Stack overflow with Dagger 2.56.2

[jrgonzalezg]

My app has been compiling fine until Dagger 2.56.1 (included). But if I try to update to Dagger/Hilt 2.56.2 I get this exception on every build:

Caused by: java.lang.StackOverflowError
        at com.squareup.kotlinpoet.TypeName.hashCode(TypeName.kt:111)
        at com.squareup.kotlinpoet.ParameterizedTypeName.hashCode(ParameterizedTypeName.kt:121)
        at com.squareup.kotlinpoet.TypeVariableName.hashCode(TypeVariableName.kt:92)
        at com.squareup.kotlinpoet.ParameterizedTypeName.hashCode(ParameterizedTypeName.kt:124)
        at com.squareup.kotlinpoet.TypeVariableName.hashCode(TypeVariableName.kt:92)
        at com.squareup.kotlinpoet.ParameterizedTypeName.hashCode(ParameterizedTypeName.kt:124)
        at com.squareup.kotlinpoet.TypeVariableName.hashCode(TypeVariableName.kt:92)
        at com.squareup.kotlinpoet.ParameterizedTypeName.hashCode(ParameterizedTypeName.kt:124)
        at com.squareup.kotlinpoet.TypeVariableName.hashCode(TypeVariableName.kt:92)

The stacktrace keeps repeating those lines which suggest some infinite recursion. And I have no information provided of which of my code leads to that.

[bcorso]

This sounds similar to square/kotlinpoet#1737.

It would be nice to have a repro, but we'd first need to know what type is causing this issue and for that we'd need more info in the stack trace to understand where this was called from in Dagger/Hilt. I'm not sure why the stacktrace isn't giving you more info about where the recursion originated from, but if you can somehow get that information we could do more to help.

[natros]

O remember having a similar problem with recursive errors. In my case I had a mix of javax with Jakarta annotations. Filipe Sousa

…

On Wed, Apr 23, 2025 at 22:53 Brad Corso ***@***.***> wrote: *bcorso* left a comment (google/dagger#4702) <#4702 (comment)> This sounds similar to square/kotlinpoet#1737 <square/kotlinpoet#1737>. It would be nice to have a repro, but we'd first need to know what type is causing this issue and for that we'd need more info in the stack trace to understand where this was called from in Dagger/Hilt. I'm not sure why the stacktrace isn't giving you more info about where the recursion originated from, but if you can somehow get that information we could do more to help. — Reply to this email directly, view it on GitHub <#4702 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABGDZQOCJZESNAF5IXGKLA323ADXHAVCNFSM6AAAAAB3W23T4SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDQMRVGU3TQNRTG4> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[jrgonzalezg]

Sorry, but I have not yet been able to dig much into this due to some personal problems. If I get some free time and if I can create a reproducible example I will update later.

From the comments above and the kotlinpoet issue my current guess is that it may be from types with self referencing generics for emulating self types like:

interface PaginatedElement {
  val identifier: String
}

interface PaginatedSection<Element : PaginatedElement, Section : PaginatedSection<Element, Section>> {
  val identifier: String
  val elements: List<Element>

  fun withNewElements(elements: List<Element>): Section
}

and/or the actual classes implementing PaginatedSection and similar classes / interfaces.

[jrgonzalezg]

I have been able to make a bit more progress here. I have not yet been able to identify which code causes the failure exactly, but there seems to be some incompatibility or odd interaction with the hilt extensions on AndroidX (JetPack). If I remove the dependency and the ksp compiler from that there is no longer an stack overflow happening. Upgrading from version 1.2.0 to version 1.3.0-alpha01 of AndroidX hilt does not help either. Only if I remove their hilt-compiler the build goes through with Dagger 2.56.2. But on Dagger 2.56.1 the build succeeded without issues with both ksp compilers included.