Rikki Gibson

@dotnet/roslyn-compiler for a second review

Does this mean we aren't properly doing nullable reinference on pattern variables?

Yes but I think using Span.ToArray isn't the right solution here. Stack alloc'd spans are only used by this feature when the length is a compile time constant. But length...

It looks like test/CI failures in compiler layer are meaningful. Will have to investigate.

How should we document the experimental diagnostics? Perhaps we should add a markdown doc similar to https://github.com/dotnet/roslyn/blob/main/docs/Breaking%20API%20Changes.md which lists the "currently active experiments", providing context for each diagnostic ID we...

@AlekseyTs Changes under Compilers are ready for another review pass.

@jcouv I seem to recall a decision that nullability attributes didn't factor into constraint checking, or something like that. Is it possible that a similar decision was made here, for...

@cston @jcouv Please take a look when you have the chance, thanks.

Test `Microsoft.CodeAnalysis.MSBuild.UnitTests.NewlyCreatedProjectsFromDotNetNew.ValidateCSharpTemplateProjects ` is failing in the latest CI run: ``` System.MissingFieldException : Field not found: 'Microsoft.CodeAnalysis.GeneratorDriverState.TrackIncrementalSteps'. at Microsoft.CodeAnalysis.ExternalAccess.RazorCompiler.HostOutputNode`1.UpdateStateTable(Builder graphState, NodeStateTable`1 previousTable, CancellationToken cancellationToken) at Microsoft.CodeAnalysis.DriverStateTable.Builder.GetLatestStateTableForNode[T](IIncrementalGeneratorNode`1 source) in /_/src/Compilers/Core/Portable/SourceGeneration/Nodes/DriverStateTable.cs:line 60...

Thanks for helping me understand the issue!