I just noticed that wordnet_structures.py is quite the resource hog. Specifically, for conversion of the current wordnet db, it grabs well over a hundred megabytes of ram before finishing.
I tweaked it a bit to try to reduce the effect, and the ram usage is somewhat smaller in the current version, which I'm attaching to this mail. How much it needs exactly depends heavily on the architecture it runs on: it's about 169Mb on x86 and 335Mb on x86_64. I'd guess these are probably fairly representative for 32bit and 64bit archs in general. Further optimization wouldn't be easy. While eating this amount of memory isn't nice, I expect that it shouldn't pose a problem on the typical build system, and it doesn't need to run for very long in any event. Regards, Sebastian Hagen
wordnet_structures.py.gz
Description: application/gzip
