Bug in ByteArrayByteBank

Daniel-Alievsky

setBytes method always allocates a new byte array without any necessity. Please see:

	public void setBytes(final long startpos, final byte[] bytes,
		final int offset, final int length)
	{
		// ensure we have space
		checkWritePos(startpos, startpos + length);
		final int neededCapacity = (int) (size + length);
		buffer.ensureCapacity(neededCapacity);

		// copy the data
		System.arraycopy(bytes, offset, buffer.getArray(), (int) startpos, length);
		buffer.setSize(neededCapacity);
		updateSize(startpos + length);
	}

final int neededCapacity = (int) (size + length);
This is incorrect. The new capacity should depend on the starting position, something like this: startpos + length
This has became a problem in by code:

        final byte[] newIFDBytes = oldIFDBytes.clone();
        final DataHandle<?> newStream = getBytesHandle(newIFDBytes, littleEndian);
        ...some writing into newStream...

where

    static BytesHandle getBytesHandle(byte[] data, boolean littleEndian) {
        Objects.requireNonNull(data, "Null data");
        final BytesHandle result = new BytesHandle(new BytesLocation(data));
        result.setLittleEndian(littleEndian);
        return result;
    }

I was hoping that writing to newStream would affect my newIFDBytes array, since I always write within the array's range—I just modified some fields (IFD elements). Unfortunately, the first write to newStream results in a reallocation of the internal byte array. This is bad—after that, I need to read the byte array back from newStream, although this is obviously unnecessary.