[KSP2] Intersection types are not modeled

[ZacSweers]

The following function reports no bounds for T in KSP2

fun <T> example() where T : Appendable, T : CharSequence {

}

In KSP 1 it correctly reports the Appendable and CharSequence bounds, but KSP2 reports no bounds

[ZacSweers]

This remains broken in 2.0.0-1.0.21. Repro: square/kotlinpoet#1885

[neetopia]

I think it should have been fixed, I pulled your repro to my local and it looks that the test is running the artifact from KCT, which is 1.0.20-RC1-1.020. I overrode the test classpath with a local build of KSP

constraints {
    implementation("com.google.devtools.ksp:symbol-processing-aa-embeddable:2.0.0-1.0.22")
  }

and test is passing, still, 1.0.21 won't work with KCT though due to #1908 therefore the local build, otherwise this type parameter bounds should be fine.

[ZacSweers]

ahh, good catch. I'll try that + the latest snapshot

[ZacSweers]

sidenote - could be useful for KSP to publish a bom given all the transitives?

[ZacSweers]

Ah, just read #595

[ZacSweers]

This remains broken in 2.0.0-1.0.21. Repro: square/kotlinpoet#1885

[neetopia]

I think it should have been fixed, I pulled your repro to my local and it looks that the test is running the artifact from KCT, which is 1.0.20-RC1-1.020. I overrode the test classpath with a local build of KSP

constraints {
    implementation("com.google.devtools.ksp:symbol-processing-aa-embeddable:2.0.0-1.0.22")
  }

and test is passing, still, 1.0.21 won't work with KCT though due to #1908 therefore the local build, otherwise this type parameter bounds should be fine.

[ZacSweers]

ahh, good catch. I'll try that + the latest snapshot

[ZacSweers]

sidenote - could be useful for KSP to publish a bom given all the transitives?

[ZacSweers]

Ah, just read #595

[ZacSweers]

Confirmed this is fixed in 2.0.0-1.0.22-SNAPSHOT