#!/bin/bash # run a process in the background, and log it's stdout and stderr to a specific logfile # returncode is stored in $_exitcode # $1 identifier # $2 command (will be eval'ed) # $3 logfile run_background () { [ -z "$1" ] && die_error "run_background: please specify an identifier to keep track of the command!" [ -z "$2" ] && die_error "run_background needs a command to execute!" [ -z "$3" ] && die_error "run_background needs a logfile to redirect output to!" debug "run_background called. identifier: $1, command: $2, logfile: $3" ( \ touch $RUNTIME_DIR/$1-running debug "run_background starting $1: $2 >>$3 2>&1" [ -f $3 ] && echo -e "\n\n\n" >>$3 echo "STARTING $1 . Executing $2 >>$3 2>&1\n" >> $3; var_exit=${1}_exitcode eval "$2" >>$3 2>&1 read $var_exit <<< $? #TODO: bash complains about 'not a valid identifier' debug "run_background done with $1: exitcode (\$$1_exitcode): "${!var_exit}" .Logfile $3" #TODO ${!var_exit} doesn't show anything --> maybe fixed now echo >> $3 rm -f $RUNTIME_DIR/$1-running ) & sleep 2 } # wait until a process is done # $1 identifier wait_for () { [ -z "$1" ] && die_error "wait_for needs an identifier to known on which command to wait!" while [ -f $RUNTIME_DIR/$1-running ] do #TODO: follow_progress dialog mode = nonblocking (so check and sleep is good), cli mode (tail -f )= blocking? (so check is probably not needed as it will be done) sleep 1 done kill $(cat $ANSWER) #TODO: this may not work when mode = cli } # $1 set (array) haystack # $2 needle check_is_in () { [ -z "$1" ] && debug "check_is_in $1 $2" && die_error "check_is_in needs a non-empty needle as \$2 and a haystack as \$1!" # haystack can be empty though local pattern="$1" element shift for element do [[ $element = $pattern ]] && return 0 done return 1 } # cleans up file in the runtime directory who can be deleted cleanup_runtime () { rm -rf $RUNTIME_DIR/.dia* &>/dev/null }