dotnet/roslyn

Version Used: 5.13.0-1.26502.3 (9e0445c) Notes The example occurs only with the try-catch block Minimal repo: using System.IO; class Program { static void Main() { int i = 0; try { if(i == 1){ goto Fallback; // error CS8648: A goto cannot jump to a location after a using declaration. } using var stream = new MemoryStream(); } catch { } Fallback: i++; } } .NET Lab. Diagnostic Id: "CS8648: A goto cannot jump to a location after a using declaration."). Expected Behavior: The code should compile with no errors Actual Behavior: (12,17): error CS8648: A goto cannot jump to a location after a using declaration. Related #39637

Version Used: Insiders [12210.170] Steps to Reproduce: #nullable enable using System.Collections.Generic; class MyCollection<T> { public int Length { get; } } static class Extensions { public static IEnumerator<T> GetEnumerator<T>(this MyCollection<T>? c) => new List<T>().GetEnumerator(); } class Program { static void Main() { MyCollection<int>? x = null; object[] y = [..x]; } } Diagnostic Id: CS8602 - Dereference of a possibly null reference Expected Behavior: CS8602 is reported for the ..x spread element because the compiler unconditionally dreferences it to access the .Length property Actual Behavior: No warnings are reported, the program crashes at runtime with a NullReferenceException

PR #84961 resolved #80052 by adding a special handling for short is patterns. We should investigate if a more general solution would be possible. See #84961 (comment) by @RikkiGibson: I was wondering whether the increased overhead of the DAG codegen, is inherent to it or not? Basically, is there is some adjustment we can instead make to the DAG codegen, which would allow it to generate code generally as good or better than the "naive" sequence of linear tests? This could either be an optimization of the lowered DAG code itself, or, some shape change to the lowered code which allows a subsequent optimization pass (e.g. temp reuse) to be able to optimize some code it couldn't before. Note that I'm not deeply opposed to the current approach or to being pragmatic about making sure we can solve the key scenario for .NET 11. Just wondering if it's possible for us to pursue a better general solution. And #84961 (comment): #68694 and #72273 already attempted a more general solution bu

The current behavior is dependent on whether we are currently constructing the struct instance, and feels a little inconsistent. Basically, it feels like we should be able to get a similar behavior for member flow states, when any default value of a struct type is used, as when we are working out the initial state of struct fields in a constructor. Note that 'null-resilient' properties using field keyword would be excepted from this. Their initial flow state would be the same as the property. Currently working on a change to make that work more consistently in constructors. (resolving #77991). I consider this issue low priority struct S { public void M() { this = default; this.Prop2.ToString(); // no warning } public S() { this = default; Prop2.ToString(); // warning } public string Prop2 { get; set; } }

Background and Motivation Some analyzers like the trimming and AOT compatibility analyzers, emit diagnostics that would also be emitted by external tools, and therefore would require [UnconditionalSuppressMessage] in order for them to be suppressed. However, suppressing them from the IDE generates either a #pragma warning disable, a [SuppressMessage] or changes to the .editorconfig, none of which will be seen by the external tool. I am proposing a new entry in the WellKnownDiagnosticTags class, that will instruct IDEs to suppress diagnostics tagged with it using [UnconditionalSuppressMessage]. Additionally: IDEs will not suggest suppressing the diagnostic with #pragma warning disable, or configuring its severity in the .editorconfig. If UnconditionalSuppressMessageAttribute is not available, we can either ignore the tag, or not suggest suppressing the diagnostic at all. Optionally, we can report a warning if the user attempts to do any of the above, indicating that the suppression