AOSP art.module.platform.api.stubs 编译失败（NaN java 错误）答案

【问题标题】：AOSP art.module.platform.api.stubs compile failed (NaN java error)AOSP art.module.platform.api.stubs 编译失败（NaN java 错误）
【发布时间】：2021-11-27 16:39:13
【问题描述】：

Failed to build aosp
[  0% 18/12890] //libcore/mmodules/core_platform_api:art.module.platform.api.stubs javac [common]
FAILED: out/soong/.intermediates/libcore/mmodules/core_platform_api/art.module.platform.api.stubs/android_common/javac/art.module.platform.api.stubs.jar

out/soong/.intermediates/libcore/mmodules/core_platform_api/art.module.platform.api.stubs/android_common/javac/srcjars/java/lang/Double.java:106: error: self-reference in initializer
public static final double NaN = NaN;
                                 ^
out/soong/.intermediates/libcore/mmodules/core_platform_api/art.module.platform.api.stubs/android_common/javac/srcjars/java/lang/Float.java:108: error: cannot find symbol
public static final float NaN = NaNf;
                                ^
  symbol:   variable NaNf   
  location: class Float
2 errors

在每个文件中 Double.java 文件 NaN = 0.0d / 0.0; 在我尝试修改这些文件后，构建重新生成它们时出现相同的错误。该文件在源中的位置以及如何修复这些错误？感谢您的回答。

【问题讨论】：

标签： java android nan android-source lineageos

【解决方案1】：

他们来自metalava。

文件src/main/java/com/android/tools/metalava/model/FieldItem.kt

我敢打赌，您已经更新了 kotlinc，因此代码现在无法按预期工作。那里有两个部分。一个用于 Double，一个用于 Float。两者都无法在 kotlin 1.5+ 中按预期工作

所以你应该替换

is Float -> {
  writer.print(" = ")
  when (value) {
     Float.POSITIVE_INFINITY -> writer.print("(1.0f/0.0f);")
     Float.NEGATIVE_INFINITY -> writer.print("(-1.0f/0.0f);")
     Float.NaN -> writer.print("(0.0f/0.0f);")
  }
}

通过

is Float -> {
  writer.print(" = ")
  when {
    value == Float.POSITIVE_INFINITY -> writer.print("(1.0f/0.0f);")
    value == Float.NEGATIVE_INFINITY -> writer.print("(-1.0f/0.0f);")
    java.lang.Float.isNaN(value) -> writer.print("(0.0f/0.0f);")
    else -> { 
      writer.print(canonicalizeFloatingPointString(value.toString()))
      writer.print("f;")
    }
  }
}

和Double一样，替换块

 is Double -> {
   writer.print(" = ")
   when (value) {
     Double.POSITIVE_INFINITY -> writer.print("(1.0/0.0);")
     Double.NEGATIVE_INFINITY -> writer.print("(-1.0/0.0);")
     Double.NaN -> writer.print("(0.0/0.0);")
     else -> { 
       writer.print(canonicalizeFloatingPointString(value.toString()))
       writer.print(";")
     }
   }
 }

通过

 is Double -> {
     writer.print(" = ")
     when {
         value == Double.POSITIVE_INFINITY -> writer.print("(1.0/0.0);")
         value == Double.NEGATIVE_INFINITY -> writer.print("(-1.0/0.0);")
         java.lang.Double.isNaN(value) -> writer.print("(0.0/0.0);")
         else -> {
             writer.print(canonicalizeFloatingPointString(value.toString()))
             writer.print(";")
         }
     }
 }

【讨论】：