Mark empty outer tags as errors when context switching from C# to markup (RZ1022)

[Copilot]

Summary of the changes

@<>...</> used as a markup template transition compiles silently but throws a runtime InvalidCharacterError because <> is not a valid HTML element name. The parser already had RZ1022 ("Outer tag is missing a name") defined but never emitted it.

• HtmlMarkupParser.ParseStartTag: Emit RZ1022 when the outermost tag of a MarkupInCodeBlock context has an empty name and is immediately closed (<>). The At(SyntaxKind.CloseAngle) guard ensures this only fires for <> and not for < at EOF, < with trailing whitespace, or < expr> patterns.
• HtmlErrorTest: Two new tests covering @{<>foo</>} and @Html.Repeat(10, @<>Foo #@item</>).
• Baseline updates: New .diag.txt baselines for the new tests in both test/TestFiles/ and legacyTest/TestFiles/; existing HtmlTagsTest.EmptyTag (@{<></> Bar}) in both test projects now correctly captures RZ1022.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>empty tags should be marked as errors when context switching from c# to markup</issue_title>
<issue_description>yesterday i ran into this error

because a coworker had created this code to make some rendering dynamic.

to force a context switch from c# to markup.

this (specifically, the empty tag in markup,
<>
should be marked as an error

this was my fix, btw: -

i initially had a different variable name for the argument, but i got an error then, saying that __builder isn't declared - so i fixed it... feels a little like magic, but i can live with that :-P</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes empty tags should be marked as errors when context switching from c# to markup #12810

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[davidwengier]

@copilot please investigate the test failures.

[davidwengier]

I checked and this method has no callers, before this PR. The diagnostic is never used 🤷‍♂️

[Copilot]

@copilot please investigate the test failures.

The failure was in the legacy test project (Microsoft.AspNetCore.Razor.Language.Legacy.Test) — the HtmlTagsTest.EmptyTag test there was missing its .diag.txt baseline, since I'd only generated it for the main test/ project but not for legacyTest/. Fixed in da25b1e by adding legacyTest/TestFiles/ParserTests/HtmlTagsTest/EmptyTag.diag.txt.

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• vb4vsblobprodcus33.vsblob.vsassets.io
  • Triggering command: /usr/share/dotnet/dotnet dotnet test src/Compiler/Microsoft.AspNetCore.Razor.Language/legacyTest/Microsoft.AspNetCore.Razor.Language.Legacy.Test.csproj (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

[davidwengier]

This is introducing a new diagnostic into an existing test, that looks like this:

razor/src/Compiler/Microsoft.AspNetCore.Razor.Language/test/Legacy/HtmlTagsTest.cs

Lines 40 to 45 in c640a1c

	public void EmptyTag()
	{
	// This can happen in situations where a user is in VS' HTML editor and they're modifying
	// the contents of an HTML tag.
	ParseDocumentTest("@{<></> Bar}");
	}

That doesn't seem like an issue to me, but feel free to disagree.

[ToddGrun]

[chsienki]

While I agree that this is the right fix, it's a breaking change. Sure, the code might have errored at runtime, but it did compile and if that code path was never executed it wouldn't have been an issue.

I don't think we can take this until we have warning waves (sigh)

[davidwengier]

if that code path was never executed it wouldn't have been an issue.

You're right :(