I feel this deserves a more general Roslyn analyzer, which would warn whenever an async lambda expression as an argument of a call is implicitly converted to a void-returning delegate type. If the developer really wants an async void function, they can replace the lambda expression with a named local function so that the void is explicit.

The analyzer could abstain from warning if there is no overload that would accept some sort of task-type-returning delegate instead. However, I suspect there wouldn't be too many warnings even without this check.

I feel this deserves a more general Roslyn analyzer, which would warn whenever an async lambda expression as an argument of a call is implicitly converted to a void-returning delegate type. If the developer really wants an async void function, they can replace the lambda expression with a named local function so that the void is explicit.

The analyzer could abstain from warning if there is no overload that would accept some sort of task-type-returning delegate instead. However, I suspect there wouldn't be too many warnings even without this check.

It's great idea, would you like to open an issue in the analyzers repo?

It's great idea, would you like to open an issue in the analyzers repo?

I found some pre-existing feature requests for warning about async void:

• "Avoid Async Void" Analyzer roslyn-analyzers#1576 rejected; they don't want to add analysers for async. The recommendation was to use analysers from vs-threading instead.
• Report a diagnostic for 'async void' analyzer callbacks roslyn-analyzers#4944, but it is scoped to only analyse custom analysers and thus would not apply to programs using System.CommandLine.

At this time then, it does not seem useful to file a new request there. But I'll be looking into whether I can start using the vs-threading analysers.