Question

我最初使用的是位于Central Maven存储库中的库。所以我只是将依赖项包含在IntelliJ中的pom文件中。

现在我需要稍微修改源代码然后我不得不重新编译外部库并在本地安装到我的.m2存储库中。在这种情况下，我是否需要从我的Pom中删除依赖项部分才能使用我自己编译的本地安装的jar文件？

编译和生成jar文件的脚本。我运行这个库源代码附带的脚本来编译和安装在本地。

#!/bin/bash -e

# helper function that ensures cmd returns 0 exit code
function checkexit {
    "$@"
    local status=$?
    if [ $status -ne 0 ]; then
        echo "error with $1" >&2
        exit 1
    fi
    return $status
}

function maybeUpdateRepo {
    if [ "$UPDATE_REPOS" == "true" ]; then
        # are there uncommited changes in the repo?
        if ! (git diff-index --quiet HEAD --) then
           echo "Some uncommited changes found in this repo! Stashing..."
           git stash
        fi
        git pull
    fi
}

# check incoming parameters
while [[ $# -gt 1 ]]
do
key="$1"
#Build type (release/debug), packaging type, chip: cpu,gpu,lib type (static/dynamic)
case $key in
    -c|--chip)
    CHIP="$2"
    shift # past argument
    ;;
    -cc|--compute)
    COMPUTE="$2"
    shift # past argument
    ;;
    -a|--march)
    NATIVE="$2"
    shift # past argument
    ;;
     -l|--libtype)
    LIBTYPE="$2"
    shift # past argument
    ;;
     --scalav)
    SCALAV="$2"
    shift # past argument
    ;;
    -s|--shallow)
    SHALLOW="YES"
    ;;
    -r|--repostrategy)
    REPO_STRATEGY="$2"
    shift # past argument
    ;;
    --deploy)
    DEPLOY="YES"
    ;;
    --testnd4j)
    TEST_ND4J="YES"
    ;;
    --testdatavec)
    TEST_DATAVEC="YES"
    ;;
    --testdl4j)
    TEST_DL4J="YES"
    ;;
    --skiplibnd4j)
    SKIP_LIBND4J="YES"
    ;;
    --skipnd4j)
    SKIP_ND4J="YES"
    ;;
    --skipdatavec)
    SKIP_DATAVEC="YES"
    ;;
    --skipdl4j)
    SKIP_DL4J="YES"
    ;;
    --mvnopts)
    MVN_OPTS="$2"
    shift
    ;;
    *)
            # unknown option
    ;;
esac
shift # past argument or value
done

# default for chip
if [[ -z "$CHIP" ]]; then
    # test for cuda libraries
    if (type ldconfig &> /dev/null); then
       if (ldconfig -p | grep -q libcuda\.so); then
           CHIP="cuda"
       else
           CHIP="cpu"
        fi
    fi
fi

# adjust scala versions
if [ "$SCALAV" == "2.10" ]; then
  SCALA="2.10.6"
fi
# adjust scala versions
if [ "$SCALAV" == "2.11" ]; then
  SCALA="2.11.7"
fi

# set git cloning to a shallow depth if the option says so
if [[ -z "$SHALLOW" ]]; then
    GIT_CLONE="git clone"
else
    GIT_CLONE="git clone --depth 1"
fi

# set git cloning to a shallow depth if the option says so
if [[ -z "$DEPLOY" ]]; then
    MVN_GOAL="install"
else
    MVN_GOAL="deploy"
fi

