Front page | perl.perl5.porters |
Postings from August 2012
On Sun, Aug 19, 2012 at 12:14:39PM -0600, Karl Williamson wrote: > I would prefer to benchmark more realistic real-world data, and > wonder if anyone has some ideas on such, or any other thoughts on > this issue. Well, one thing would be to benchmark a worst-case scenario. Pick a swatch that has a large inversion list: i.e. something that is maximally bad for binary search; then run a loop where only the same character is repeatedly looked up (so minimally small cache). And see whether the result is significantly bad. PS: If a swatch has only a small, fixed number of fields, would it be better to use an array rather than a hash to represent it? -- I've often wanted to drown my troubles, but I can't get my wife to go swimming.Thread Previous | Thread Next