WorkingwithC++DatainJavaPresentedby:

JessicaWinbladThisPresentation?2021JessicaWinbladOutlineWorkingwithC++DLLsJavaNativeInterface(JNI)I/OwithFiles&StreamsByte/BitOrderDataTypetranslationRealWorldExamplePoweringtest-equipmentfor12.5secondsProjectRequirementsSendvoltageoutaparallelportforaspecifiedtimeinseconds,toahalf-secondaccuracyRunonaWindows2000orXPmachineIntegratewithatest-suitealreadywritteninJavaEngineeringChallengeWithWindows98thiswouldhavebeeneasyWindows2000provideslesstransparentaccesstothemachine’shardwareOnlyhardwaredevicedriverscandirectlyaccessthehardwareYoucan’twritedevicedriversinJavaSolutionUseadevicedriverDLLthatallowsdirectprogrammingoftheparallelportUseJNItoaccessthedevicedriverJNI(JavaNativeInterface)classioPort{

publicnativevoidOut32(

shortPortAddress,shortdata);

static{System.loadLibrary("jnpout32pkg");}

}

…

ioPortpp=newioPort();

pp.Out32(0x378,currentVal);AnotherRealWorldExampleByteandBitOrderMatterBigEndian(“NetworkByteOrder〞)Eg:Motorolla68kprocessorLittleEndianEg:x86PCMayneedtotestfor

“endianness〞Javahasclassestohelpwith

correctingbyteorder

(eg:)Asourcevalueof0xFFFE=

1111111111111110

Couldbereadas:11111110111111111111111101111111etc.ConvertingDataTypesAppliestobothstreamsandfilesIfyouhaveadoubleinC++shouldyouuseareadDouble()methodofyourjavastream/filereadertoreadit?(No)JavaandC++donotalwayshavethesamenamesforequivalentprimitivetypes.Sometypesdon’tmapnicely.Size1byte2bytes4bytes8bytes16bytes8bits16bits32bits64bits128bitsC++char/bool

float(long)double

Javabooleanchar

floatdoubleSize1byte2bytes4bytes8bytes16bytes8bits16bits32bits64bits128bitsC++byteshortint/longlonglong__int128JavabyteshortintlongBigIntegerInternalSizingofDataTypes*C++sizesareOS/compilerdependent(Win32shown)Signed/UnsignedTypesJavaensuresconsistencybyalwaysusingsignedtypesC++supportsbothsigned&unsignedtypesUnsignedByteSignedByteUnsignedShortSignedShortSize1byte1byte2bytes2bytesValueRange0to255-128to1270to65,535-32,768to32,767PrincipleofConversionToreadinunsignedvaluesfromC++theresultingtypeinJavaneedstobelargerAlso,someextraconversionneedstobedonetofixincorrectsignextension.Na?veUnsignedConversionshortvalue=(short)in.readByte();Question:Ifashortcanholdfrom0to65,535whydoesn’tthisworkforvalues128-255?Answer:SignExtensionappliedwhencastingHowDoesSignExtensionWork?unsignedbyte:129=0x81=100000012Thesignbitisextended:11111111

100000012Intwoscompliment,ifthesignbit=1,thenumberispresumednegative.TwosComplimentRawBits

Signed

Unsigned01111111=12712700000010=2200000001=1100000000=0011111111=?125511111110=?225410000001=?12712910000000=?128128Solution:BitMaskingbyteb=in.readByte();//readsassignedshortbitmask=(short)0xff;shortvalue=(short)(b&bitmask);

111111111xxxxxxxnegative&00000000111111110xFF

000000001xxxxxxxpositiveBeCarefulwithUnsignedIntslonga=(long)(in.readInt()&0xffffffff);doesn’twork!Reason:0xffffffffisanegativevalue.Solution:longa=(long)(in.readInt()&0xffffffffL);DealingwithDecimalsGoingfromaC++doubletoaJavafloatiseasybecausebothare8-byteIEEE754values.GoingfromaC++floattoJavaisharderbecauseJavadoesnothavea4-bytefloattypeButJavagivestoolstomaketheconversioneasyinta=in.readInt();

floatb=FBitsToFloat(a);out.writeInt(Float.floatBitsToInt(floatValue));OtherPitfalls&IssuesIfyourC++codeusedbit-fields,youwillhavetodobitmaskingandshiftingtoreadouttheindividualfiel