Fix overload resolution of static member extension

docs/release-notes/.FSharp.Compiler.Service/11.0.100.md

@@ -48,6 +48,7 @@
 * Fix signature generation: SRTP constraints use postfix syntax that fails conformance, now uses explicit type param declarations. ([Issue #19594](https://github.com/dotnet/fsharp/issues/19594), [PR #19609](https://github.com/dotnet/fsharp/pull/19609))
 * Fix signature generation: type params with special characters missing backtick escaping. ([Issue #19595](https://github.com/dotnet/fsharp/issues/19595), [PR #19609](https://github.com/dotnet/fsharp/pull/19609))
 * Fix internal error when using custom attribute with `[<Optional>]` value type parameter and no `[<DefaultParameterValue>]`. ([Issue #8353](https://github.com/dotnet/fsharp/issues/8353), [PR #19484](https://github.com/dotnet/fsharp/pull/19484))
+* Fix overload resolution of static member extension if one or more intrinsics candidates exist ([Issue #19664](https://github.com/dotnet/fsharp/issues/19664), [PR #19698](https://github.com/dotnet/fsharp/pull/19698))
 * Fix parallel compilation of scripts ([PR #19649](https://github.com/dotnet/fsharp/pull/19649))
 
 ### Added

src/Compiler/Checking/NameResolution.fs

@@ -4164,8 +4164,10 @@ let ResolveNestedField sink (ncenv: NameResolver) nenv ad recdTy lid =
 /// determine any valid members
 //
 // QUERY (instantiationGenerator cleanup): it would be really nice not to flow instantiationGenerator to here.
-let private ResolveExprDotLongIdent (ncenv: NameResolver) m ad nenv ty (id: Ident) rest (typeNameResInfo: TypeNameResolutionInfo) findFlag maybeArgExpr =
-    let lookupKind = LookupKind.Expr LookupIsInstance.Yes
+let private ResolveExprDotLongIdent (ncenv: NameResolver) m ad nenv ty (id: Ident) rest (typeNameResInfo: TypeNameResolutionInfo) findFlag staticOnly maybeArgExpr =
+    let lookupKind = 
+        if staticOnly then LookupKind.Expr LookupIsInstance.No
+        else LookupKind.Expr LookupIsInstance.Yes
     let adhocDotSearchAccessible = AtMostOneResult m (ResolveLongIdentInTypePrim ncenv nenv lookupKind ResolutionInfo.Empty 1 m ad id rest findFlag typeNameResInfo ty maybeArgExpr)
     match adhocDotSearchAccessible with
     | Exception _ ->
@@ -4324,7 +4326,7 @@ let ResolveExprDotLongIdentAndComputeRange (sink: TcResultsSink) (ncenv: NameRes
         let resInfo, item, rest =
             match lid with
             | id :: rest ->
-                ResolveExprDotLongIdent ncenv wholem ad nenv ty id rest typeNameResInfo findFlag maybeAppliedArgExpr
+                ResolveExprDotLongIdent ncenv wholem ad nenv ty id rest typeNameResInfo findFlag staticOnly maybeAppliedArgExpr
             | _ -> error(InternalError("ResolveExprDotLongIdentAndComputeRange", wholem))
         let itemRange = ComputeItemRange wholem lid rest
         resInfo, item, rest, itemRange

tests/FSharp.Compiler.ComponentTests/Conformance/BasicGrammarElements/MethodResolution/StaticMethodResolution.fs

