Report primitive type in literal-to-primitive relation complaints #38049

JoshuaKGoldberg · 2020-04-19T19:22:11Z

tests/baselines/reference/classPropertyErrorOnNameOnly.errors.txt

DanielRosenwasser · 2020-04-20T19:21:11Z

typescript-bot · 2020-04-20T19:21:17Z

Heya @DanielRosenwasser, I've started to run the tarball bundle task on this PR at 035b9ac. You can monitor the build here.

DanielRosenwasser · 2020-04-20T19:26:11Z

Do we have a regression for something like this?

let x: "hello" | "world";
let y: "world" | number;
y = x;

My fear is that this gives an error for string is not assignable to string | number but GitHub makes it so hard to review the changes.

typescript-bot · 2020-04-20T19:37:21Z

Hey @DanielRosenwasser, I've packed this into an installable tgz. You can install it for testing by referencing it in your package.json like so:

{
    "devDependencies": {
        "typescript": "https://typescript.visualstudio.com/cf7ac146-d525-443c-b23c-0d58337efebc/_apis/build/builds/71654/artifacts?artifactName=tgz&fileId=4F84A2EE356018F5BE2BA2B18081291312BF84EC551EC982861945C22B635B1802&fileName=/typescript-3.9.0-insiders.20200420.tgz"
    }
}

and then running npm install.

There is also a playground for this build.

JoshuaKGoldberg · 2020-04-20T20:39:37Z

Ah, correct, that is a regression. I can take that on soon!

JoshuaKGoldberg · 2020-04-21T03:36:17Z

d12b741 excludes type unions from either the source or target types in reporting. Looks like that extends through reporting chains, where either can be unions.

DanielRosenwasser · 2020-04-21T18:57:05Z

tests/baselines/reference/bigintIndex.errors.txt

@@ -1,6 +1,6 @@
 tests/cases/compiler/a.ts(2,6): error TS1023: An index signature parameter type must be either 'string' or 'number'.
 tests/cases/compiler/a.ts(8,11): error TS2538: Type '1n' cannot be used as an index type.
-tests/cases/compiler/a.ts(14,1): error TS2322: Type 'bigint' is not assignable to type 'string | number | symbol'.
+tests/cases/compiler/a.ts(14,1): error TS2322: Type '123n' is not assignable to type 'string | number | symbol'.


Why did this happen?

Oh, the target is a union.

src/compiler/checker.ts

DanielRosenwasser · 2020-04-22T19:08:42Z

tests/baselines/reference/complicatedIndexedAccessKeyofReliesOnKeyofNeverUpperBound.errors.txt

-                      Type '"text"' is not assignable to type 'T'.
+                Type 'string' is not assignable to type 'ChannelOfType<T, TextChannel>["type"]'.
+                  Type 'string' is not assignable to type 'ChannelOfType<T, TextChannel>["type"]'.
+                    Type 'string' is not assignable to type 'T & "text"'.


DanielRosenwasser · 2020-04-22T19:09:23Z

tests/baselines/reference/conditionalTypes1.errors.txt

          '0' is assignable to the constraint of type 'T', but 'T' could be instantiated with a different subtype of constraint 'string | number'.
-            Type '"" | 0' is not assignable to type 'T'.
+            Type 'string | number' is not assignable to type 'T'.
              '"" | 0' is assignable to the constraint of type 'T', but 'T' could be instantiated with a different subtype of constraint 'string | number'.


DanielRosenwasser · 2020-04-22T19:44:29Z

tests/baselines/reference/genericCallWithFunctionTypedArguments.errors.txt

 !!! error TS2345:       'T' could be instantiated with an arbitrary type which could be unrelated to '1'.
        var r11b = foo3(1, (x: T) => '', 1); // error
                           ~~~~~~~~~~~~
 !!! error TS2345: Argument of type '(x: T) => string' is not assignable to parameter of type '(a: 1) => string'.
 !!! error TS2345:   Types of parameters 'x' and 'a' are incompatible.
-!!! error TS2345:     Type '1' is not assignable to type 'T'.
+!!! error TS2345:     Type 'number' is not assignable to type 'T'.
 !!! error TS2345:       'T' could be instantiated with an arbitrary type which could be unrelated to '1'.


