On 10 March 2015 at 09:01, John SJ Anderson <geneh...@genehack.org> wrote:
> And the average latency of the network connection you're serving this > captcha over is... ? Its probably a bigger issue if you have > 2,000 customers at any time fetching captchas =). But the question is more "Not is the subroutine slower" but "How does that slowness compare to the entire codebase". Comparing vs itself is meaningless, you could make the subroutine faster, and make the application itself *slower*. ( Seems diabolical, but happens, and happened in Perl https://www.youtube.com/watch?v=-HlGQtAuZuY ) If you're not comparing the *whole* program with the *whole* performance against the previous version, and then seeing the performance gains as being substantial enough in *that context*, then you're benchmarking the wrong thing. -- Kent *KENTNL* - https://metacpan.org/author/KENTNL
