cdk-virtual-scroll: newly-rendered list items seem to inherit internal state from the _templateCache
- Vorherrschende Sprache
- TypeScript
- Sterne
- 25k
- Forks
- 6.8k
- Ø Merge
- 1 T. 8 Std.
- Gemergte PRs (30 T.)
- 91
Beschreibung
#### What is the expected behavior?
New components rendered by `CdkVirtualForOf` are pristine by default.
#### What is the current behavior?
New components created by `CdkVirtualForOf` within the `` inherit internal state of recycled views by default.
#### What are the steps to reproduce?
https://stackblitz.com/edit/angular-virtual-scrolling-with-component-state
1. Click on *Data Item 3*
1. Scroll down
1. Observe *Data Item 12* is "active"
1. Scroll down
1. Observe *Data Item 23* is "active".
1. Scroll up
1. Observe *Data Item 10* is "active" (a different one, now)
#### Which versions of Angular, Material, OS, TypeScript, browsers are affected?
```
Angular CLI: 7.3.8
Node: 11.12.0
OS: win32 x64
Angular: 7.2.12
... animations, common, compiler, compiler-cli, core, forms
... http, language-service, platform-browser
... platform-browser-dynamic, platform-server, router
... service-worker
Package Version
-----------------------------------------------------------
@angular-devkit/architect 0.13.6
@angular-devkit/build-angular 0.13.6
@angular-devkit/build-optimizer 0.13.6
@angular-devkit/build-webpack 0.13.6
@angular-devkit/core 7.1.4
@angular-devkit/schematics 7.1.4
@angular/cdk 7.3.7
@angular/cli 7.3.8
@angular/pwa 0.11.4
@ngtools/webpack 7.3.6
@schematics/angular 7.1.4
@schematics/update 0.13.8
rxjs 6.4.0
typescript 3.2.4
webpack 4.29.0
```
- Microsoft Windows 10 Pro 10.0.17763 N/A Build 17763
- Chrome Version 73.0.3683.103 (Official Build) (64-bit)
- FireFox Developer Edition 67.0b9 (64-bit)
#### Is there anything else we should know?
Surely this is the very intent of [view recycling](https://github.com/angular/material2/blob/master/src/cdk/scrolling/scrolling.md#view-recycling), but it seems very strange that the default behaviour of `*cdkVirtualFor` is very different from `*ngFor`: my bound parameters changed, the data rendered via the `@Input()` changed, so why didn't the internal state reset?
It's more than _template_ caching, which most of us would take to mean the HTML: it caches whole components and internal state.
Obviously, setting `templateCacheSize: 0` fixes this problem.
Maybe simply highlighting this in the documentation might help, since you have to read between the lines to understand what's happening? Perhaps I'm the only one with this expectation. 🤷♂️
Beitragsleitfaden
Rechercherichtung
Reproduzieren Sie das Durchsickern des Zustands mit dem verknüpften StackBlitz und untersuchen Sie anschließend CdkVirtualForOf sowie das Verhalten der Ansichts-Wiederverwendung von templateCacheSize. Bestimmen Sie, ob wiederverwendete Komponenten ihren internen Zustand zurücksetzen sollten oder ob das Verhalten klarer dokumentiert werden muss, und überprüfen Sie das gewählte Ergebnis anhand der gemeldeten Scroll-Schritte.
Vom Indexierungsmodell aus dem Issue-Text verfasst.
Bewertung
- Tech-Stack
- angular, typescript
- Bereich
- frontend
- Issue-Typ
- Bug
- Schwierigkeit
- 4/5
- Geschätzter Aufwand
- 3-5 Tage
- Aktivitätsstatus
- Veraltet
- Klarheit
- Größtenteils klar
- Anfängerfreundlichkeit
- 35/100