DanielRosenwasser

Okay, one thing I've noticed is that this doesn't work with weird intersections of singleton types and with generics constrained to the type of interest. Maybe instead of forEachType(target, isUnitType), try using the following helper function.

function typeCouldHaveNoTopLevelSingletonTypes(type: Type) {
    return forEachType(type, worker);
}

function typeCouldHaveNoTopLevelSingletonTypesWorker(type: Type) {
    return (type.flags & TypeFlags.Intersection)
        ? forEach((type as IntersectionType).types, typeCouldHaveNoTopLevelSingletonTypesWorker)
        : isUnitType(type) || !!(type.flags & TypeFlags.Instantiable);
}

DanielRosenwasser · 2020-04-28T06:18:27Z

src/compiler/checker.ts

+                return forEachType(type, typeCouldHaveNoTopLevelSingletonTypesWorker);
+            }
+
+            function typeCouldHaveNoTopLevelSingletonTypesWorker(type: Type): boolean {


Okay, I think we're getting there. First, I would move these functions out so that they don't add more to the current closure on every call.

Second, I think we have to go deeper on constraints, and we can get rid of the "Worker" function.

function typeCouldHaveNoTopLevelSingletonTypes(type: Type): boolean { if (type.flags & TypeFlags.UnionOrIntersection) { return !!forEach((type as IntersectionType).types, typeCouldHaveNoTopLevelSingletonTypes); } if (type.flags & TypeFlags.Instantiable) { const constraint = getConstraintOfType(type); if (constraint) { return typeCouldHaveNoTopLevelSingletonTypes(constraint); } } return isUnitType(type); }

tests/baselines/reference/constructorImplementationWithDefaultValues2.errors.txt

sandersn · 2020-05-21T17:37:20Z

@JoshuaKGoldberg Do you want to keep working on this? It looks like this just needs a small refactoring suggested by @DanielRosenwasser (although I didn't understand his comment "I think we have to go deeper on constraints".)

JoshuaKGoldberg · 2020-05-21T17:40:01Z

I'd be fine dropping this PR & letting you / someone else pick it up.

I was planning on spending some time to really go through and understand the constraints & reportings here, which are a good bit deeper than I realized going in... it'd be better IMO if the PR gets merged before I do that (and I have other PRs that are actually ready for review) 😄.

…e-relation-reporting

DanielRosenwasser · 2020-05-22T00:42:28Z

I'll take it on

DanielRosenwasser · 2020-05-26T20:14:44Z

tests/baselines/reference/extractInferenceImprovement.errors.txt

@@ -32,7 +32,7 @@ tests/cases/compiler/extractInferenceImprovement.ts(28,26): error TS2345: Argume
    // Should fail
    prop = getProperty2(obj, s);
                             ~
-!!! error TS2345: Argument of type 'unique symbol' is not assignable to parameter of type 'never'.
+!!! error TS2345: Argument of type 'typeof s' is not assignable to parameter of type 'never'.


Hm, this one definitely feels arguable. Not sure if @rbuckton has strong feelings about it, but I don't think it's a blocker.

JoshuaKGoldberg added 2 commits April 19, 2020 13:26

Report primitive type in literal-to-primitive relation complaints

b14b231

Fixed a couple of tsserver unit tests

af4ea60

JoshuaKGoldberg commented Apr 19, 2020

View reviewed changes

tests/baselines/reference/classPropertyErrorOnNameOnly.errors.txt Show resolved Hide resolved

JoshuaKGoldberg added 4 commits April 19, 2020 20:25

Eased up on some of the more aggressive reporting changes

769a2f2

Again fixed up manual unit tests

cb19b3d

This had better be the last unit test fix...