# Report argument values
echo CHIP          = "${CHIP}"
echo COMPUTE       = "${COMPUTE}"
echo DEPLOY        = "${DEPLOY}"
echo NATIVE        = "${NATIVE}"
echo LIBTYPE       = "${LIBTYPE}"
echo SCALAV        = "${SCALAV}"
echo SHALLOW       = "${SHALLOW}"
echo REPO_STRATEGY = "${REPO_STRATEGY}"
echo TEST_ND4J     = "${TEST_ND4J}"
echo TEST_DATAVEC  = "${TEST_DATAVEC}"
echo TEST_DL4J     = "${TEST_DL4J}"
echo SKIP_LIBND4J  = "${SKIP_LIBND4J}"
echo SKIP_ND4J     = "${SKIP_ND4J}"
echo SKIP_DATAVEC  = "${SKIP_DATAVEC}"
echo SKIP_DL4J     = "${SKIP_DL4J}"
echo MVN_OPTS      = "${MVN_OPTS}"

###########################
# Script execution starts #
###########################

pushd ..

# removes lingering snapshot artifacts from existing maven cache to ensure a
# clean build
JAVA_PROJECTS="nd4j datavec deeplearning4j"
for dirName in $JAVA_PROJECTS; do
    if [ -d "$dirName" ]; then
        pushd "$dirName"
        mvn dependency:purge-local-repository -DreResolve=false -DactTransitively=false
        popd
    fi
done

# What to do with existing repos
case $REPO_STRATEGY in
    "delete")
        DELETE_REPOS="true"
        UPDATE_REPOS=""
        ;;
    "update")
        UPDATE_REPOS="true"
        DELETE_REPOS=""
        ;;
    *) # unknown : do nothing
        UPDATE_REPOS=""
        DELETE_REPOS=""
        ;;
esac

# removes any existing repositories to ensure a clean build
if [[ ! (-z "$DELETE_REPOS") ]]; then
    PROJECTS="libnd4j nd4j datavec deeplearning4j"
    for dirName in $PROJECTS; do
        find . -maxdepth 1 -iname "$dirName" -exec rm -rf "{}" \;
    done
fi

# compile libnd4j
if [[ -z "$SKIP_LIBND4J" ]]; then
    if [[ ! (-z "$DELETE_REPOS") || ! (-d libnd4j) ]]; then
        checkexit $GIT_CLONE https://github.com/deeplearning4j/libnd4j.git
    fi
    pushd libnd4j
    maybeUpdateRepo
    if [[ -z "$NATIVE" ]]; then
        checkexit bash buildnativeoperations.sh "$@" -a native
    else
        checkexit bash buildnativeoperations.sh "$@"
    fi

    if [ "$CHIP" == "cuda" ]; then
        if [[ -z "$COMPUTE" ]]; then
            checkexit bash buildnativeoperations.sh -c cuda
        else
            checkexit bash buildnativeoperations.sh -c cuda -cc "$COMPUTE"
        fi
    fi
    LIBND4J_HOME=$(pwd)
    export LIBND4J_HOME
    popd
fi

# build and install nd4j to maven locally
if [[ -z "$SKIP_ND4J" ]]; then
    if [[ ! (-z "$DELETE_REPOS") || ! (-d nd4j) ]]; then
        checkexit $GIT_CLONE https://github.com/deeplearning4j/nd4j.git
    fi
    if [[ -z "$TEST_ND4J" ]]; then
        ND4J_OPTIONS="-DskipTests"
    else
        ND4J_OPTIONS=""
    fi
    pushd nd4j
    maybeUpdateRepo
    if [ "$CHIP" == "cpu" ]; then
        checkexit bash buildmultiplescalaversions.sh clean $MVN_GOAL -Dmaven.javadoc.skip=true -pl '!nd4j-backends/nd4j-backend-impls/nd4j-cuda,!nd4j-backends/nd4j-backend-impls/nd4j-cuda-platform,!nd4j-backends/nd4j-tests' $ND4J_OPTIONS $MVN_OPTS
    else
        checkexit bash buildmultiplescalaversions.sh clean $MVN_GOAL -Dmaven.javadoc.skip=true $ND4J_OPTIONS $MVN_OPTS
    fi
    popd
fi

