did anything change about how diagnostic factories...

# compiler

[Zac Sweers]

did anything change about how diagnostic factories repackage PsiElement in 2.3.0? I'm trying out a preview of 2.3.0 and finding enabling IDE support now fails in the KTIJ plugin, but not sure if this is an issue on my end or IJ 2025.3 EAP

java.lang.IllegalArgumentException: class org.jetbrains.kotlin.psi.KtAnnotationEntry is not a subtype of class org.jetbrains.kotlin.com.intellij.psi.PsiElement for factory SUGGEST_CLASS_INJECTION
	at org.jetbrains.kotlin.diagnostics.KtDiagnosticKt.checkPsiTypeConsistency(KtDiagnostic.kt:108)
	at org.jetbrains.kotlin.diagnostics.KtDiagnosticKt.access$checkPsiTypeConsistency(KtDiagnostic.kt:1)
	at org.jetbrains.kotlin.diagnostics.KtPsiSimpleDiagnostic.<init>(KtDiagnostic.kt:121)
	at org.jetbrains.kotlin.diagnostics.KtDiagnosticFactory0.on(KtDiagnosticFactory.kt:80)
	at org.jetbrains.kotlin.diagnostics.KtDiagnosticReportHelpersKt.reportOn(KtDiagnosticReportHelpers.kt:53)
	at org.jetbrains.kotlin.diagnostics.KtDiagnosticReportHelpersKt.reportOn$default(KtDiagnosticReportHelpers.kt:48)
	at dev.zacsweers.metro.compiler.fir.checkers.InjectConstructorChecker.check(InjectConstructorChecker.kt:99)

Reason I ask is because my plugin does do the workaround to check for both PsiElement locations: https://github.com/ZacSweers/metro/blob/d7367550c004941978326d3f15ec5532dbb3fb44/c[…]ler/src/main/kotlin/dev/zacsweers/metro/compiler/diagnostics.kt

[Zac Sweers]

Seems so, if I switch and just use the unshaded one it starts working

[hfhbd]

Yeah, why do you use PsiElement and not the KtElement?

[hfhbd]

Like this: https://github.com/hfhbd/validation/blob/main/kotlin-plugin/src/main/kotlin/app/softwork/validation/plugin/kotlin/fir/ValidationErrors.kt#L8

[hfhbd]

I am able to use my plugin inside the IDE without a problem, because KtElement is not affected by shading.

[Zac Sweers]

tribal knowledge copied from other projects blob upside down

[Zac Sweers]

I'll give it a shot

[Zac Sweers]

lgtm, thanks!