I am trying to solve an optimization problem and to find the most efficient way of performing the following commands:
whois -> sed -> while (exit while) ->perform action
while loop currently look like
while [x eq smth]; do
x=$((x+1))
done
some action
Maybe it is more efficient to have while true with an if inside (if clause the same as for while). Also, what is the best case using bash to evaluate the time required for every single step?
The by far biggest performance penalty and most common performance problem in Bash is unnecessary forking.
while [[ something ]]
do
var+=$(echo "$expression" | awk '{print $1}')
done
will be thousands of times slower than
while [[ something ]]
do
var+=${expression%% *}
done
Since the former will cause two forks per iteration, while the latter causes none.
Things that cause forks include but are not limited to pipe | lines, $(command expansion), <(process substitution), (explicit subshells), and using any command not listed in help (which type somecmd will identify as 'builtin' or 'shell keyword').
Well for starters you could remove $(, this creates a subshell and is sure
to slow the task down somewhat
while [ x -eq smth ]
do
(( x++ ))
done
sed(or any other child process) is probably your main overhead, rewriting the loop is unlikely to make a significant difference. Consider using shell expansion instead ofsed, or write the whole lot inawk, or maybe evenperlif you have complex code.