# build and install datavec
if [[ -z "$SKIP_DATAVEC" ]]; then
    if [[ ! (-z "$DELETE_REPOS") || ! (-d datavec) ]]; then
        checkexit $GIT_CLONE https://github.com/deeplearning4j/datavec.git
    fi
    if [[ -z "$TEST_DATAVEC" ]]; then
        DATAVEC_OPTIONS="-DskipTests"
    else
        if [ "$CHIP" == "cuda" ]; then
            DATAVEC_OPTIONS="-Ptest-nd4j-cuda-8.0"
        else
            DATAVEC_OPTIONS="-Ptest-nd4j-native"
        fi
    fi
    pushd datavec
    maybeUpdateRepo
    if [ "$SCALAV" == "" ]; then
        checkexit bash buildmultiplescalaversions.sh clean $MVN_GOAL -Dmaven.javadoc.skip=true $DATAVEC_OPTIONS $MVN_OPTS
    else
        checkexit mvn clean $MVN_GOAL -Dmaven.javadoc.skip=true -Dscala.binary.version="$SCALAV" -Dscala.version="$SCALA" $DATAVEC_OPTIONS $MVN_OPTS
    fi
    popd
fi

# build and install deeplearning4j
if [[ -z "$SKIP_DL4J" ]]; then
    if [[ ! (-z "$DELETE_REPOS") ||  ! (-d deeplearning4j) ]]; then
        checkexit $GIT_CLONE https://github.com/deeplearning4j/deeplearning4j.git
    fi
    if [[ -z "$TEST_DL4J" ]] ; then
        DL4J_OPTIONS="-DskipTests"
    else
        if [ "$CHIP" == "cuda" ]; then
            DL4J_OPTIONS="-Ptest-nd4j-cuda-8.0"
        else
            DL4J_OPTIONS="-Ptest-nd4j-native"
        fi
    fi
    pushd deeplearning4j
    maybeUpdateRepo
    if [ $DELETE_REPOS == "true" ]; then
        # reset the working diectory to the latest version of the tracking branch
        git remote update
        TRACKING_BRANCH=$(git rev-parse --abbrev-ref --symbolic-full-name '@{u}')
        git reset --hard "$TRACKING_BRANCH"
    fi
    if [ "$SCALAV" == "" ]; then
        if [ "$CHIP" == "cpu" ]; then
            checkexit bash buildmultiplescalaversions.sh clean $MVN_GOAL -Dmaven.javadoc.skip=true -pl '!./deeplearning4j-cuda/' $DL4J_OPTIONS $MVN_OPTS
        else
            checkexit bash buildmultiplescalaversions.sh clean $MVN_GOAL -Dmaven.javadoc.skip=true $DL4J_OPTIONS $MVN_OPTS
        fi
    else
        checkexit mvn clean $MVN_GOAL -Dmaven.javadoc.skip=true -Dscala.binary.version="$SCALAV" -Dscala.version="$SCALA" $DL4J_OPTIONS $MVN_OPTS
    fi
    popd
fi

popd

Answer 1

我建议您按照以下更好的做法和工作线：

现在我需要稍微修改源代码，然后我必须这样做重新编译外部库并在本地安装到我的.m2中库中。

您必须将其版本从x.y.z更新为x.y.z+1-SNAPSHOT以标记其更改，然后发布工件。我们现在就说你当地人。

注意 - SNAPSHOT因为这似乎正在开发中，否则您也可以将更改标记为x.y.z+1发布版本。

在这种情况下，我是否需要从Pom中删除依赖项部分为了使用我编译的本地安装的jar文件自己？

当您使用依赖项

时，现在在原始项目中

<groupId>somegroup</groupId>
<artifactId>someartifactId</artifactId>
<version>x.y.z</version>

只需将版本更新为

<version>x.y.z+1-SNAPSHOT</version>

修改：这可能有助于理解该方法，What exactly is a Maven Snapshot and why do we need it?

如果在本地安装，我仍然在Pom.xml中包含依赖项吗？

1 个答案: