You can't use switch expressions in Blazor (sometimes)

[giscus[bot]]

You can't use switch expressions in Blazor

Did you ever try to have a switch expression in Blazor? Yes, and did it go well? Probably not. Let's see why!

https://steven-giesel.com/blogPost/b839cab5-55ad-4fd8-90ca-7799cde8ace0

[Ghost4Man]

I'm pretty sure the C# compiler (Roslyn) uses lowering (likely in LocalRewriter_SwitchExpression.cs) to compile switch expressions, not source generators.

Razor does not have a problem with switch expressions. It just seems to have a problem with < as the first token of a {} block. This is understandable, since code blocks can contain tags/components (<Foo>), so it gets parsed as a MarkupBlock.

If you put something like _ and < 3 as the first guard, then it compiles fine. See this .NET Lab example. Your GetGreeting example also works fine in Razor.

[linkdotnet]

Interesting. I have to check that. If so, I'll update the blog post.

[Ghost4Man]

I also found an open issue dotnet/razor#7230 which mentions that one can also use parentheses around the first pattern as a workaround.

[linkdotnet]

Uhh yeah - that said I will update the blog post and credit you

[davidwengier]

Can confirm this is 100% not a "source generators can't see the output of other source generators" issue, though that certainly is real and does stop some things from working in Razor, like being able to use the Regex generator etc. As pointed out by @Ghost4Man the act of "rewriting" a switch expression to conditional logic is done during by lowering, and that happens with something called "bound nodes", the C# compiler never produces actual C# code for it.

The simple explanation for why some things don't work in Razor files well, is that the Razor compiler has to parse all of the C# in your Razor files, and its C# parser is not up with all of the latest and greatest C# features. In .NET 10 there is a new "Roslyn tokenizer" that it takes advantage of (see https://devblogs.microsoft.com/dotnet/enhancing-razor-productivity-with-new-features/#roslyn-c#-tokenizer for more details) which has unlocked things like raw string literals, but there is still more work to do before all C# features are available. Sometimes you can get away with thing, like if you don't have a < after a {, and the Razor parser will just blindly output whatever is in your Razor file to the generated C#, but that is a long way from being "supported".

[linkdotnet]

@davidwengier The confusion about lowering/source generated is 100% on me and for full transparency I will keep it in the blog post (with the disclaimer on top that it's bs :D)

But really glad to see that there is good progress and conversation around the topic!