Type checker: don't suppress errors while checking expressions #18311
Conversation
❗ Release notes required
|
|
Well - 11 tests, that's much less than I would have expected here :) |
| @@ -51,14 +51,10 @@ neg45.fs(79,28,79,33): typecheck error FS0193: A type parameter is missing a con | |||
|
|
|||
| neg45.fs(79,20,79,22): typecheck error FS0340: The signature and implementation are not compatible because the declaration of the type parameter 'T' requires a constraint of the form 'T :> System.IComparable | |||
|
|
|||
There was a problem hiding this comment.
Sanity check - is this expected? I see that the errors here are duplicates, so it's fine anyway, just asking if that's anyhow logical side effect.
There was a problem hiding this comment.
I did not expect it, but since they're duplicates I thought it was fine to just update the baseline.
| @@ -16,6 +16,7 @@ | |||
| * Add support for C# `Experimental` attribute. ([PR #18253](https://github.com/dotnet/fsharp/pull/18253)) | |||
| * Nullness warnings are issued for signature<>implementation conformance ([PR #18186](https://github.com/dotnet/fsharp/pull/18186)) | |||
| * Symbols: Add FSharpAssembly.IsFSharp ([PR #18290](https://github.com/dotnet/fsharp/pull/18290)) | |||
| * Type checker: don't suppress errors while checking expressions ([PR #18311](https://github.com/dotnet/fsharp/pull/18311)) | |||
There was a problem hiding this comment.
I think we will need better real-life IDE tests and what it does when typing incomplete snippets of code.
I assume there was a reason to suppress typecheck errors in case of parser errors, to prevent a wall of semantic errors jumping around when typing.
I admit the reasons behind it might have changed, but it would be safer if we could either test this more while typing.
Fixes #17787.
I guess many tests may need to be updated. 🙂