Quote:
Originally Posted by Turbocapitalist
Here is an unscientific test. Ignore the 'real' times and notice the 'user' and 'sys' times for perl versus sed, for as similar usage as I could think of. For more complex tasks, it will be a different matter:
|
Very interesting. The much lower user time for perl confirms that it is much better at regular expressions than sed.
But, to be more accurate, you'd have to call sed for every regular expression like this:
Code:
$ time for i in $(seq 1 100000); do
sed -e 's/a/A/g;' x > /dev/null
sed -e 's/b/B/g;' x > /dev/null
sed -e 's/c/C/g;' x > /dev/null
sed -e ' s/d/D/g;' x > /dev/null
sed -e 's/e/E/g;' x > /dev/null
sed -e 's/f/F/g;' x > /dev/null
sed -e 's/g/G/g;' x > /dev/null
sed -e 's/h/H/g;' x > /dev/null
sed -e 's/i/I/g;' x > /dev/null
sed -e 's/j/J/g;' x > /dev/null
sed -e 's/k/K/g;' x > /dev/null
sed -e 's/l/L/g;' x > /dev/null
sed -e 's/m/M/g;' x > /dev/null
sed -e 's/n/N/g;' x > /dev/null
sed -e 's/o/O/g;' x > /dev/null
sed -e 's/p/P/g;' x > /dev/null
sed -e 's/q/Q/g;' x > /dev/null
sed -e ' s/r/R/g;' x > /dev/null
sed -e 's/s/S/g;' x > /dev/null
sed -e 's/t/T/g;' x > /dev/null
sed -e 's/u/U/g;' x > /dev/null
sed -e 's/v/V/g;' x > /dev/null
sed -e 's/w/W/g;' x > /dev/null
sed -e 's/x/X/g;' x > /dev/null
sed -e 's/y/Y/g;' x > /dev/null
sed -e 's/z/Z/g;' x > /dev/null
done
That would surely raise your sys time dramatically.
