Stuff with the Perl Foundation. A couple of patches in the Perl core. A few CPAN modules. That about sums it up.
A short session with Devel::Profile led me from this:
time elapsed (wall): 17.6237
time running program: 13.2164 (74.99%)
time profiling (est.): 4.4072 (25.01%)
number of calls: 526792
To this:
time elapsed (wall): 11.4355
time running program: 8.7706 (76.70%)
time profiling (est.): 2.6649 (23.30%)
number of calls: 301198
Next, I'll be taking a look at Devel::Cover and Devel::SmallProf (which, coincidentally, is by the author of Language::Prolog::Yaswi, a "competitor" module to AI::Prolog).
Regrettably, to get that sort of performance enhancement, I had to engage in some serious encapsulation violations. This will make some further refactorings more difficult. I wonder if I made a mistake?
I've also noticed that some people's tests are failing. I can't replicate it on my machine, but it looks suspiciously like Scalar::Util::looks_like_number fails on 5.8.4 if the number is undefined. This doesn't seem right, though, as this is such a common function. Of course, I see that both of those failures are on boxes my boss owns. Perhaps a plot? On the other hand, all but one of the test reports are from boxes he owns. How depressing. I guess people aren't downloading it after all.
looks_like_number on 5.8.4 (Score:2)
Re:looks_like_number on 5.8.4 (Score:2)
Yup, that's a bug.
Re:looks_like_number on 5.8.4 (Score:2)
Though it occurs to me to wonder what version of Scalar::Util you have? If it's an older version, maybe I can just require the latest?
Re:looks_like_number on 5.8.4 (Score:1)
http://tinyurl.com/5k6t7/ [tinyurl.com]
http://tinyurl.com/598qj [tinyurl.com]
Re:looks_like_number on 5.8.4 (Score:2)
That looks like the problem. I'll just need to add a test for defined in the code.