TypeScript Bot (@typescript-bot) pack this
TypeScript Bot (@typescript-bot) test this
TypeScript Bot (@typescript-bot) user test this
TypeScript Bot (@typescript-bot) run dt
TypeScript Bot (@typescript-bot) perf test this

Heya Daniel Rosenwasser (@DanielRosenwasser), I've started to run the extended test suite on this PR at dd07cdc. You can monitor the build here.

Heya Daniel Rosenwasser (@DanielRosenwasser), I've started to run the tarball bundle task on this PR at dd07cdc. You can monitor the build here.

Heya Daniel Rosenwasser (@DanielRosenwasser), I've started to run the parallelized community code test suite on this PR at dd07cdc. You can monitor the build here.

Heya Daniel Rosenwasser (@DanielRosenwasser), I've started to run the perf test suite on this PR at dd07cdc. You can monitor the build here.

Update: The results are in!

Heya Daniel Rosenwasser (@DanielRosenwasser), I've started to run the parallelized Definitely Typed test suite on this PR at dd07cdc. You can monitor the build here.

The user suite test run you requested has finished and failed. I've opened a PR with the baseline diff from master.

Daniel Rosenwasser (@DanielRosenwasser)
The results of the perf run you requested are in!

Comparison Report - main..46266

Download Benchmark

TypeScript Bot (@typescript-bot) perf test this

Heya Anders Hejlsberg (@ahejlsberg), I've started to run the perf test suite on this PR at 76fbe25. You can monitor the build here.

Update: The results are in!

TypeScript Bot (@typescript-bot) perf test this faster

Heya Anders Hejlsberg (@ahejlsberg), I've started to run the abridged perf test suite on this PR at cf546e8. You can monitor the build here.

Update: The results are in!

Heya Anders Hejlsberg (@ahejlsberg), I've started to run the perf test suite on this PR at cf546e8. You can monitor the build here.

Update: The results are in!

Anders Hejlsberg (@ahejlsberg)
The results of the perf run you requested are in!

Comparison Report - main..46266

Download Benchmark

Anders Hejlsberg (@ahejlsberg)
The results of the perf run you requested are in!

Comparison Report - main..46266

Download Benchmark

This is super exciting, thanks for working on this Anders Hejlsberg (@ahejlsberg)! I just ran into this today when working with Result types.

Daniel Rosenwasser (@DanielRosenwasser) maybe it's worth adding this feature to the 4.6 beta blog post:

https://devblogs.microsoft.com/typescript/announcing-typescript-4-6-beta/#control-flow-analysis-for-dependent-parameters

The blog post does not mention object restructuring, only tuple/union destructuring, but I checked the 4.6 nightly playground and it works with objects too now, which is highly anticipated by React users (more than the example showcased in the blog post IMHO)

Edit: oups actually I'm wrong it's not yet 100% good enough for React, and the following still fails:

import React from "react";

type ItemData =
    | { kind: 'user', name: string }
    | { kind: 'company', id: string }

function Item({ kind, ...data }: ItemData) {
    if (kind === 'user') {
        <div>{data.id}</div>
    }
    if (kind === 'company') {
        <div>{data.name}</div>
    }
    return null;
}

But the following now works:

type ItemData2 =
    | { kind: 'user', data: {name: string} }
    | { kind: 'company', data: {id: string} }

function Item2({ kind, data }: ItemData2) {
    if (kind === 'user') {
        <div>{data.name}</div>
    }
    if (kind === 'company') {
        <div>{data.id}</div>
    }
    return null;
}

Still worth highlighting that in the post :)

Seems like the difference there is the spread, right?

Okay, that's a good call, we must have missed that. I will likely add that to the RC post.

Awesome! Thank you so much❤️

Just a question of understanding: Is the support for the spread syntax planned for a bugfix release or for a minor version? Or is it currently not being planned to be changed / given priority?

Luxalpa (@luxalpa) tracked here: #46680

Would also like to see this implemented 😄

I've noticed that this feature doesn't seem to play nice with default values:

type T = {
	flag: true;
	a: number;
} | {
	flag: false;
	a: string;
}

// This code works if you remove the "= true"
({ flag = true, a }: T) => {
	if (flag) {
		a.toFixed(); // thinks it can be either a string or number again
	} else {
		a.toUpperCase(); // same as above
	}
}

Is this also covered by #46680 or is there another reason it doesn't work?

Gareth Chen (@50an6xy06r6n) that feels like a bug

It's a design limitation, but one we could consider removing. We only do CFA for dependent destructured parameters when they're declared without initializers because it adds complexity to consider the contribution the initializer might make to the inferred declared type of the parameter.

Anders Hejlsberg (@ahejlsberg) intuitively it feels like it should work. In my case (react), the way props work means you have to pass parameters as an object, though I realized that there's other ways to set the default value.

This also does not work when destructuring a nested discriminated union type. See playground link here.

I assume this is because of this line in the PR description:

and the parent type for the destructuring is a discriminated union type

Since the object is nested, the parent type being destructured is not the discriminated union. Still, it's a shame this does not work.

I am currently running into issues with this (or a related) analysis. I have a union [A, B] | [C, D] | [E, F] and functions isA() isC() isE() and regardless of destructuring or not, narrowing the first tuple member cannot narrow the second tuple member.

That's a great feature, I personally use it to make global variables dependant on each other:

interface BrowserAPI {
    context: "browser";
    doBrowserThing(): void;
    doNodeThing: never;
}

interface NodeAPI {
    context: "node";
    doNodeThing(): void;
    doBrowserThing: never;
}

declare const api: BrowserAPI | NodeAPI;

const { context, doNodeThing, doBrowserThing } = api;

if (context === "browser") {
    doBrowserThing();
} else {
    doNodeThing();
}

But it lacks one thing necessary for my usecase: a way to use destructuring in a declaration file (.d.ts). I can't actually perform destructuring, because all the API is defined on the global object at run-time.