2010/8/17, Vineet Jain <vinjv...@gmail.com>: > Looking at the source code, it seems that the row object returned back > from pytables will always try and look up the key as name and then > revert to an index by int. If I get a numpy array back and use it in > cython code then I can index (not database index) some_numpy_array[i] > would be translated to direct c code without any overhead.
Ah, right. Yeah, you can use `Table.readWhere()` then. But not underestimate the `Row` object: it is coded in Cython, and it has been very fine-tuned, so I would not be surprised if it can be close enough to an ndarray object. But the best is to benchmark, of course... -- Francesc Alted ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Pytables-users mailing list Pytables-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/pytables-users
