Repository navigation
Defer resolution of indexed access types with reducible object types - #53098
Conversation
TypeScript Bot (typescript-bot)
added
Author: Team
For Milestone Bug
|
TypeScript Bot (@typescript-bot) test this |
|
Heya Anders Hejlsberg (@ahejlsberg), I've started to run the abridged perf test suite on this PR at 8334e30. You can monitor the build here. Update: The results are in! |
|
Heya Anders Hejlsberg (@ahejlsberg), I've started to run the parallelized Definitely Typed test suite on this PR at 8334e30. You can monitor the build here. Update: The results are in! |
|
Heya Anders Hejlsberg (@ahejlsberg), I've started to run the diff-based top-repos suite on this PR at 8334e30. You can monitor the build here. Update: The results are in! |
|
Heya Anders Hejlsberg (@ahejlsberg), I've started to run the extended test suite on this PR at 8334e30. You can monitor the build here. |
|
Heya Anders Hejlsberg (@ahejlsberg), I've started to run the diff-based user code test suite on this PR at 8334e30. You can monitor the build here. Update: The results are in! |
|
Anders Hejlsberg (@ahejlsberg) Here are the results of running the user test suite comparing Everything looks good! |
|
Anders Hejlsberg (@ahejlsberg) Here they are:Comparison Report - main..53098
System
Hosts
Scenarios
Developer Information: |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hey Anders Hejlsberg (@ahejlsberg), the results of running the DT tests are ready. |
|
Anders Hejlsberg (@ahejlsberg) Here are the results of running the top-repos suite comparing Everything looks good! |
|
|
||
| // Additional repro from 53030 | ||
|
|
||
| type GetPayload<P extends Payload, K extends keyof P> = P extends { dataType: K } ? P["data"] : never; |
There was a problem hiding this comment.
perhaps it would still be worthwhile to include a test from #51161? that one was even more suspicious as the index access in the true branch was using the same property name as the one in the extendsType. This additional repro is using a different one
| type GetPayload<P extends Payload, K extends keyof P> = P extends { dataType: K } ? P["data"] : never; | |
| type GetPayload<P extends Payload, K extends keyof P> = P extends { dataType: K } ? P["data"] : never; | |
| // repro from 51161 | |
| export type AnyOneof = { oneofKind: string; [k: string]: unknown } | { oneofKind: undefined }; | |
| export type AnyOneofKind<T extends AnyOneof> = T extends { oneofKind: keyof T } | |
| ? T['oneofKind'] | |
| : never; | |
There was a problem hiding this comment.
Sure, I'll add the extra test.
|
TypeScript Bot (@typescript-bot) pack this edit: I thought anyone could request a build... I think this fixes an issue I'm having so I'd like to test it in the playground. |
|
TypeScript Bot (@typescript-bot) pack this |
|
Heya Anders Hejlsberg (@ahejlsberg), I've started to run the tarball bundle task on this PR at 20827a5. You can monitor the build here. |
|
Hey Anders Hejlsberg (@ahejlsberg), I've packed this into an installable tgz. You can install it for testing by referencing it in your and then running There is also a playground for this build and an npm module you can use via |
|
Thanks! Looks like it doesn't resolve my issue, unfortunately. It looks like the same basic pattern to me -- is this expected after the fix? const SomeRpcEndpointId = 'someRpcEndpoint' as const ;
type SomeRpcEndpointId = typeof SomeRpcEndpointId;
interface SomeRpcEndpointActions {
doFoo: {
in: string;
out: number;
};
doBar: {
in: boolean;
out: boolean;
};
}
type AllRpcEndpoints = {
[SomeRpcEndpointId]: SomeRpcEndpointActions;
};
type InvokeRpc = <H extends keyof AllRpcEndpoints, A extends keyof AllRpcEndpoints[H]>(
handler: H,
action: A,
payload: AllRpcEndpoints[H][A]['in'],
) => Promise<AllRpcEndpoints[H][A]['out']>;
declare const invokeRpc: InvokeRpc;
// Despite the compiler errors above, this does type-check correctly.
const result: boolean = await invokeRpc('someRpcEndpoint', 'doBar', true);
// Workaround inspired by https://github.com/microsoft/TypeScript/issues/53030#issuecomment-1448934975.
type GetIn<T> = T extends { in: any } ? T['in'] : never;
type GetOut<T> = T extends { out: any } ? T['out'] : never;
type InvokeRpcWorkaround = <H extends keyof AllRpcEndpoints, A extends keyof AllRpcEndpoints[H]>(
handler: H,
action: A,
payload: GetIn<AllRpcEndpoints[H][A]>,
) => Promise<GetOut<AllRpcEndpoints[H][A]>>;
declare const invokeRpcWorkaround: InvokeRpcWorkaround;
// Type-checks as expected.
const resultWorkaround: boolean = await invokeRpcWorkaround('someRpcEndpoint', 'doBar', true);
export {}; // appease the compiler |
|
Sean Kelley (@seansfkelley) you are looking for a fix for this issue. |
5 participants
This reuses some of the machinery introduced in #46812 to also defer resolution of indexed access types with reducible object types. The additional issue I mention here is actually caused by #46812 always deferring
keyoffor reducible types, making it impossible to explore their constraints. That's fixed in this PR by a new flag ongetIndexTypethat disables deferral.EDIT: Turns out the additional issue is a duplicate of #51161. This also means that this PR supercedes #51176.
Fixes #51161.
Fixes #53030.