I have a class that wraps a large file and tries to make it look like a string w.r.t. slicing. Here, "large file" means on the order of hundreds of GB. All the slicing/indexing stuff through __getitem__() works fine, but len() is quite broken. It seems to be converting the value returned by __len__() to a 32-bit integer. If the conversion yields a negative number, it raises an exception.
I'm running Python 2.4.1 on an Opteron running RedHat FC3. It's a 64-bit processor, and Python ints appear to be 64-bit as well, so even if len() only works with ints, it should still be able to handle 64-bit values. Here's a simple example that shows what I'm talking about: >>> import sys >>> sys.version '2.4.1 (#1, Jun 22 2005, 16:00:46) \n[GCC 3.4.2 20041017 (Red Hat 3.4.2-6.fc3)]' >>> type(1<<50) <type 'int'> >>> type(1<<70) <type 'long'> >>> class LenTest: ... def __len__(self): ... return 49000000000 ... >>> x = LenTest() >>> x.__len__() 49000000000 >>> len(x) 1755359744 Is this a bug, a design decision, or do I have something misconfigured in my Python build? -Josh -- http://mail.python.org/mailman/listinfo/python-list
