🤖 User test baselines have changed for refs/heads/master

[typescript-bot]

This test run was triggerd by a request on #33716
Please review the diff and merge if no changes are unexpected.
You can view the build log here.

cc @microsoft/typescript

[sandersn]

Found a bug from the vs code errors. I'll continue looking.

[sandersn]

Control flow error is wrong, appears to be using the declared type: #37480

[sandersn]

CSSStyleDeclaration.zoom no longer has null in the type.

[sandersn]

class MsPointerHandler, for IE10 and IE11 compatibility, are now broken.

[sandersn]

office-ui-fabric errors look correct, and I would have suspected external libraries changing if I hadn't tested with an old build of the compiler.

[sandersn]

this error is correct as far as I can tell; one CSS wrapper has a precise type and the other only has 'string'.

[sandersn]

it does disappear when building with dd6811f (from March 2), so office-ui-fabric will have to fix it somehow

[sandersn]

this is the other unique error; haven't figured out what is going on here

[sandersn]

object is no longer assignable to Record<string, ComponentSlotStyleInput<{}, {}>> (because the target has an index signature and the source doesn't -- the Record's value type isn't important.)

This is also correct as far as I can tell, but also disappears with the build from March 2.

[sandersn]

Seems correct, ReturnType<Function> isn't allowed because Function is callable but doesn't have any signatures. I'm not sure why we didn't complain about this before.

It's pretty easy to fix: change

export type ExtractComputedReturns<T extends any> = {
  [key in keyof T]: T[key] extends { get: Function }
    ? ReturnType<T[key]['get']>
    : ReturnType<T[key]>
}

to

export type ExtractComputedReturns<T extends Record<string, (...args:any[]) => any>> = {
    [key in keyof T]: T[key] extends { get: (...args:any[]) => any }
    ? ReturnType<T[key]['get']>
    : ReturnType<T[key]>
}

I'll either contact the vue team or send a PR.

[sandersn]

I'm going to merge this, and send a followup PR to add "node" zone.js' tsconfig

[sandersn]

These errors are correct, and repro on 3.8 as well.

[sandersn]

Well, locally, maxNodeModuleJsDepth: 0 -- when you change it to maxNodeModuleJsDepth: 1, the errors appear in the editor, but otherwise only appear in tsc. Not sure what is going on here.

Aside: why isn't it maxNodeModulesJSDepth to matches node_modules and JS?

[sandersn]

defaultView now has the same type as window

[sandersn]

tsconfig types entry needs to have "node" added

[sandersn]

getUsedName is used 3 times with no arguments, so it's likely intentional -- or perhaps a repeated error.

The usage is odd, though, so it's probably fine to force an optionality annotation.

[sandersn]

other 2 usages of getUsedName

[sandersn]

context object is technically optional although all the other calls provide a starting context

[sandersn]

ha ha spreading a RegExpExecArray will surely never go wrong.

seriously, this is annoying if You Know What You're Doing, but I would feel comfortable making people write a cast for this.

[sandersn]

I looked at uglify; the errors there come from two sources, both of which are intended to be optional but are not documented as such.

[sandersn]

this is clearly intended to be optional; the parameter name is false_by_default

[sandersn]

TreeTransformer takes two functions and the second is indeed optional, but it's not immediately obvious because there's no naming convention and the optionality check is hundreds of lines from the parameter.