Tutorial: When perl is not quite fast enough, Nicholas Clark

1: Avoiding optimising
2: Compromises:
 a: Mention XS, but only to avoid it
 b: Introduction to Inline
  Putting an Inline module on CPAN. (not in talk)
 c: Compiling your own perl with tweaked compiler options.
 d: Comparison of speed of perl versions
3: Tests
4: The step by step approach:
 a: profile
 b: identify slow things that may go faster
 c: test ideas
 d: record results
5: Devel::DProf (and other profilers)
6: What causes slowness (part 1)
7: Benchmark (and other timers)
8: What causes slowness (part 2)
9: Speed traps (where you are using something else)
 a: when hashes would be better
 b: when regexps would be better
 c: when pack or unpack would be better
 d: when doing nothing would be better (avoid needless calculations)
10: using AutoSplit and AutoLoader
11: Memoize
12: General tricks (not perl specific)
13: exporting and importing - Exporter is slow
14: regexp traps for the unwary
 and the merits of tr///
15: dynamic (regexps, pack, unpack, sprintf) building regexps and templates at run time
16: better hash keys
 [idea from Merijn, not in the YAPC talk]
17: passing data between processes
 [idea from someone else at YAPC, cannot remember name] 18: Conclusion

If you would like to be a speaker on the Scandinavian Perl Workshop, please read the Call For Papers or contact us at <contact@copenhagen.pm.org>.