Week of 06/29/2008 - 02:00 to 07/06/2008 - 01:59
You probably have heard about MapReduce, the method to highly parallelize certain algorithms so that they can be run on large processing farms. Several more-or-less good introductions of how this works exist, but most of them suffer from lots of technological noise, distracting from the core concepts.
Here - mostly for my own understanding - I try to recapture what MapReduce does, and I try to be as minimalistic as possible, even throwing away first MapReduce's biggest feature: parallelism.
Of course I made sure that the Xcode 3.0 developer tools were installed from the shipped DVD, and amazingly enough a cpan frontend to the CPAN mirror servers was already installed together with the Perl 5.8.8 there.
Then the usual incantation:
Yes, I managed not to get beaten up by a drunken soccer fan and am jolly on the spot with the truth, nothing but the truth.
Category EURO 2008
Available on CPAN: Perl TM 1.42
The CTM support is experimental, read-only (and will remain so) and it lacks some features:
- reification of maps
- %mergemap vs. %include directive
I found implementing CTM not quite easy, fortunately lheuer has prepared a number of useful