1b3c7a2

Lint fix

035b9ac

Excluded type unions from the generale source types

d12b741

DanielRosenwasser reviewed Apr 21, 2020

View reviewed changes

DanielRosenwasser requested changes Apr 21, 2020

View reviewed changes

src/compiler/checker.ts Outdated Show resolved Hide resolved

Used DH's suggested heuristic

99cdb64

JoshuaKGoldberg requested a review from DanielRosenwasser April 22, 2020 18:37

DanielRosenwasser reviewed Apr 22, 2020

View reviewed changes

I did everything you said I followed all the rules

d4825ab

JoshuaKGoldberg requested a review from DanielRosenwasser April 28, 2020 00:03

DanielRosenwasser reviewed Apr 28, 2020

View reviewed changes

tests/baselines/reference/constructorImplementationWithDefaultValues2.errors.txt Outdated Show resolved Hide resolved

sandersn assigned DanielRosenwasser May 5, 2020

sandersn added the For Backlog Bug PRs that fix a backlog bug label May 5, 2020

DanielRosenwasser added 3 commits May 22, 2020 00:09

Merge remote-tracking branch 'origin/master' into literal-to-primitiv…

8cf2466

…e-relation-reporting

Check constraints deeply on singleton types.

a04ecb5

Updated baselines.

b38b0c7

DanielRosenwasser added 2 commits May 22, 2020 01:02

Feed generalized source type to elaboration.

afe90f1

Updated baselines.

a008792

DanielRosenwasser requested a review from sandersn May 22, 2020 01:13

DanielRosenwasser approved these changes May 22, 2020

View reviewed changes

Less code

d3d282c

DanielRosenwasser reviewed May 26, 2020

View reviewed changes

DanielRosenwasser merged commit f2d6eda into microsoft:master May 26, 2020

DanielRosenwasser mentioned this pull request May 26, 2020

Report base primitive sources against boolean targets #38792

Merged

dfreeman mentioned this pull request Jul 1, 2020

Get CI passing typed-ember/ember-cli-typescript#1167

Merged

JoshuaKGoldberg deleted the literal-to-primitive-relation-reporting branch August 15, 2020 00:40

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Report primitive type in literal-to-primitive relation complaints #38049

Report primitive type in literal-to-primitive relation complaints #38049

JoshuaKGoldberg commented Apr 19, 2020

DanielRosenwasser commented Apr 20, 2020

typescript-bot commented Apr 20, 2020 •

edited

Loading

DanielRosenwasser commented Apr 20, 2020 •

edited

Loading

typescript-bot commented Apr 20, 2020 •

edited

Loading

JoshuaKGoldberg commented Apr 20, 2020

JoshuaKGoldberg commented Apr 21, 2020

DanielRosenwasser Apr 21, 2020

DanielRosenwasser Apr 21, 2020

DanielRosenwasser Apr 22, 2020

DanielRosenwasser Apr 22, 2020

DanielRosenwasser Apr 22, 2020

DanielRosenwasser left a comment

DanielRosenwasser Apr 28, 2020

sandersn commented May 21, 2020

JoshuaKGoldberg commented May 21, 2020

DanielRosenwasser commented May 22, 2020

DanielRosenwasser May 26, 2020

Report primitive type in literal-to-primitive relation complaints #38049

Report primitive type in literal-to-primitive relation complaints #38049

Conversation

JoshuaKGoldberg commented Apr 19, 2020

DanielRosenwasser commented Apr 20, 2020

typescript-bot commented Apr 20, 2020 • edited Loading

DanielRosenwasser commented Apr 20, 2020 • edited Loading

typescript-bot commented Apr 20, 2020 • edited Loading

JoshuaKGoldberg commented Apr 20, 2020

JoshuaKGoldberg commented Apr 21, 2020

DanielRosenwasser Apr 21, 2020

Choose a reason for hiding this comment

DanielRosenwasser Apr 21, 2020

Choose a reason for hiding this comment

DanielRosenwasser Apr 22, 2020

Choose a reason for hiding this comment

DanielRosenwasser Apr 22, 2020

Choose a reason for hiding this comment

DanielRosenwasser Apr 22, 2020

Choose a reason for hiding this comment

DanielRosenwasser left a comment

Choose a reason for hiding this comment

DanielRosenwasser Apr 28, 2020

Choose a reason for hiding this comment

sandersn commented May 21, 2020

JoshuaKGoldberg commented May 21, 2020

DanielRosenwasser commented May 22, 2020

DanielRosenwasser May 26, 2020

Choose a reason for hiding this comment

typescript-bot commented Apr 20, 2020 •

edited

Loading

DanielRosenwasser commented Apr 20, 2020 •

edited

Loading

typescript-bot commented Apr 20, 2020 •

edited

Loading