Unefficient use of io.netty.util.ByteProcessor$IndexOfProcessor

[comtel2000]

Expected behavior

ByteBufUtils.indexOf(..) or in my case ByteBuf$bytesBefore for any ByteBuf
should work as efficient as a simple for loop without object allocation for ByteProcessor$IndexOfProcessor

Actual behavior

Usage of ByteBuf$indexOf or ByteBuf$bytesBefore stress the garbage collector by allocate ByteProcessor$IndexOfProcessor for each call

By replace all indexOf and bytesBefore with simple for loops reduce the garabge in my App by 50%

Steps to reproduce

compare object allocation / memory footprint for:

AbstractByteBuf$bytesBefore(byte b)

and a simple:

    protected static int bytesBefore(ByteBuf in, byte b) {
      for (int i = in.readerIndex(); i < in.writerIndex(); i++) {
        if (in.getByte(i) == b) {
          return i - in.readerIndex();
        }
      }
      return -1;
    }

Minimal yet complete reproducer code (or URL to code)

Netty version

4.1.39-Final

JVM version (e.g. java -version)

1.8.212 x64

OS version (e.g. uname -a)

Windows 10

[normanmaurer]

@comtel2000 that's a good one... we can definitely add some optimisation here.. Will open a pr shortly

[normanmaurer]

@comtel2000 PTAL #9502