#!/bin/bash ## # use named pipes to get the last row of a given DBGEN data chunk ## usage() { [ $# -gt 1 ] && echo "ERROR: $2" echo "USAGE: `basename $1` " echo " gather final row of table (e.g., -T ) at named and " exit } ## # MAIN ## SF=1 DOP=1 # parse command line case $# in 1) [ $1 = "-h" ] && usage $0 [ $1 = "--help" ] && usage $0 t=$1 ;; 2) t=$1 SF=$2 ;; 3) t=$1 SF=$2 DOP=$3 ;; *) usage $0 "invalid argument count" ;; esac # assure a setting for DSS_PATH [ -z "$DSS_PATH" ] && DSS_PATH="." [ ! -d "$DSS_PATH/$SF" ] && mkdir $DSS_PATH/$SF chmod 777 $DSS_PATH DSS_PATH=$DSS_PATH/$SF export DSS_PATH # set the other args case $t in n) f="nation"; DOP=1;; # special case for tiny table r) f="region"; DOP=1;; # special case for tiny table c) f="customer";; s) f="supplier";; P) f="part";; S) f="partsupp";; O) f="orders";; L) f="lineitem";; *) usage "bad table abreviation: $t" esac if [ $DOP -eq 1 ] then PIPE="$DSS_PATH/${f}.tbl" else PIPE="$DSS_PATH/${f}.tbl.${DOP}" fi # create a named pipe for each table rm -rf $PIPE mknod $PIPE p # generate data into it ./dbgen -q -f -s $SF -T $t -S $DOP -C $DOP & # assure that data is being written to the pipe sleep 30 # read the last row from the pipe tail -1 $PIPE > $f.last_row.$SF.$DOP & wait rm $PIPE