ktfmt fails to parse angle brackets in certain cases

[myabc]

Example 1. In the case of a class inheriting from more than one class or interface, angle bracket syntax is used to denote the supertype from which the inherited implementation is taken.

package com.myandroidapp

import android.app.Application
import androidx.lifecycle.DefaultLifecycleObserver
import androidx.lifecycle.LifecycleOwner

class App : Application(), DefaultLifecycleObserver {

  override fun onCreate() {
    // com.google.googlejavaformat.FormattingError: 11:15: error: did not generate token "<"
    super<Application>.onCreate()
   // etc.
  }
}

Expected behaviour: ktfmt parses/reformats the source code.
Actual behaviour: results in an error similar to com.google.googlejavaformat.FormattingError: 11:15: error: did not generate token "<"

Update: appears to be fixed in 2c17040

Example 2 Type Parameters in Annotations: for example, the @TypeParceler annotation from Kotlinx's Parcelize library allows you to explicitly specify an implementation of Parcelable for types that do not have built-in serialisation support.

import java.util.UUID

import kotlinx.android.parcel.*

@Parcelize
@TypeParceler<UUID, UUIDParceler>()
data class User(
  val id: UUID
)

Expected behaviour: ktfmt parses/reformats the source code.
Actual behaviour: results in an error similar to com.google.googlejavaformat.FormattingError: 5:21: error: expected token: '('; generated < instead or 5:21: error: expected token: 'data'; generated < instead if the empty parentheses are removed.

---

Using ktfmt 0.13 with Spotless plugin for Gradle 4.4.0; Gradle 6.5.0. Targeting Kotlin 1.3.72.

[myabc]

Since looked through the commit history. It appears this commit addresses the issue highlighted in the first example: 2c17040

[cgrushko]

@myabc thanks! if I understand correctly, it's a matter of releasing a new version?
(also, we should think about how to automate releases...)

[myabc]

@cgrushko I built ktfmt from source and installed it into my local Maven repository. I can confirm that 2c17040 does the trick: ktfmt no longer crashes on calls to super with angle bracket syntax to specify the supertype.

I understand correctly, it's a matter of releasing a new version?

If only... 🙂

AFAICT the second issue I listed is still present on master. While it'd be great to see a release with the fix for multiple implementation inheritance, that release would not address all angle bracket syntax bugs.