@@ -0,0 +1,41 @@
+// Copyright (c) Microsoft Corporation.  All Rights Reserved.  See License.txt in the project root for license information.
+
+namespace Conformance.BasicGrammarElements
+
+open FSharp.Test.Compiler
+open Xunit
+
+module StaticMethodResolution =
+
+    // Regression test for https://github.com/dotnet/fsharp/issues/19664
+    //
+    // When a static extension method is defined in a *different* [<AutoOpen>] module than
+    // the generic type it extends, and shares its name with an intrinsic static member,
+    // resolving the call via the explicit-type-argument syntax `Type<TArg>.Member(...)`
+    // previously failed with FS0505. The non-generic dotted form `Type.Member(...)`
+    // resolved correctly, so any regression test that omits the explicit type arguments
+    // does not actually exercise the bug. See the discussion at
+    // https://github.com/dotnet/fsharp/issues/19675#issuecomment-4373059900.
+    [<Fact>]
+    let ``Static extension on generic type resolves with explicit type arguments``() =
+        Fsx """
+module Extensions =
+
+    type StaticGeneric<'T> =
+        static member Bar() = ()
+        static member Bar(_: int, _: int) = ()
+
+    [<AutoOpen>]
+    module StaticGenericExtensions =
+        type StaticGeneric<'T> with
+            static member Bar(_: int) = ()
+
+module Program =
+    open Extensions
+
+    StaticGeneric<int>.Bar()        // intrinsic, 0 args
+    StaticGeneric<int>.Bar(42)      // regressed: extension, 1 arg, see issue 19664 (FS0505)
+    StaticGeneric<int>.Bar(42, 0)   // intrinsic, 2 args
+        """
+        |> typecheck
+        |> shouldSucceed

tests/FSharp.Compiler.ComponentTests/Conformance/BasicGrammarElements/PropertyResolution/StaticPropertyResolution.fs

@@ -0,0 +1,45 @@
+// Copyright (c) Microsoft Corporation.  All Rights Reserved.  See License.txt in the project root for license information.
+
+namespace Conformance.BasicGrammarElements
+
+open FSharp.Test.Compiler
+open Xunit
+
+module StaticPropertyResolution =
+
+    // Regression test for static property accessors (getter/setter) resolution.
+    // Related to https://github.com/dotnet/fsharp/issues/19797
+    //
+    // When a static extension 'set' or 'get' accessor is defined in a *different* module 
+    // than the generic type it extends, and the corresponding property has the other intrinsic 
+    // static accessor 
+    // (i.e. the intrinsic property has a 'get' and the extension has a 'set', or vice versa)
+    //
+    // For instance, resolving the assignment via explicit-type-argument syntax, 
+    // where the 'set' accessor is an extension and the 'get' accessor is intrinsic,
+    // Type<TArg>.Property <- value previously failed with FS0810.
+    //
+    // The non-generic dotted form `Type.Member`
+    // resolved correctly, so any regression test that omits the explicit type arguments
+    // does not actually exercise the bug. See the discussion at
+    // https://github.com/dotnet/fsharp/issues/19675#issuecomment-4373059900.
+    [<Fact>]
+    let ``Static property on generic type resolves the extension accessor when the other is intrinsic``() =
+        Fsx """
+module Lib =
+
+    type Label<'T> =
+        static member Text with get() = "Static Intrinsic"
+
+    [<AutoOpen>]
+    module Utils =
+        type Label<'T> with
+            static member Text with set (v) = printfn "Extension"
+
+module Program =
+    open Lib
+
+    Label<int>.Text <- "Text" // regressed: extension setter, see issue 19797 (FS0810)
+        """
+        |> typecheck
+        |> shouldSucceed

tests/FSharp.Compiler.ComponentTests/FSharp.Compiler.ComponentTests.fsproj

@@ -64,6 +64,8 @@
     <Compile Include="Conformance\BasicGrammarElements\MemberDefinitions\OptionalDefaultParamArgs\OptionalDefaultParamArgs.fs" />
     <Compile Include="Conformance\BasicGrammarElements\MemberDefinitions\OverloadingMembers\OverloadingMembers.fs" />
     <Compile Include="Conformance\BasicGrammarElements\MethodResolution\MethodResolution.fs" />
+	<Compile Include="Conformance\BasicGrammarElements\MethodResolution\StaticMethodResolution.fs" />
+	<Compile Include="Conformance\BasicGrammarElements\PropertyResolution\StaticPropertyResolution.fs" />
     <Compile Include="Conformance\BasicGrammarElements\ModuleAbbreviations\ModuleAbbreviations.fs" />
     <Compile Include="Conformance\BasicGrammarElements\ModuleDefinitions\ModuleDefinitions.fs" />
     <Compile Include="Conformance\BasicGrammarElements\NamespaceDeclGroups\NamespaceDeclGroups.fs" />