使用read -r读取文件对于MINGW来说确实很奇怪

Question

我有以下代码段，可以在ubuntu的终端应用程序中正常工作，但是相同的代码在MINGW（git bash在Windows中使用的shell）中表现异常。

while true;
do
    while IFS=, read -r x owner login_name
    do
       echo "$x $owner $login_name" # putting this makes it work on MINGW
        getVisitors "$x" "$owner" "$login_name"
    done<"${machines}"
done

function join_by { local IFS="$1"; shift; echo "$*"; }

function getVisitors
{
    machine="$1"
    team="$2"
    user="$3"
    echo "Running ${cmd} on ${machine}"


        result=$(ssh -q -i "${key_file}" -o ConnectTimeout=1 -o BatchMode=yes -o StrictHostKeyChecking=no -t ${user}@${machine} ${cmd})
        isMachineUp=$?
        ips=()
        for visitor in ${result}
        do
            shopt -s nocasematch
            if [[ ! "$visitor" =~ $myHostName.* ]];then
                ips+=("${visitor}")
            fi
            shopt -u nocasematch
        done
        if [[ $isMachineUp -ne 0 ]];then
            ips=("-");
        else
            if [[ $ips == "" ]];then
                ips="-";
            fi
        fi
        echo "Users found on $machine === > $ips"
        addMachine $(join_by , "${ips}") "${machine}" "${team}"
}

上面的代码应该永远循环，并逐行解析“ $ {machines}”文件，该文件是一个csv文件，类似于下面给出的文件。

foo,-,admin bar,-,admin dor,-,admin

但是，该代码段仅解析MINGW shell环境中的第一行。