Elaborate type of invoked expression in error message #5784

erictsangx · 2015-11-25T05:31:41Z

No description provided.

msftclas · 2015-11-25T05:31:44Z

Hi @erictsangx, I'm your friendly neighborhood Microsoft Pull Request Bot (You can call me MSBOT). Thanks for your contribution!

In order for us to evaluate and accept your PR, we ask that you sign a contribution license agreement. It's all electronic and will take just minutes. I promise there's no faxing. https://cla.microsoft.com.

TTYL, MSBOT;

msftclas · 2015-11-25T05:44:04Z

@erictsangx, Thanks for signing the contribution license agreement so quickly! Actual humans will now validate the agreement and then evaluate the PR.

Thanks, MSBOT;

sandersn · 2015-11-25T20:08:24Z

tests/baselines/reference/unionTypeCallSignatures.errors.txt

    unionOfDifferentParameterTypes();// error - no call signatures
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-!!! error TS2349: Cannot invoke an expression whose type lacks a call signature.
+!!! error TS2658: Cannot invoke an expression whose type is a union with incompatible call signatures.

    var unionOfDifferentNumberOfSignatures: { (a: number): number; } | { (a: number): Date; (a: string): boolean; };
    unionOfDifferentNumberOfSignatures(); // error - no call signatures


Shouldn't this call get the new error too?

sandersn · 2015-11-25T21:49:03Z

This is a nice improvement, although I'd rather see an elaboration of the original error, something like

Cannot invoke an expression whose type lacks a call signature.
  Type '{ (a: number): number } | { (a: string): Date }' has no compatible call signatures.

DanielRosenwasser · 2015-11-26T05:47:56Z

I don't think this is a sufficient check; for instance, you shouldn't report a different error in the following case:

Basically, this is for people who are dumbfounded by the fact that each of their constituent types is callable but the result of the union is not callable. This new error message is even less clear because it throws more jargon into the mix for no reason.

I think if you also checked whether at least one of the constituent types had a call signature, that would be a sufficient condition to issue the new message.

DanielRosenwasser · 2015-11-26T05:48:34Z

Also, can you add the test I showed above if it turns out that my suggested change doesn't affect any tests? Thanks!

erictsangx · 2015-11-26T11:05:39Z

Thank you for the clarification, guys. It seems I was doing it wrong.
I have reworked the code and the error message.
The new error is thrown only if any constituent type is not callable. Otherwise, it should not throw TS2349: Cannot invoke an expression whose type lacks a call signature.

DanielRosenwasser · 2015-11-29T00:05:52Z

src/compiler/checker.ts

+                        let unionCallSignatures: Signature[] = [];
+                        let hasError = false;
+                        forEach(unionApparentType.types, type => {
+                            const eachCallSignature = getSignaturesOfType(type, SignatureKind.Call);


Extract the logic here out to a function called typeHasCallSignatures and place it next to typeHasConstructSignatures. Don't base the new function on typeHasConstructSignatures

Then use (apparentType.flags & TypeFlags.Union && forEach((<UnionType>apparentType).types, typeHasCallSignatures)

erictsangx · 2015-12-02T09:24:00Z

Updated. I will create another pr for Not all types in '{0}' have compatible call signatures. If this pr has no other problems.

DanielRosenwasser · 2015-12-02T10:11:52Z

tests/baselines/reference/constructorOverloads4.errors.txt

@@ -23,7 +23,7 @@ tests/cases/compiler/constructorOverloads4.ts(11,1): error TS2348: Value of type

    (new M.Function("return 5"))();
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-!!! error TS2349: Cannot invoke an expression whose type lacks a call signature.
+!!! error TS2349: Cannot invoke an expression whose type lacks a call signature. Type 'Function' has no compatible call signatures.


This is a regression

Specifically, we're not looking for call signatures here. We're looking for a construct signature, so we're not appropriately reporting an error.

Can you explain the semantics here? I don't think that class Function and function Function will merge (hence the duplicate identifier errors), so the error that changes should be based on only one of the definitions. The first definition has no call signatures, so if the error is based on the first definition, then it looks correct to me.

I must have misread the error span.

DanielRosenwasser · 2015-12-02T10:15:53Z

Actually, never mind my last comment. We can work on this change in this PR.

