Package one.microstream.chars

Class MemoryCharConversion_doubleUTF8

java.lang.Object
one.microstream.chars.MemoryCharConversion_doubleUTF8
public final class MemoryCharConversion_doubleUTF8
extends Object

  • Method Summary

    Modifier and Type Method Description
    static long put​(double value, long address)
    This algorithm is somewhere from 3 times to 25 times (depending on exponent) faster than the infinite spaghetti code used in JDK.

    Methods inherited from class java.lang.Object

    clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

  • Method Details

    • put

      public static final long put​(double value, long address)
      This algorithm is somewhere from 3 times to 25 times (depending on exponent) faster than the infinite spaghetti code used in JDK. It generates discrepancies of +/- 1 at the 16th digit and +/- 8 in the 17th digit compared to the stricter JDK algorithm. As digits 16 and 17 cannot be relied upon anyway due to the technical floating point inaccuracy (e.g. the JDK double parser generates comparable discrepancies), the algorithm is still deemed correct ("enough").

      Otherwise, the behavior of the algorithm is the same as that of the JDK implementation (i.e. decimal point character '.', denormalized representation in range ]1E7; 1E-3], exponent character 'E', etc.

      Note that this method is intended as an implementation detail and a "know-what-you-are-doing" tool that does not perform array bound checks. If array bound checking is desired, use VarString.add(double)explicitly.

      Parameters:
      value - the value to be represented as a character sequence.
      address - the memory address where the character sequence shall start.
      Returns:
      the address pointing to after the last character of the assembled sequence.