An exercise in embarassing parallelization of Bash functions, done through a file archival example

archive_data.sh

#!/bin/sh

PREFILLED=false
SLOTS=8
END_YEAR=2099

year=2018

declare -a procs[8]

function archive {
    test -z "${1}" && echo 'Missing year parameter!' && return
    local year=${1}
    echo "Zipping files for ${year}..."
    if [[ `find Individual_Hit_Data/ -regex ".*/${year}.*" -printf '.' | wc -c` -eq 0 ]]; then
        continue
    fi
    echo "time find Individual_Hit_Data/ -regex .*/${year}.* | tar jcvf ${year}-individualHits.tar.bz2 -T -"
    echo
}

function spawn {
    local i=${1}
    archive ${year} &    # parallelize simply by forking into the background
    procs[${i}]=$!
    echo "Spawing parallel job at PID $! for year ${year}"
    year=$((${year}+1))
}

while true; do
    for i in $(seq 0 $((${SLOTS}-1))); do
        if ! ${PREFILLED}; then
            spawn ${i}
            echo
            if [ ${#procs[@]} -eq ${SLOTS} ]; then
                PREFILLED=true
            fi
        else
            if ! ps -p ${procs[$i]} >&-; then
                echo "Job PID ${procs[$i]} finished..."
                echo
                unset procs[$i]
                spawn ${i}
            fi
        fi
    done

    if [[ ${year} -gt ${END_YEAR} ]]; then
        exit 0
    fi
done