DanielRosenwasser · 2015-12-02T10:17:44Z

src/compiler/checker.ts

@@ -9438,7 +9443,12 @@ namespace ts {
                    error(node, Diagnostics.Value_of_type_0_is_not_callable_Did_you_mean_to_include_new, typeToString(funcType));
                }
                else {
-                    error(node, Diagnostics.Cannot_invoke_an_expression_whose_type_lacks_a_call_signature);
+                    if (apparentType.flags & TypeFlags.Union && forEach((<UnionType>apparentType).types, typeHasCallSignatures)) {
+                        // TODO check Not all types in '{0}' have compatible call signatures.


Just get rid of this portion entirely and use it in the separate PR. I'd rather not temporarily disable error messages until another PR comes in.

… union of types with call signatures add diagnostic message 2658: Cannot invoke an expression whose type lacks a call signature. Type '{0}' has no compatible call signatures. Closes microsoft#5640

erictsangx · 2015-12-08T10:17:39Z

@DanielRosenwasser updated

sandersn · 2015-12-08T17:18:15Z

👍

mhegazy · 2016-09-14T05:33:36Z

manually merged. thanks

msftclas added the cla-required label Nov 25, 2015

msftclas added cla-signed and removed cla-required labels Nov 25, 2015

sandersn reviewed Nov 25, 2015
View reviewed changes

erictsangx force-pushed the fixUnionTypeCallSignature branch 2 times, most recently from a63f800 to 19f9379 Compare November 26, 2015 11:04

DanielRosenwasser reviewed Nov 29, 2015
View reviewed changes

erictsangx force-pushed the fixUnionTypeCallSignature branch from 19f9379 to 8b3f46c Compare December 2, 2015 09:21

erictsangx force-pushed the fixUnionTypeCallSignature branch from 8b3f46c to 4737d12 Compare December 2, 2015 09:59

DanielRosenwasser reviewed Dec 2, 2015
View reviewed changes

DanielRosenwasser changed the title ~~Improve error message when calling value whose type is a union of types with call signatures~~ Elaborate type of invoked expression Dec 2, 2015

DanielRosenwasser changed the title ~~Elaborate type of invoked expression~~ Elaborate type of invoked expression in error message Dec 2, 2015

DanielRosenwasser reviewed Dec 2, 2015
View reviewed changes

erictsangx added 4 commits December 8, 2015 17:47

Fix issue: Confusing error message when calling value whose type is a…

f645432

… union of types with call signatures add diagnostic message 2658: Cannot invoke an expression whose type lacks a call signature. Type '{0}' has no compatible call signatures. Closes microsoft#5640

add tests for the issue

bc2b105

improvements based on comments

67573be

baseline accept

557433e

erictsangx force-pushed the fixUnionTypeCallSignature branch from 4737d12 to 557433e Compare December 8, 2015 10:16

mhegazy self-assigned this Sep 14, 2016

mhegazy merged commit 557433e into microsoft:master Sep 14, 2016

microsoft locked and limited conversation to collaborators Jun 19, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Elaborate type of invoked expression in error message #5784

Elaborate type of invoked expression in error message #5784

erictsangx commented Nov 25, 2015

msftclas commented Nov 25, 2015

msftclas commented Nov 25, 2015

sandersn Nov 25, 2015

sandersn commented Nov 25, 2015

DanielRosenwasser commented Nov 26, 2015

DanielRosenwasser commented Nov 26, 2015

erictsangx commented Nov 26, 2015

DanielRosenwasser Nov 29, 2015

DanielRosenwasser Nov 29, 2015

erictsangx commented Dec 2, 2015

DanielRosenwasser Dec 2, 2015

DanielRosenwasser Dec 2, 2015

sandersn Dec 8, 2015

DanielRosenwasser Jan 22, 2016

DanielRosenwasser commented Dec 2, 2015

DanielRosenwasser Dec 2, 2015

erictsangx commented Dec 8, 2015

sandersn commented Dec 8, 2015

mhegazy commented Sep 14, 2016

Elaborate type of invoked expression in error message #5784

Elaborate type of invoked expression in error message #5784

Conversation

erictsangx commented Nov 25, 2015

msftclas commented Nov 25, 2015

msftclas commented Nov 25, 2015

Choose a reason for hiding this comment

sandersn commented Nov 25, 2015

DanielRosenwasser commented Nov 26, 2015

DanielRosenwasser commented Nov 26, 2015

erictsangx commented Nov 26, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

erictsangx commented Dec 2, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

DanielRosenwasser commented Dec 2, 2015

Choose a reason for hiding this comment

erictsangx commented Dec 8, 2015

sandersn commented Dec 8, 2015

mhegazy commented Sep 14, 2016