwasser | Sorry, just to clarify for myself, is the idea that plain property declarations with initializers would automatically be marked as reactive? I think that would be **way** better from a UX POV, as well as a type system POV. I guess the downside is that this really could only work in a Proxy-based reactivity model; you c... | 2018-10-16T06:14:51 | 2018-10-16T06:15:12 | {} | CONTRIBUTOR |
vuejs | core | 430,245,730 | 4 | yyx990803 | @DanielRosenwasser yeah, plain initializers would become reactive. I think it's possible to work in older model too, as we are already doing that in `vue-class-component`. | 2018-10-16T13:52:39 | 2018-10-16T13:52:39 | {} | MEMBER |
vuejs | core | 434,154,293 | 12 | HerringtonDarkholme | Mapping current API to hooks looks fantastic 😍 It feels like authentic Vue in hooks and type checks better! (lest lifecycle renaming fuss 😄 ). I'll try with Vue2 to see if it works well.
@posva About minimizing, if we cannot rename properties on object in most compressor, but hooks expose themselves as local var... | 2018-10-30T02:58:14 | 2018-10-30T02:59:45 | {} | MEMBER |
vuejs | core | 434,755,605 | 12 | Jinjiang | Could there be a setter for `useComputed`? And `useData` seems only accept an object or array which is a little constraint for that I think. | 2018-10-31T16:30:01 | 2018-10-31T16:30:01 | {} | MEMBER |
vuejs | core | 434,840,155 | 13 | znck | I share your concerns, I checked all the places I have used provide/inject and in most the injected state is used either in a method or in a computed property. | 2018-10-31T20:41:13 | 2018-10-31T20:41:39 | {} | MEMBER |
vuejs | core | 441,580,408 | 12 | Jinjiang | > > `useData` seems only (to) accept an object or array which is a little constraint for that I think.
>
> @Jinjiang I'm a bit lost here – why is it a constraint? The only other option for `data` is a function, which is, as far as my understanding goes, not relevant or necessary anymore in a hook context as the stat... | 2018-11-26T09:54:23 | 2018-11-26T09:54:23 | {} | MEMBER |
vuejs | core | 434,162,111 | 8 | Justineo | https://github.com/vuejs/vue/issues/4946
This is where we decided to bring `.sync` back for 2.x and the original feature request was `v-model` for multiple values.
Here are some potential problems I can think of:
With the current syntax being `v-model="val"`, the actual prop in the binding is encapsulated, esp... | 2018-10-30T03:48:00 | 2018-10-30T03:51:31 | {} | MEMBER |
vuejs | core | 441,550,118 | 4 | Jinjiang | > Update: don't know since when but latest Chrome Canary now enables class fields by default, so we can already play with the following API without a transpiler:
So the `Data` interface in TS could be removed if it already works, right?
``` ts
import { h, Component, ComponentWatchOptions } from '@vue/renderer-do... | 2018-11-26T08:04:33 | 2018-11-26T08:04:33 | {} | MEMBER |
vuejs | core | 443,069,593 | 8 | chrisvfritz | > So how could we bind an event named update:value in template without v-model? Is <MyComponent @update:value="xxx" /> or <MyComponent v-on:update:value="xxx" /> (two colons?) a valid syntax?
I don't see why we couldn't make it valid, if it isn't already. 🙂 I don't _think_ we currently allow for multiples arguments... | 2018-11-30T02:34:30 | 2018-11-30T02:34:30 | {} | CONTRIBUTOR |
vuejs | core | 428,624,499 | 2 | chrisvfritz | > Maybe it's better to make `$attrs` implicit and print warnings when its a fragment/portal/closed component IMO.
@Jinjiang I really like the idea of a warning for users, but with any implicit pass-through, I really think we're only _delaying_ pain by allowing users to rely on an unstable API that the component auth... | 2018-10-10T15:45:40 | 2018-10-10T15:46:41 | {} | CONTRIBUTOR |
vuejs | core | 429,017,449 | 2 | yyx990803 | Open separate issue for attrs fallthrough behavior: #5 | 2018-10-11T16:12:15 | 2018-10-11T16:12:15 | {} | MEMBER |
vuejs | core | 429,030,574 | 5 | chrisvfritz | Great overview!
> Or maybe $attrs should just point to $props in this case?
I think that makes sense. Then users never have to refactor `v-bind="$props"` to `v-bind="$attrs"` when they start defining props. | 2018-10-11T16:46:48 | 2018-10-11T16:46:48 | {} | CONTRIBUTOR |
vuejs | core | 429,630,312 | 7 | Akryum | Well, that's the approach taken by the official [react-apollo](https://www.apollographql.com/docs/react/features/server-side-rendering.html#getDataFromTree) ([source](https://github.com/apollographql/react-apollo/blob/master/src/getDataFromTree.ts)). BTW it seems so much easier to do in React :sob: | 2018-10-14T14:20:42 | 2018-10-14T14:20:42 | {} | MEMBER |
vuejs | core | 433,432,456 | 11 | yyx990803 | The current `renderToString` implementation actually already handles async component (i.e. waiting on a component before it resolves) so it should be simple to use similar logic to wait on data fetching. We just need to have a convention or internal mechanism for a component to signal that it needs to wait for a preFet... | 2018-10-26T14:44:42 | 2018-10-26T14:44:42 | {
"+1": 1
} | MEMBER |
vuejs | core | 433,907,663 | 12 | posva | Funny thing, the latest version I released for vue-promised uses named imports so it could also export a hook version.
The pattern seems powerful but as you said it's hard for beginners so we have to be careful not to replace existing patterns that are much easier to grasp for newcomers because that's one of Vue goo... | 2018-10-29T13:22:27 | 2018-10-29T13:22:27 | {} | MEMBER |
vuejs | core | 434,101,356 | 12 | yyx990803 | @chrisvfritz yeah, I think we would introduce hooks as an advanced feature for code reuse.
Re `count/setCount`: you are right, if you create an object with `useState()`, you can still directly mutate it and it will work. I added a new paragraph showing how we can provide hooks that maps Vue's current API (and I thin... | 2018-10-29T22:16:28 | 2018-10-29T22:16:28 | {
"+1": 2
} | MEMBER |
vuejs | core | 437,121,824 | 7 | Akryum | Closing in favor of https://github.com/vuejs/vue-next/issues/11 | 2018-11-08T19:16:34 | 2018-11-08T19:16:34 | {} | MEMBER |
vuejs | core | 440,473,573 | 17 | chrisvfritz | @yyx990803 @LinusBorg @posva I think you're thinking of "container" components, but I'm referring to components that use the `provide`/`inject` interface to render non-XML (e.g. Canvas, WebGL, Audio, etc). For example:
```vue
<MapboxMap>
<MapboxMarkers :items="cities" />
<MapboxNavigation />
</MapboxMap>
``... | 2018-11-20T23:59:21 | 2018-11-21T00:02:02 | {
"+1": 2
} | CONTRIBUTOR |
vuejs | core | 428,415,356 | 2 | yyx990803 | > write a component which is designed to accept style but forget $attrs
The best/common practice would be always merging `$attrs` in its entirety. It would be a conscious decision to only allow `style` falling through so I don't think it's possible to "forget".
> use a class to a fragment component or "closed (wi... | 2018-10-10T02:20:39 | 2018-10-10T02:20:39 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.