First, PyTables wins! The API has taken some getting used to (more on that in a different post), but once I got things working, I've been very happy with the performance.
I dug out a mailing list post from last year and turned it into http://www.pytables.org/moin/UserDocuments/CustomDataTypes. But I still have some questions and thoughts: First, how does PyTables map back from _c_classId to class instances? It appears to be metaclass magic, but I don't see it documented anywhere. In general, what does a Table or Leaf subclass need in order to work? More significantly, should monkey-patching the creation function into the File object be the recommended approach? For example, in the sparse-matrix object I'm working on (http://bazaar.launchpad.net/~commonsense/divisi/trunk/annotate/head%3A/csc/divisi/pytables_tensor.py), I use 'create' and 'open' classmethods. (The 'open' would be less important if I subclassed Table instead of wrapping it. In general, some people recommend wrapping external classes over subclassing; any thoughts here?) In fact, in my implementation, I want to hide as much of the details of PyTables as possible from my users, so I only require a filename (and use a default location within the HDF5 file). I use a file-handle pool to avoid opening a file multiple times. So far it works very well despite being extremely simplistic; what would you think about making a proper implementation of a file-handle pool within PyTables proper? -Ken ------------------------------------------------------------------------------ Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge _______________________________________________ Pytables-users mailing list Pytables-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/pytables-users
