hi,
nice work on 11l!
as the author of shedskin, I was intrigued by the large performance difference with shedskin for the 'prime number' benchmark.. =) after all, the result should be more or less the same, especially since we both seem to map Python lists to STL vectors.
turns out this uncovered an issue in shedskin, where we use 4 bytes to store a boolean, something I never encountered as a problem (at least not knowingly). this should be patched in a maintenance release of shedskin soon, so thanks for triggering!
all the best, cheers,
mark.
nice work on 11l!
as the author of shedskin, I was intrigued by the large performance difference with shedskin for the 'prime number' benchmark.. =) after all, the result should be more or less the same, especially since we both seem to map Python lists to STL vectors.
turns out this uncovered an issue in shedskin, where we use 4 bytes to store a boolean, something I never encountered as a problem (at least not knowingly). this should be patched in a maintenance release of shedskin soon, so thanks for triggering!
all the best, cheers,
mark.